Introduction
A ∞ A_\infty A ∞ -algebras were invented in 1963 by Stasheff [85, 86] as a tool in the study of ‘group-like’ topological spaces in homotopy theory, and their study was mostly confined to Algebraic Topology until the 1990’s. A ∞ A_\infty A ∞ -categories were introduced by Fukaya [27] in 1993 to define Fukaya categories in Symplectic Geometry. These formed part of Kontsevich’s Homological Mirror Symmetry Conjecture [53] in 1995, and then received much attention. A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories are now important in many areas of mathematics.
By ∞ \infty ∞ -category we mean quasicategories, a special kind of simplicial set which are now the most popular model for ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -categories. They were introduced by Boardman and Vogt [7], and developed by Joyal [46] and Lurie [60, 61, 63]. This book is heavily influenced by Lurie’s philosophy and results.
A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories belong in an ∞ \infty ∞ -categorical setting since one usually cares about them not up to isomorphism but up to ‘quasi-equivalence’ or ‘homotopy equivalence’, and ∞ \infty ∞ -categories generalize homotopy theory to a wider context. In applications it is usually necessary to make some arbitrary choices to construct an A ∞ A_\infty A ∞ -algebra or A ∞ A_\infty A ∞ -category, and different choices yield homotopy equivalent (but not necessarily isomorphic) answers.
There is a literature on ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories [11, 12, 24, 72, 73, 76, 88], reviewed in §4.10–§4.11. Most previous work has studied ∞ \infty ∞ -categories of the form A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_\infty\operatorname{Cat}^{\mathrm{un}}_R[QE_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ Q E un − 1 ] ∞ , in which one begins with an ordinary category A ∞ Cat R u n A_\infty\operatorname{Cat}^{\mathrm{un}}_R A ∞ Cat R un of (for example, strictly unital) A ∞ A_\infty A ∞ -categories over a commutative ring R R R , and then localizes at the ‘quasi-equivalences’ Q E u n QE_{\mathrm{un}} Q E un to get an ∞ \infty ∞ -category. Note that A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_\infty\operatorname{Cat}^{\mathrm{un}}_R[QE_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ Q E un − 1 ] ∞ is not really known explicitly: by general abstract theory, it exists as the solution to a complicated ∞ \infty ∞ -categorical universal property, and the challenge is to prove something useful about it.
Our approach is very different to most of the existing literature, and indeed to the mainstream of work on ∞ \infty ∞ -categories. We will define our ∞ \infty ∞ -categories, such as the ∞ \infty ∞ -category A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ of unital A ∞ A_\infty A ∞ -categories over R R R in §7.2, completely explicitly as simplicial sets , by saying what the N N N -simplices are. This is unusual: conventional wisdom is that one uses ∞ \infty ∞ -categories for abstract theory, but model categories for calculations and examples.
We quote from the introduction of Land’s useful book [55] on ∞ \infty ∞ -categories:
‘. . . the theory of ∞ \infty ∞ -categories is useful in practice. However, it also has some drawbacks. In my opinion, the biggest one is the following (exaggerated on purpose):
⚠ It is practically impossible to construct anything by hand.’
Most of this book is a counterexample to this, as we construct by hand many ∞ \infty ∞ -categories and ∞ \infty ∞ -functors, so our theory serves as a worked example of parts of ∞ \infty ∞ -category theory. Our definitions are usually quite complicated.
Background on A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories
We briefly recall some basic definitions and facts for A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories, which are given in much more detail in Chapters 3–4. This material is all due to other authors, attributed in Chapters 3–4. Let R R R be a commutative ring throughout.
Definition 1.1. An A ∞ A_{\infty} A ∞ -category A \mathcal{A} A over R R R consists of:
(a) A set, or proper class, Obj A \operatorname{Obj}\mathcal{A} Obj A of objects of A \mathcal{A} A .
(b) For all objects x , y ∈ Obj A x,y \in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , a Z \mathbb{Z} Z -graded R R R -module
Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) . \operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y). Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) . We also write Hom A ( x , y ) \operatorname{Hom}_{\mathcal{A}}(x,y) Hom A ( x , y ) when we want to identify the A ∞ A_{\infty} A ∞ -category A \mathcal{A} A .
(c) For all k ≥ 1 k\geq1 k ≥ 1 and objects x 0 , x 1 , … , x k x_{0},x_{1},\ldots,x_{k} x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) m_{k}^{x_{0},\ldots,x_{k}}:\operatorname{Hom}(x_{0},x_{1})\otimes_{R}\cdots\otimes_{R}\operatorname{Hom}(x_{k-1},x_{k})\longrightarrow\operatorname{Hom}(x_{0},x_{k}) m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) of degree 2 − k 2-k 2 − k . These should satisfy for all n ≥ 1 n\geq1 n ≥ 1 and objects x 0 , … , x n x_{0},\ldots,x_{n} x 0 , … , x n
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (1) \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{x_{0},\ldots,x_{i},x_{i+j},\ldots,x_{n}}\circ\left(\operatorname{id}_{\operatorname{Hom}(x_{0},x_{1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{i-1},x_{i})}\otimes m_{j}^{x_{i},\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{n-1},x_{n})}\right)=0. \tag*{(1)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (1)
Equation (1.1) for n = 1 n=1 n = 1 becomes m 1 x 0 , x 1 ∘ m 1 x 0 , x 1 = 0 m_{1}^{x_{0},x_{1}}\circ m_{1}^{x_{0},x_{1}}=0 m 1 x 0 , x 1 ∘ m 1 x 0 , x 1 = 0 , with m 1 x 0 , x 1 m_{1}^{x_{0},x_{1}} m 1 x 0 , x 1 of degree 1 1 1 . Thus ( Hom ( x 0 , x 1 ) , m 1 x 0 , x 1 ) (\operatorname{Hom}(x_{0},x_{1}),m_{1}^{x_{0},x_{1}}) ( Hom ( x 0 , x 1 ) , m 1 x 0 , x 1 ) is a complex, and we form its cohomology H ∗ ( Hom ( x 0 , x 1 ) ) H^{*}(\operatorname{Hom}(x_{0},x_{1})) H ∗ ( Hom ( x 0 , x 1 )) . For all objects x 0 , x 1 , x 2 x_{0},x_{1},x_{2} x 0 , x 1 , x 2 in A \mathcal{A} A we have a well-defined, associative, R R R -bilinear product, which we call composition ∘ \circ ∘
∘ : H p ( Hom ( x 0 , x 1 ) ) × H q ( Hom ( x 1 , x 2 ) ) ⟶ H p + q ( Hom ( x 0 , x 2 ) ) (2) \circ:H^{p}(\operatorname{Hom}(x_{0},x_{1}))\times H^{q}(\operatorname{Hom}(x_{1},x_{2}))\longrightarrow H^{p+q}(\operatorname{Hom}(x_{0},x_{2})) \tag*{(2)} ∘ : H p ( Hom ( x 0 , x 1 )) × H q ( Hom ( x 1 , x 2 )) ⟶ H p + q ( Hom ( x 0 , x 2 )) (2)
given by
( a 1 + Im m 1 x 0 , x 1 ) ∘ ( a 2 + Im m 1 x 1 , x 2 ) = m 2 x 0 , x 1 , x 2 ( a 1 ⊗ a 2 ) + Im m 1 x 0 , x 2 . (a_{1}+\operatorname{Im}m_{1}^{x_{0},x_{1}})\circ(a_{2}+\operatorname{Im}m_{1}^{x_{1},x_{2}})=m_{2}^{x_{0},x_{1},x_{2}}(a_{1}\otimes a_{2})+\operatorname{Im}m_{1}^{x_{0},x_{2}}. ( a 1 + Im m 1 x 0 , x 1 ) ∘ ( a 2 + Im m 1 x 1 , x 2 ) = m 2 x 0 , x 1 , x 2 ( a 1 ⊗ a 2 ) + Im m 1 x 0 , x 2 .
We can now define a non-unital R R R -linear category H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) to have objects x x x in Obj A \operatorname{Obj}\mathcal{A} Obj A , morphism spaces Hom H 0 ( A ) ( x , y ) = H 0 ( Hom A ( x , y ) ) \operatorname{Hom}_{H^{0}(\mathcal{A})}(x,y)=H^{0}(\operatorname{Hom}_{\mathcal{A}}(x,y)) Hom H 0 ( A ) ( x , y ) = H 0 ( Hom A ( x , y )) , and composition of morphisms ∘ : H 0 ( Hom ( x , y ) ) × H 0 ( Hom ( y , z ) ) ⟶ H 0 ( Hom ( x , z ) ) \circ:H^{0}(\operatorname{Hom}(x,y))\times H^{0}(\operatorname{Hom}(y,z))\longrightarrow H^{0}(\operatorname{Hom}(x,z)) ∘ : H 0 ( Hom ( x , y )) × H 0 ( Hom ( y , z )) ⟶ H 0 ( Hom ( x , z )) as in (1.2). We call H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) the cohomology category of A \mathcal{A} A . Note that in general H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) is not a category in the usual sense, as there need not exist units 1 x ∈ Hom H 0 ( A ) ( x , x ) 1_{x}\in\operatorname{Hom}_{H^{0}(\mathcal{A})}(x,x) 1 x ∈ Hom H 0 ( A ) ( x , x ) for x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A .
Definition 1.2. An A ∞ A_{\infty} A ∞ -category A \mathcal{A} A over R R R is called unital (or strictly unital ) if for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A there exists an identity 1 x ∈ Hom 0 ( x , x ) 1_{x}\in\operatorname{Hom}^{0}(x,x) 1 x ∈ Hom 0 ( x , x ) satisfying
m 2 x 0 , x 0 , x 1 ( 1 x 0 ⊗ a ) = m 2 x 0 , x 1 , x 1 ( a ⊗ 1 x 1 ) = a (3) m_{2}^{x_{0},x_{0},x_{1}}(1_{x_{0}}\otimes a)=m_{2}^{x_{0},x_{1},x_{1}}(a\otimes1_{x_{1}})=a \tag*{(3)} m 2 x 0 , x 0 , x 1 ( 1 x 0 ⊗ a ) = m 2 x 0 , x 1 , x 1 ( a ⊗ 1 x 1 ) = a (3)
for all objects x 0 , x 1 x_0,x_1 x 0 , x 1 in A \mathcal{A} A and a ∈ Hom ( x 0 , x 1 ) a \in\operatorname{Hom}(x_0,x_1) a ∈ Hom ( x 0 , x 1 ) , and
m k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (4) m_{k+1}^{x_0,\ldots,x_i,x_i,\ldots,x_k}(a_1 \otimes\cdots\otimes a_i \otimes1_{x_i} \otimes a_{i+1} \otimes\cdots\otimes a_k)=0 \tag*{(4)} m k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (4)
for all k ≠ 1 k \ne1 k = 1 , i = 0 , … , k i=0,\ldots,k i = 0 , … , k , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k \in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and a j ∈ Hom ( x j − 1 , x j ) a_j \in\operatorname{Hom}(x_{j-1},x_j) a j ∈ Hom ( x j − 1 , x j ) . Identities are unique if they exist, so to be unital is a property of A \mathcal{A} A .
When A \mathcal{A} A is unital, the cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) has identities [ 1 x ] [1_x] [ 1 x ] in H 0 ( Hom ( x , x ) ) H^0(\operatorname{Hom}(x,x)) H 0 ( Hom ( x , x )) , so it is an honest R R R -linear category.
There are several notions of unitality for A ∞ A_\infty A ∞ -categories, of which our unital is the strongest. In §4.1 we also define cohomologically unital and weakly cohomologically unital A ∞ A_\infty A ∞ -categories. Note that names are not consistent across the literature: some authors call our unital A ∞ A_\infty A ∞ -categories ‘strictly unital’, and some authors use ‘unital’ to mean either cohomologically unital, or weakly cohomologically unital, in our sense.
Definition 1.3. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_\infty A ∞ -categories over R R R . An A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B consists of:
(a) A map F O b j : Obj A → Obj B \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{A}\to\operatorname{Obj}\mathcal{B} F Obj : Obj A → Obj B , sometimes just written F \mathcal{F} F .
(b) For all k ≥ 1 k\ge1 k ≥ 1 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
F k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F O b j ( x 0 ) , F O b j ( x k ) ) , \mathcal{F}_k^{x_0,\ldots,x_k}:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{F}_{\mathrm{Obj}}(x_k)), F k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , F Obj ( x k )) , of degree 1 − k 1-k 1 − k . These must satisfy for all n ≥ 1 n\ge1 n ≥ 1 and x 0 , … , x n ∈ Obj A x_0,\ldots,x_n\in\operatorname{Obj}\mathcal{A} x 0 , … , x n ∈ Obj A
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) = ∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F O b j ( x 0 ) , F O b j ( x n 1 ) , … , F O b j ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) . (5) \begin{aligned}
&\sum_{\substack{i,k\ge0,\ j\ge1\\i+j+k=n}}(-1)^{i+jk}\mathcal{F}_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}\circ
\bigl(\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_i)}\\
&\qquad\otimes m_j^{\mathcal{A};x_i,\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{n-1},x_n)}\bigr)\\
&=\sum_{\substack{n=n_1+\cdots+n_k\\n_a\ge1}}(-1)^{\sum_{i=1}^k(n_i+1)(k-i)}
m_k^{\mathcal{B};\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{F}_{\mathrm{Obj}}(x_{n_1}),\ldots,\mathcal{F}_{\mathrm{Obj}}(x_n)}
\bigl(\mathcal{F}_{n_1}^{x_0,\ldots,x_{n_1}}\otimes\mathcal{F}_{n_2}^{x_{n_1},\ldots,x_{n_1+n_2}}\otimes\cdots\otimes\mathcal{F}_{n_k}^{x_{n-n_k},\ldots,x_n}\bigr). \tag*{(5)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) = n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F Obj ( x 0 ) , F Obj ( x n 1 ) , … , F Obj ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) . (5) The identity functor id A : A → A \operatorname{id}^{\mathcal{A}}:\mathcal{A}\to\mathcal{A} id A : A → A has id O b j A = id Obj A \operatorname{id}^{\mathcal{A}}_{\mathrm{Obj}}=\operatorname{id}_{\operatorname{Obj}\mathcal{A}} id Obj A = id Obj A , and id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) \operatorname{id}^{\mathcal{A},x}_1=\operatorname{id}:\operatorname{Hom}_{\mathcal{A}}(x,x)\to\operatorname{Hom}_{\mathcal{A}}(x,x) id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) , and id k A = 0 \operatorname{id}^{\mathcal{A}}_k=0 id k A = 0 for k > 1 k>1 k > 1 .
When n = 1 n=1 n = 1 , equation (1.5) becomes
F 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ F 1 x 0 , x 1 : Hom A ( x 0 , x 1 ) ⟶ Hom B ( F ( x 0 ) , F ( x 1 ) ) . \mathcal{F}_1^{x_0,x_1}\circ m_1^{\mathcal{A};x_0,x_1}=m_1^{\mathcal{B};\mathcal{F}(x_0),\mathcal{F}(x_1)}\circ\mathcal{F}_1^{x_0,x_1}:
\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_1)). F 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ F 1 x 0 , x 1 : Hom A ( x 0 , x 1 ) ⟶ Hom B ( F ( x 0 ) , F ( x 1 )) . Thus F 1 x 0 , x 1 \mathcal{F}_1^{x_0,x_1} F 1 x 0 , x 1 induces morphisms on cohomology
H k ( F 1 x 0 , x 1 ) : H k ( Hom A ( x 0 , x 1 ) ) ⟶ H k ( Hom B ( F ( x 0 ) , F ( x 1 ) ) ) . (6) H^k(\mathcal{F}_1^{x_0,x_1}):H^k(\operatorname{Hom}_{\mathcal{A}}(x_0,x_1))\longrightarrow H^k(\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_1))). \tag*{(6)} H k ( F 1 x 0 , x 1 ) : H k ( Hom A ( x 0 , x 1 )) ⟶ H k ( Hom B ( F ( x 0 ) , F ( x 1 ))) . (6) If F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , G : B → C \mathcal{G}:\mathcal{B}\to\mathcal{C} G : B → C are functors of A ∞ A_\infty A ∞ -categories over R R R , the composition G ∘ F : A → C \mathcal{G}\circ\mathcal{F}:\mathcal{A}\to\mathcal{C} G ∘ F : A → C is given by ( G ∘ F ) O b j = G O b j ∘ F O b j (\mathcal{G}\circ\mathcal{F})_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}}\circ\mathcal{F}_{\mathrm{Obj}} ( G ∘ F ) Obj = G Obj ∘ F Obj and
( G ∘ F ) n x 0 , … , x n = (7) (\mathcal{G}\circ\mathcal{F})_n^{x_0,\ldots,x_n}= \tag*{(7)} ( G ∘ F ) n x 0 , … , x n = (7) ∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) , \sum_{\substack{n=n_1+\cdots+n_k\\n_a\ge1}}(-1)^{\sum_{i=1}^k(n_i+1)(k-i)}
\mathcal{G}_k^{\mathcal{F}(x_0),\mathcal{F}(x_{n_1}),\mathcal{F}(x_{n_1+n_2}),\ldots,\mathcal{F}(x_n)}
\bigl(\mathcal{F}_{n_1}^{x_0,\ldots,x_{n_1}}\otimes\mathcal{F}_{n_2}^{x_{n_1},\ldots,x_{n_1+n_2}}\otimes\cdots\otimes\mathcal{F}_{n_k}^{x_{n-n_k},\ldots,x_n}\bigr), n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) ,
for all n ≥ 1 n \ge1 n ≥ 1 and x 0 , … , x n ∈ Obj A x_0,\ldots,x_n \in\operatorname{Obj}\mathcal{A} x 0 , … , x n ∈ Obj A . Composition is associative. If A , B \mathcal{A},\mathcal{B} A , B are unital, we call an A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B unital if F 1 x , x ( 1 x ) = 1 F Obj ( x ) \mathcal{F}_1^{x,x}(1_x)=1_{\mathcal{F}_{\operatorname{Obj}}(x)} F 1 x , x ( 1 x ) = 1 F Obj ( x ) for all objects x x x in A \mathcal{A} A , and
F k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 \mathcal{F}_{k+1}^{x_0,\ldots,x_i,x_i,\ldots,x_k}(a_1\otimes\cdots\otimes a_i\otimes1_{x_i}\otimes a_{i+1}\otimes\cdots\otimes a_k)=0 F k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0
for all k ≥ 1 k\ge1 k ≥ 1 , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A , i = 0 , … , k i=0,\ldots,k i = 0 , … , k and a j ∈ Hom A ( x j − 1 , x j ) a_j\in\operatorname{Hom}_{\mathcal{A}}(x_{j-1},x_j) a j ∈ Hom A ( x j − 1 , x j ) for j = 1 , … , k j=1,\ldots,k j = 1 , … , k .
Write A ∞ Cat R A_\infty\operatorname{Cat}_{R} A ∞ Cat R for the category with objects A ∞ A_\infty A ∞ -categories A \mathcal{A} A over R R R and morphisms A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B . Write A ∞ Cat R u n ⊂ A ∞ Cat R A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}\subset A_\infty\operatorname{Cat}_{R} A ∞ Cat R un ⊂ A ∞ Cat R for the non-full subcategory with objects unital A ∞ A_\infty A ∞ -categories and morphisms unital A ∞ A_\infty A ∞ -functors. In §4.1–§4.2 we also define further subcategories
A ∞ Cat R u n ⊂ A ∞ Cat R c u ⊂ A ∞ Cat R w c ⊂ A ∞ Cat R (8) A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}\subset A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}}\subset A_\infty\operatorname{Cat}_{R}^{\mathrm{wc}}\subset A_\infty\operatorname{Cat}_{R} \tag*{(8)} A ∞ Cat R un ⊂ A ∞ Cat R cu ⊂ A ∞ Cat R wc ⊂ A ∞ Cat R (8)
with different notions of unitality for objects and morphisms.
Here A ∞ Cat R c u A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}} A ∞ Cat R cu has objects cohomologically unital A ∞ A_\infty A ∞ -categories, that is, units 1 x ∈ Hom 0 ( x , x ) 1_x\in\operatorname{Hom}^{0}(x,x) 1 x ∈ Hom 0 ( x , x ) need not exist, but there should exist (nonunique) cohomological units e x ∈ Hom 0 ( x , x ) e_x\in\operatorname{Hom}^{0}(x,x) e x ∈ Hom 0 ( x , x ) for x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A such that left and right multiplication by e x , e y e_x,e_y e x , e y on the complex ( Hom ∗ ( x , y ) , m 1 x , y ) (\operatorname{Hom}^{*}(x,y),m_1^{x,y}) ( Hom ∗ ( x , y ) , m 1 x , y ) using m 2 x , x , y , m 2 x , y , y m_2^{x,x,y},m_2^{x,y,y} m 2 x , x , y , m 2 x , y , y are chain homotopic to the identity, and A ∞ Cat R w c A_\infty\operatorname{Cat}_{R}^{\mathrm{wc}} A ∞ Cat R wc has objects weakly cohomologically unital A ∞ A_\infty A ∞ -categories A \mathcal{A} A , that is, there exist (unique) units 1 x ∈ H 0 ( Hom ∗ ( x , x ) , m 1 x , x ) 1_x\in H^0(\operatorname{Hom}^{*}(x,x),m_1^{x,x}) 1 x ∈ H 0 ( Hom ∗ ( x , x ) , m 1 x , x ) making the cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) unital.
In both of A ∞ Cat R c u , A ∞ Cat R w c A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}},A_\infty\operatorname{Cat}_{R}^{\mathrm{wc}} A ∞ Cat R cu , A ∞ Cat R wc , morphisms are cohomologically unital A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , that is, H 0 ( F ) : H 0 ( A ) → H 0 ( B ) H^0(\mathcal{F}):H^0(\mathcal{A})\to H^0(\mathcal{B}) H 0 ( F ) : H 0 ( A ) → H 0 ( B ) takes identities [ 1 x ] , [ e x ] , 1 x [1_x],[e_x],1_x [ 1 x ] , [ e x ] , 1 x in H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) to identities in H 0 ( B ) H^0(\mathcal{B}) H 0 ( B ) .
If F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a morphism in A ∞ Cat R u n A_\infty\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , the morphisms (1.6) induce an R R R -linear functor H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) H^0(\mathcal{F}_1):H^0(\mathcal{A})\to H^0(\mathcal{B}) H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) of ordinary R R R -linear categories.
Definition 1.4. A unital A ∞ A_\infty A ∞ -category A \mathcal{A} A over R R R has split units if for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A the map R → Hom 0 ( x , x ) R\to\operatorname{Hom}^{0}(x,x) R → Hom 0 ( x , x ) , λ ↦ λ 1 x \lambda\mapsto\lambda1_x λ ↦ λ 1 x , is injective and splits as a map of R R R -modules. An A ∞ A_\infty A ∞ -category over R R R is called h-projective if for all x , y ∈ Obj A x,y\in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , the complex ( Hom ( x , y ) , m 1 x , y ) (\operatorname{Hom}(x,y),m_1^{x,y}) ( Hom ( x , y ) , m 1 x , y ) is h-projective (see Definition 4.47).
We write A ∞ Cat R u n , h s ⊂ A ∞ Cat R u n , s u ⊂ A ∞ Cat R u n A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}}\subset A_\infty\operatorname{Cat}_{R}^{\mathrm{un,su}}\subset A_\infty\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , hs ⊂ A ∞ Cat R un , su ⊂ A ∞ Cat R un for the full subcategories of objects with split units, and of h-projective objects with split units. In the cohomologically unital case, we write A ∞ Cat R c u , h p ⊂ A ∞ Cat R c u A_\infty\operatorname{Cat}_{R}^{\mathrm{cu,hp}}\subset A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}} A ∞ Cat R cu , hp ⊂ A ∞ Cat R cu for the full subcategory of h-projective objects.
A unital A ∞ A_\infty A ∞ -category A \mathcal{A} A behaves like a ‘cofibrant object’, in a model-theoretic sense, if it is h-projective with split units. A cohomologically unital A ∞ A_\infty A ∞ -category A \mathcal{A} A behaves like a ‘cofibrant object’ if it is h-projective.
If R R R is a field then every A \mathcal{A} A is h-projective. If A \mathcal{A} A is unital then it has split units provided it has no zero objects , that is, no objects z z z with 1 z = 0 1_z=0 1 z = 0 .
Definition 1.5. A differential graded category, or dg -category, over R R R is a unital A ∞ A_\infty A ∞ -category A \mathcal{A} A over R R R with m k x 0 , … , x k = 0 m_k^{x_0,\ldots,x_k}=0 m k x 0 , … , x k = 0 for all k ≥ 3 k\ge3 k ≥ 3 and objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in A \mathcal{A} A . If A , B \mathcal{A},\mathcal{B} A , B are dg -categories, a dg -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a unital A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B such that F k x 0 , … , x k = 0 \mathcal{F}_k^{x_0,\ldots,x_k}=0 F k x 0 , … , x k = 0 for all k ≥ 2 k\ge2 k ≥ 2 and objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in A \mathcal{A} A . We write D G C a t R ⊂ A ∞ C a t R u n \mathrm{DGCat}_{R} \subset\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} DGCat R ⊂ A ∞ Cat R un for the non-full subcategory with objects dg-categories and morphisms dg-functors. We also write D G C a t R A ∞ \mathrm{DGCat}_{R}^{\mathcal{A}_{\infty}} DGCat R A ∞ for the full subcategory of A ∞ C a t R u n \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un whose objects are dg-categories, but whose morphisms are A ∞ \mathcal{A}_{\infty} A ∞ -functors. Then D G C a t R ⊂ D G C a t R A ∞ ⊂ A ∞ C a t R u n \mathrm{DGCat}_{R} \subset\mathrm{DGCat}_{R}^{\mathcal{A}_{\infty}} \subset\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} DGCat R ⊂ DGCat R A ∞ ⊂ A ∞ Cat R un .
Dg-categories are important in many areas of mathematics where A ∞ \mathcal{A}_{\infty} A ∞ -categories are not obviously relevant.
Definition 1.6. An A ∞ \mathcal{A}_{\infty} A ∞ -algebra is equivalent to an A ∞ \mathcal{A}_{\infty} A ∞ -category A \mathcal{A} A with one object ∗ \ast ∗ , so that Obj A = { ∗ } \operatorname{Obj}\mathcal{A}=\{\ast\} Obj A = { ∗ } . We use different notation for A ∞ \mathcal{A}_{\infty} A ∞ -algebras: we generally write A = Hom A ( ∗ , ∗ ) A=\operatorname{Hom}_{\mathcal{A}}(\ast,\ast) A = Hom A ( ∗ , ∗ ) as a graded R R R -module, and m k = m ∗ , … , ∗ k \mathbf{m}_{k}=\mathbf{m}_{\ast,\ldots,\ast}^{k} m k = m ∗ , … , ∗ k , and m = ( m k ) k ≥ 1 \mathbf{m}=(\mathbf{m}_{k})_{k\geq1} m = ( m k ) k ≥ 1 , and then write A ∞ \mathcal{A}_{\infty} A ∞ -algebras as ( A , m ) (A,\mathbf{m}) ( A , m ) rather than A \mathcal{A} A . We write
A ∞ A l g R u n ⊂ A ∞ A l g R c u ⊂ A ∞ A l g R w c ⊂ A ∞ A l g R \mathcal{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un}} \subset\mathcal{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{cu}} \subset\mathcal{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{wc}} \subset\mathcal{A}_{\infty}\mathrm{Alg}_{R} A ∞ Alg R un ⊂ A ∞ Alg R cu ⊂ A ∞ Alg R wc ⊂ A ∞ Alg R for the full subcategories of objects A \mathcal{A} A in the categories (1.8) with Obj A = { ∗ } \operatorname{Obj}\mathcal{A}=\{\ast\} Obj A = { ∗ } .
The theory of A ∞ \mathcal{A}_{\infty} A ∞ -algebras follows from that of A ∞ \mathcal{A}_{\infty} A ∞ -categories by restricting to A ∞ \mathcal{A}_{\infty} A ∞ -categories with one object ∗ \ast ∗ , but it is simpler both notationally and conceptually. We discuss A ∞ \mathcal{A}_{\infty} A ∞ -algebras in Chapters 3 and 6 before the corresponding material for A ∞ \mathcal{A}_{\infty} A ∞ -categories in Chapters 4 and 7.
Definition 1.7. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ \mathcal{A}_{\infty} A ∞ -categories over R R R . As in §4.3, we can define an A ∞ \mathcal{A}_{\infty} A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) whose objects are A ∞ \mathcal{A}_{\infty} A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B . If A , B \mathcal{A},\mathcal{B} A , B are unital, or cohomologically unital, there are A ∞ \mathcal{A}_{\infty} A ∞ -subcategories Fun A ∞ u n ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) , Fun A ∞ c u ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{cu}}(\mathcal{A},\mathcal{B}) Fun A ∞ cu ( A , B ) with objects (cohomologically) unital A ∞ \mathcal{A}_{\infty} A ∞ -functors, which are themselves (cohomologically) unital. They extend to a functor
Fun A ∞ : ( A ∞ C a t R ) o p × A ∞ C a t R ⟶ A ∞ C a t R , \operatorname{Fun}_{\mathcal{A}_{\infty}}:(\mathcal{A}_{\infty}\mathrm{Cat}_{R})^{\mathrm{op}}\times\mathcal{A}_{\infty}\mathrm{Cat}_{R}\longrightarrow\mathcal{A}_{\infty}\mathrm{Cat}_{R}, Fun A ∞ : ( A ∞ Cat R ) op × A ∞ Cat R ⟶ A ∞ Cat R , and similarly for Fun A ∞ u n \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{un}} Fun A ∞ un , Fun A ∞ c u \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{cu}} Fun A ∞ cu .
Definition 1.8. Let A , B \mathcal{A},\mathcal{B} A , B be (possibly unital, or cohomologically unital) A ∞ \mathcal{A}_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be (possibly unital, or cohomologically unital) A ∞ \mathcal{A}_{\infty} A ∞ -functors. As explained in §4.4, there are three notions of ‘2-morphism’ H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G , called A ∞ \mathcal{A}_{\infty} A ∞ -homotopies, A ∞ \mathcal{A}_{\infty} A ∞ -natural transformations, and A ∞ \mathcal{A}_{\infty} A ∞ -natural isomorphisms, which can be unital, or cohomologically unital.
Here A ∞ \mathcal{A}_{\infty} A ∞ -natural transformations are (closed) morphisms in Fun A ∞ ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) , Fun A ∞ u n ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) , Fun A ∞ c u ( A , B ) \operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{cu}}(\mathcal{A},\mathcal{B}) Fun A ∞ cu ( A , B ) . A ∞ \mathcal{A}_{\infty} A ∞ -natural isomorphisms are only defined in the (cohomologically) unital cases, and are A ∞ \mathcal{A}_{\infty} A ∞ -natural transformations which induce isomorphisms in H 0 ( Fun A ∞ u n ( A , B ) ) H^{0}(\operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ un ( A , B )) or H 0 ( Fun A ∞ c u ( A , B ) ) H^{0}(\operatorname{Fun}_{\mathcal{A}_{\infty}}^{\mathrm{cu}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ cu ( A , B )) . A ∞ \mathcal{A}_{\infty} A ∞ -homotopies have a different definition, but can be identified with special A ∞ \mathcal{A}_{\infty} A ∞ -natural isomorphisms in the unital case. An A ∞ \mathcal{A}_{\infty} A ∞ -natural transformation (isomorphism) H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G induces an ordinary natural transformation (isomorphism) of functors H 0 ( H ) : H 0 ( F ) ⇒ H 0 ( G ) H^{0}(\mathcal{H}):H^{0}(\mathcal{F})\Rightarrow H^{0}(\mathcal{G}) H 0 ( H ) : H 0 ( F ) ⇒ H 0 ( G ) .
A (cohomologically) unital A ∞ \mathcal{A}_{\infty} A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is called a homotopy equivalence if there exist a (cohomologically) unital A ∞ \mathcal{A}_{\infty} A ∞ -functor G : B → A \mathcal{G}:\mathcal{B}\to\mathcal{A} G : B → A and A ∞ \mathcal{A}_{\infty} A ∞ -natural isomorphisms S : G ∘ F ⇒ id A \mathcal{S}:\mathcal{G}\circ\mathcal{F}\Rightarrow\operatorname{id}_{\mathcal{A}} S : G ∘ F ⇒ id A and T : F ∘ G ⇒ id B \mathcal{T}:\mathcal{F}\circ\mathcal{G}\Rightarrow\operatorname{id}_{\mathcal{B}} T : F ∘ G ⇒ id B .
We call a (cohomologically) unital F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B a quasi-equivalence if the morphisms (1.6) are isomorphisms for all k , x 0 , x 1 k,x_{0},x_{1} k , x 0 , x 1 , and $H^{0}(\mathcal{F}_{1}):H^{0}(\mathcal{A})\to
H 0 ( B ) H^0(\mathcal{B}) H 0 ( B ) is an equivalence of categories. All homotopy equivalences are quasi-equivalences.
If R R R is a field, and in the unital case A , B \mathcal{A},\mathcal{B} A , B do not have zero objects (objects z z z with 1 z = 0 1_z=0 1 z = 0 ), then homotopy equivalences and quasi-equivalences coincide.
More generally, in the unital case, if A , B \mathcal{A},\mathcal{B} A , B are h-projective with split units then F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a quasi-equivalence if and only if it is a homotopy equivalence, and in the cohomologically unital case, if A , B \mathcal{A},\mathcal{B} A , B are h-projective then F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a quasi-equivalence if and only if it is a homotopy equivalence.
One can argue that for A ∞ A_\infty A ∞ -algebras, A ∞ A_\infty A ∞ -homotopy is the best notion of when two A ∞ A_\infty A ∞ -morphisms are ‘the same’, but for A ∞ A_\infty A ∞ -categories, A ∞ A_\infty A ∞ -natural isomorphism is best. Quasi-equivalence, and homotopy equivalence, are two competing notions of when two (cohomologically) unital A ∞ A_\infty A ∞ -categories are ‘the same’. Notice that Definition 1.8 already makes A ∞ A_\infty A ∞ -categories into something like a 2-category.
Definition 1.9. In §4.8 we define the simplicial nerve functor N A ∞ : A ∞ C a t R u n → C a t ∞ 1 \mathcal{N}_{A_\infty}:A_\infty\mathrm{Cat}^{\mathrm{un}}_R\to\mathrm{Cat}_\infty^1 N A ∞ : A ∞ Cat R un → Cat ∞ 1 , where C a t ∞ 1 ⊂ S S e t \mathrm{Cat}_\infty^1\subset\mathrm{SSet} Cat ∞ 1 ⊂ SSet is the ordinary category of ∞ \infty ∞ -categories. It is only defined for unital A ∞ A_\infty A ∞ -categories A \mathcal{A} A . There is a strict isomorphism of categories H 0 ( A ) ≅ H o ( N A ∞ ( A ) ) H^0(\mathcal{A})\cong\mathrm{Ho}(\mathcal{N}_{A_\infty}(\mathcal{A})) H 0 ( A ) ≅ Ho ( N A ∞ ( A )) , where H o ( C ) \mathrm{Ho}(\mathcal{C}) Ho ( C ) is the homotopy category of an ∞ \infty ∞ -category C \mathcal{C} C . That is, N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) is an ∞ \infty ∞ -categorical enhancement of the cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) .
Definition 1.10. In §4.9 we define when an A ∞ A_\infty A ∞ -category A \mathcal{A} A is pretriangulated . If A \mathcal{A} A is also (cohomologically) unital then the cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) is a triangulated category. We write A ∞ Δ C a t R ⊂ A ∞ C a t R A^\Delta_\infty\mathrm{Cat}_R\subset A_\infty\mathrm{Cat}_R A ∞ Δ Cat R ⊂ A ∞ Cat R , A ∞ Δ C a t R u n ⊂ A ∞ C a t R u n A^\Delta_\infty\mathrm{Cat}^{\mathrm{un}}_R\subset A_\infty\mathrm{Cat}^{\mathrm{un}}_R A ∞ Δ Cat R un ⊂ A ∞ Cat R un , A ∞ Δ C a t R c u ⊂ A ∞ C a t R c u A^\Delta_\infty\mathrm{Cat}^{\mathrm{cu}}_R\subset A_\infty\mathrm{Cat}^{\mathrm{cu}}_R A ∞ Δ Cat R cu ⊂ A ∞ Cat R cu for the full subcategories of pretriangulated objects.
We define the twist functor T w : A ∞ C a t R → A ∞ C a t R \mathrm{Tw}:A_\infty\mathrm{Cat}_R\to A_\infty\mathrm{Cat}_R Tw : A ∞ Cat R → A ∞ Cat R , which maps A ∞ C a t R u n → A ∞ C a t R u n A_\infty\mathrm{Cat}^{\mathrm{un}}_R\to A_\infty\mathrm{Cat}^{\mathrm{un}}_R A ∞ Cat R un → A ∞ Cat R un and A ∞ C a t R c u → A ∞ C a t R c u A_\infty\mathrm{Cat}^{\mathrm{cu}}_R\to A_\infty\mathrm{Cat}^{\mathrm{cu}}_R A ∞ Cat R cu → A ∞ Cat R cu . The essential images of these functors are A ∞ Δ C a t R A^\Delta_\infty\mathrm{Cat}_R A ∞ Δ Cat R , A ∞ Δ C a t R u n A^\Delta_\infty\mathrm{Cat}^{\mathrm{un}}_R A ∞ Δ Cat R un , A ∞ Δ C a t R c u A^\Delta_\infty\mathrm{Cat}^{\mathrm{cu}}_R A ∞ Δ Cat R cu . For any A \mathcal{A} A there is an embedding A ↪ T w A \mathcal{A}\hookrightarrow\mathrm{Tw}\,\mathcal{A} A ↪ Tw A as a full A ∞ A_\infty A ∞ -subcategory. We can think of T w \mathrm{Tw} Tw as being like the derived category functor D b : A ↦ D b A D^b:\mathcal{A}\mapsto D^b\mathcal{A} D b : A ↦ D b A which maps an abelian category A \mathcal{A} A to its bounded derived category D b A D^b\mathcal{A} D b A .
If A \mathcal{A} A is unital and pretriangulated then the simplicial nerve N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) is a stable ∞ \infty ∞ -category. These are the ∞ \infty ∞ -categorical generalization of triangulated categories, discussed in §2.7.
Curved and filtered A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories
Next we very briefly describe the material of Chapter 5.
Definition 1.11. Let R R R be a commutative ring. A curved A ∞ A_\infty A ∞ -category, or weak A ∞ A_\infty A ∞ -category, A \mathcal{A} A over R R R consists of:
(a) A set, or proper class, Obj A \operatorname{Obj}\mathcal{A} Obj A of objects of A \mathcal{A} A .
(b) For all objects x , y ∈ Obj A x,y\in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , a Z \mathbb{Z} Z -graded R R R -module
Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) . \operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y). Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) .
(c) For all k ≥ 0 k \ge0 k ≥ 0 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) m_k^{x_0,\ldots,x_k}:\operatorname{Hom}(x_0,x_1)\otimes_R\operatorname{Hom}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}(x_{k-1},x_k)
\longrightarrow\operatorname{Hom}(x_0,x_k) m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k )
of degree 2 − k 2-k 2 − k . For all n ≥ 0 n \ge0 n ≥ 0 and objects x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n these must satisfy
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (9) \sum_{\substack{i,j,k\ge0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}\circ\left(\operatorname{id}_{\operatorname{Hom}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{i-1},x_i)}\otimes m_j^{x_i,\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{n-1},x_n)}\right)=0. \tag*{(9)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (9)
This is the same as an A ∞ A_\infty A ∞ -category in Definition 1.1, except that we have maps m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k for k ≥ 0 k \ge0 k ≥ 0 rather than k ≥ 1 k \ge1 k ≥ 1 , and in (1.9) we allow j ≥ 0 j \ge0 j ≥ 0 rather than j ≥ 1 j \ge1 j ≥ 1 in (1.1). We define unital curved A ∞ A_\infty A ∞ -categories as in Definition 1.1.
A curved A ∞ A_\infty A ∞ -algebra is equivalent to a curved A ∞ A_\infty A ∞ -category with one object ∗ \ast ∗ , as in Definition 1.6.
Curved A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories occur naturally in many places in Mathematics and Physics. The obvious way to define curved A ∞ A_\infty A ∞ -functors of curved A ∞ A_\infty A ∞ -categories is to follow Definition 1.3, but defining F ~ k \widetilde{F}_k F k for k ≥ 0 k \ge0 k ≥ 0 rather than k ≥ 1 k \ge1 k ≥ 1 . However, we immediately run into a problem: equation (1.5) then involves an infinite sum, as on the right hand side we should sum over decompositions n = n 1 + ⋯ + n k n=n_1+\cdots+n_k n = n 1 + ⋯ + n k with k ≥ 0 k \ge0 k ≥ 0 and n i ≥ 0 n_i \ge0 n i ≥ 0 , rather than n i ≥ 1 n_i \ge1 n i ≥ 1 . The same applies to compositions in (1.7), and the definitions of A ∞ A_\infty A ∞ -homotopies and A ∞ A_\infty A ∞ -natural transformations.
Because of this, it is natural to study curved A ∞ A_\infty A ∞ -algebras and curved A ∞ A_\infty A ∞ -categories in settings in which we have notions of convergence of infinite sums . Chapters 5 and 8 develop a theory of filtered A ∞ A_\infty A ∞ -algebras and filtered A ∞ A_\infty A ∞ -categories, in which the R R R -modules Hom d ( x , y ) \operatorname{Hom}^d(x,y) Hom d ( x , y ) are equipped with complete filtrations over a partially ordered monoid ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) , and convergent infinite sums in Hom d ( x , y ) \operatorname{Hom}^d(x,y) Hom d ( x , y ) are well defined. Our definitions are new, but are based closely on work by Fukaya–Oh–Ohta–Ono [28–31, 33] and De Deken and Lowen [18].
As in [28–31,33], (derived) Fukaya categories of general symplectic manifolds should be constructed using filtered A ∞ A_\infty A ∞ -categories over a Novikov ring.
Cohomology categories H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) are not defined for curved or filtered A ∞ A_\infty A ∞ -categories. So we do not have a well-behaved notion of ‘cohomologically unital’ in the filtered case. We write A ∞ f i C a t R u n ⊂ A ∞ f i C a t R \mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}^{\mathrm{un}}_R\subset\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_R A ∞ fi Cat R un ⊂ A ∞ fi Cat R for the categories of (unital) filtered A ∞ A_\infty A ∞ -categories over a complete filtered commutative ring R R R .
Define subcategories A ∞ f i C a t R , ∞ ⊂ A ∞ f i C a t R , 1 ⊂ A ∞ f i C a t R , 0 = A ∞ f i C a t R \mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_{R,\infty}\subset\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_{R,1}\subset\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_{R,0}=\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_R A ∞ fi Cat R , ∞ ⊂ A ∞ fi Cat R , 1 ⊂ A ∞ fi Cat R , 0 = A ∞ fi Cat R , such that A ∞ f i C a t R , 1 \mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_{R,1} A ∞ fi Cat R , 1 has objects strict filtered A ∞ A_\infty A ∞ -categories A \mathcal{A} A over R R R , with m 0 x = 0 m_0^x=0 m 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A and morphisms all filtered A ∞ A_\infty A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B , including those with F ~ 0 x ≠ 0 \widetilde{\mathcal{F}}_0^x\ne0 F 0 x = 0 , and A ∞ f i C a t R , ∞ \mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}_{R,\infty} A ∞ fi Cat R , ∞ has objects strict filtered A ∞ A_\infty A ∞ -categories A \mathcal{A} A , with m 0 x = 0 m_0^x=0 m 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A , and morphisms strict filtered A ∞ A_\infty A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B , with F ~ 0 x = 0 \widetilde{\mathcal{F}}_0^x=0 F 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A . Similarly we define A ∞ f i C a t R , ∞ u n ⊂ A ∞ f i C a t R , 1 u n ⊂ A ∞ f i C a t R , 0 u n = A ∞ f i C a t R u n \mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}^{\mathrm{un}}_{R,\infty}\subset\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}^{\mathrm{un}}_{R,1}\subset\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}^{\mathrm{un}}_{R,0}=\mathsf{A}^{\mathrm{fi}}_\infty\mathsf{Cat}^{\mathrm{un}}_R A ∞ fi Cat R , ∞ un ⊂ A ∞ fi Cat R , 1 un ⊂ A ∞ fi Cat R , 0 un = A ∞ fi Cat R un . Now objects and morphisms of A ∞ f i Cat R , ∞ A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty} A ∞ fi Cat R , ∞ , A ∞ f i Cat R , ∞ u n A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}^{\mathrm{un}} A ∞ fi Cat R , ∞ un are just A ∞ A_{\infty} A ∞ -categories and A ∞ A_{\infty} A ∞ -functors, with extra filtration data. Thus forgetting filtrations gives functors
Π A ∞ f i A ∞ : A ∞ f i Cat R , ∞ ⟶ A ∞ Cat R , Π A ∞ f i A ∞ : A ∞ f i Cat R , ∞ u n ⟶ A ∞ Cat R u n . (10) \begin{aligned}
\Pi_{A_{\infty}^{\mathrm{fi}}}^{A_{\infty}} &: A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty} \longrightarrow A_{\infty}\operatorname{Cat}_{R},\\
\Pi_{A_{\infty}^{\mathrm{fi}}}^{A_{\infty}} &: A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}^{\mathrm{un}} \longrightarrow A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}. \tag*{(10)}
\end{aligned} Π A ∞ fi A ∞ Π A ∞ fi A ∞ : A ∞ fi Cat R , ∞ ⟶ A ∞ Cat R , : A ∞ fi Cat R , ∞ un ⟶ A ∞ Cat R un . (10)
Functor categories Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) , Fun A ∞ f i u n ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi un ( A , B ) can be defined in the filtered case, and surprisingly, they are A ∞ A_{\infty} A ∞ -categories rather than curved A ∞ A_{\infty} A ∞ -categories. So the definitions of A ∞ A_{\infty} A ∞ -natural transformation, A ∞ A_{\infty} A ∞ -natural isomorphism and homotopy equivalence also work in the filtered case. There are functors
Str : A ∞ Cat R ⟶ A ∞ f i Cat R , ∞ , Str : A ∞ Cat R u n ⟶ A ∞ f i Cat R , ∞ u n . (11) \begin{aligned}
\operatorname{Str} &: A_{\infty}\operatorname{Cat}_{R} \longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty},\\
\operatorname{Str} &: A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} \longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}^{\mathrm{un}}. \tag*{(11)}
\end{aligned} Str Str : A ∞ Cat R ⟶ A ∞ fi Cat R , ∞ , : A ∞ Cat R un ⟶ A ∞ fi Cat R , ∞ un . (11)
which convert a filtered A ∞ A_{\infty} A ∞ -category A \mathcal{A} A into a strict filtered A ∞ A_{\infty} A ∞ -category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) , called the associated strict A ∞ A_{\infty} A ∞ -category . Objects of Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) are pairs ( x , b ) (x,b) ( x , b ) of an object x x x in A \mathcal{A} A and a ‘bounding cochain’ b b b for x x x . Composing with Π A ∞ f i A ∞ \Pi_{A_{\infty}^{\mathrm{fi}}}^{A_{\infty}} Π A ∞ fi A ∞ gives functors
Π A ∞ f i A ∞ ∘ Str : A ∞ f i Cat R ⟶ A ∞ Cat R , Π A ∞ f i A ∞ ∘ Str : A ∞ f i Cat R u n ⟶ A ∞ Cat R u n . (12) \begin{aligned}
\Pi_{A_{\infty}^{\mathrm{fi}}}^{A_{\infty}} \circ\operatorname{Str} &: A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R} \longrightarrow A_{\infty}\operatorname{Cat}_{R},\\
\Pi_{A_{\infty}^{\mathrm{fi}}}^{A_{\infty}} \circ\operatorname{Str} &: A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R}^{\mathrm{un}} \longrightarrow A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}. \tag*{(12)}
\end{aligned} Π A ∞ fi A ∞ ∘ Str Π A ∞ fi A ∞ ∘ Str : A ∞ fi Cat R ⟶ A ∞ Cat R , : A ∞ fi Cat R un ⟶ A ∞ Cat R un . (12)
These are an important part of the programme of Fukaya–Oh–Ohta–Ono [27–31, 33] for defining Fukaya categories of symplectic manifolds ( X , ω ) (X,\omega) ( X , ω ) . First they define a filtered A ∞ A_{\infty} A ∞ -category F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) over the Novikov ring Λ Q n o v \Lambda_{Q}^{\mathrm{nov}} Λ Q nov . Then they apply the functors (1.12) to make it into an A ∞ A_{\infty} A ∞ -category.
Our ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories
Chapter 7, the central chapter in the book, defines many different ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R , which fit into a diagram (7.1) of inclusions of ∞ \infty ∞ -categories:
A ∞ C a t R u n , ∞ = A ∞ C a t R , 2 u n , ∞ ↾ A ∞ C a t R , 3 u n , ∞ = A ˙ ∞ C a t R , 3 u n , ∞ ⟶ A ∞ C a t R , 3 c u , ∞ ⟶ A ∞ C a t R , 3 ∞ ↾ ↾ A ∞ C a t R , 4 u n , ∞ = A ˙ ∞ C a t R , 4 u n , ∞ ⟶ A ∞ C a t R , 4 c u , ∞ ⟶ A ∞ C a t R , 4 ∞ ↾ ↾ A ∞ C a t R , 5 u n , ∞ = A ˙ ∞ C a t R , 5 u n , ∞ ⟶ A ∞ C a t R , 5 c u , ∞ ⟶ A ∞ C a t R , 5 ∞ ↾ ↾ ⋮ ⋮ A ∞ C a t R , ∞ u n , ∞ = A ˙ ∞ C a t R , ∞ u n , ∞ ⟶ A ∞ C a t R , ∞ c u , ∞ ⟶ A ∞ C a t R , ∞ ∞ N ( A ∞ C a t R u n ) = N ( A ∞ C a t R u n ) ⟶ N ( A ∞ C a t R c u ) ⟶ N ( A ∞ C a t R ) . (13) \begin{aligned}
A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} &= A_{\infty}\mathrm{Cat}_{R,2}^{\mathrm{un},\infty} \\
&\upharpoonright\\
A_{\infty}\mathrm{Cat}_{R,3}^{\mathrm{un},\infty} &= \dot{A}_{\infty}\mathrm{Cat}_{R,3}^{\mathrm{un},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,3}^{\mathrm{cu},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,3}^{\infty} \\
&\upharpoonright\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\upharpoonright\\
A_{\infty}\mathrm{Cat}_{R,4}^{\mathrm{un},\infty} &= \dot{A}_{\infty}\mathrm{Cat}_{R,4}^{\mathrm{un},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,4}^{\mathrm{cu},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,4}^{\infty} \\
&\upharpoonright\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\upharpoonright\\
A_{\infty}\mathrm{Cat}_{R,5}^{\mathrm{un},\infty} &= \dot{A}_{\infty}\mathrm{Cat}_{R,5}^{\mathrm{un},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,5}^{\mathrm{cu},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,5}^{\infty} \\
&\upharpoonright\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\upharpoonright\\
&\vdots\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\vdots\\
A_{\infty}\mathrm{Cat}_{R,\infty}^{\mathrm{un},\infty} &= \dot{A}_{\infty}\mathrm{Cat}_{R,\infty}^{\mathrm{un},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,\infty}^{\mathrm{cu},\infty} \longrightarrow A_{\infty}\mathrm{Cat}_{R,\infty}^{\infty} \\
\mathcal{N}(A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) &= \mathcal{N}(A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) \longrightarrow\mathcal{N}(A_{\infty}\mathrm{Cat}_{R}^{\mathrm{cu}}) \longrightarrow\mathcal{N}(A_{\infty}\mathrm{Cat}_{R}). \tag*{(13)}
\end{aligned} A ∞ Cat R un , ∞ A ∞ Cat R , 3 un , ∞ A ∞ Cat R , 4 un , ∞ A ∞ Cat R , 5 un , ∞ A ∞ Cat R , ∞ un , ∞ N ( A ∞ Cat R un ) = A ∞ Cat R , 2 un , ∞ ↾ = A ˙ ∞ Cat R , 3 un , ∞ ⟶ A ∞ Cat R , 3 cu , ∞ ⟶ A ∞ Cat R , 3 ∞ ↾ ↾ = A ˙ ∞ Cat R , 4 un , ∞ ⟶ A ∞ Cat R , 4 cu , ∞ ⟶ A ∞ Cat R , 4 ∞ ↾ ↾ = A ˙ ∞ Cat R , 5 un , ∞ ⟶ A ∞ Cat R , 5 cu , ∞ ⟶ A ∞ Cat R , 5 ∞ ↾ ↾ ⋮ ⋮ = A ˙ ∞ Cat R , ∞ un , ∞ ⟶ A ∞ Cat R , ∞ cu , ∞ ⟶ A ∞ Cat R , ∞ ∞ = N ( A ∞ Cat R un ) ⟶ N ( A ∞ Cat R cu ) ⟶ N ( A ∞ Cat R ) . (13)
Chapter 6 first constructs the analogous diagram (160) of ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -algebras, obtained from (13) by restricting to A ∞ A_{\infty} A ∞ -categories with one object ∗ \ast ∗ . Here in (13) :
The ∞ \infty ∞ -categories in the first two columns, with superscripts ‘un’, have objects unital A ∞ A_{\infty} A ∞ -categories and 1-morphisms unital A ∞ A_{\infty} A ∞ -functors.
The ∞ \infty ∞ -categories in the third column, with superscript ‘cu’, have objects cohomologically unital A ∞ A_{\infty} A ∞ -categories and 1-morphisms cohomologically unital A ∞ A_{\infty} A ∞ -functors.
The ∞ \infty ∞ -categories in the fourth column have objects all A ∞ A_{\infty} A ∞ -categories, and 1-morphisms all A ∞ A_{\infty} A ∞ -functors.
The ∞ \infty ∞ -categories in the first and second columns, linked by ‘=’, are strictly isomorphic, but with definitions written in a different way.
In each row we have subscripts ‘R R R , K K K ’ for K = 2 , 3 , 4 , … , ∞ K = 2, 3, 4, \ldots, \infty K = 2 , 3 , 4 , … , ∞ , where we call K K K the type of the ∞ \infty ∞ -category.
N ( A ∞ C a t R u n ) , … , N ( A ∞ C a t R ) \mathcal{N}(A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}), \ldots, \mathcal{N}(A_{\infty}\mathrm{Cat}_{R}) N ( A ∞ Cat R un ) , … , N ( A ∞ Cat R ) are the nerves of the ordinary categories A ∞ C a t R u n , … , A ∞ C a t R A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}, \ldots, A_{\infty}\mathrm{Cat}_{R} A ∞ Cat R un , … , A ∞ Cat R from §1.1.
The 2-morphisms in A ∞ C a t R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ of type 2 in the first row of (13) correspond to unital A ∞ A_{\infty} A ∞ -natural isomorphisms in Definition 1.8. But in the ∞ \infty ∞ -categories of type K ≥ 3 K \ge3 K ≥ 3 in the subsequent rows, the 2-morphisms correspond to (unital) A ∞ A_{\infty} A ∞ -homotopies of A ∞ A_{\infty} A ∞ -functors, as in Definition 1.8. In the author’s view, A ∞ C a t R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ will be the best ∞ \infty ∞ -category of A ∞ A_{\infty} A ∞ -categories for most applications, as A ∞ A_{\infty} A ∞ -natural isomorphisms are the best notion of when two A ∞ A_{\infty} A ∞ -functors are ‘the same’.
The first column of {(xref)}(1.13) can only be defined for unital A ∞ A_{\infty} A ∞ -categories, as the degeneracy maps in the simplicial sets involve identities 1 x 1_x 1 x . The last three columns are defined by a different method which does not need identities, but requires a more restrictive notion of 2-morphism.
To give a flavour of the definitions of the ∞ \infty ∞ -categories in {(xref)}(1.13), we give a partial definition of A ∞ C a t R , u n ∞ A_{\infty}\mathrm{Cat}_{R,\mathrm{un}}^{\infty} A ∞ Cat R , un ∞ . For N ≥ 0 N \ge0 N ≥ 0 , an N N N -simplex in A ∞ C a t R , u n ∞ A_{\infty}\mathrm{Cat}_{R,\mathrm{un}}^{\infty} A ∞ Cat R , un ∞ is a quadruple ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \bigl((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\emptyset\ne I\subseteq\{0,\ldots,N\}}\bigr) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where:
(a) O i \mathcal{O}^{i} O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , thought of as the set of objects Obj A i \operatorname{Obj}\mathcal{A}^{i} Obj A i of an A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i at vertex i i i of the N N N -simplex Δ N \Delta^{N} Δ N .
(b) f i j : O i → O j f^{ij}:\mathcal{O}^{i}\to\mathcal{O}^{j} f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\le i\le j\le N 0 ≤ i ≤ j ≤ N , thought of as the map F Obj i j : Obj A i → Obj A j \mathcal{F}^{ij}_{\operatorname{Obj}}:\operatorname{Obj}\mathcal{A}^{i}\to\operatorname{Obj}\mathcal{A}^{j} F Obj ij : Obj A i → Obj A j in an A ∞ A_{\infty} A ∞ -functor F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j living on the edge [ i , j ] [i,j] [ i , j ] of Δ N \Delta^{N} Δ N , with f i i = id O i f^{ii}=\operatorname{id}_{\mathcal{O}^{i}} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
Suppose ∅ ≠ I ⊆ { 0 , 1 , … , N } \emptyset\ne I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } with I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M . Define f I = f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 : O i 0 → O i M f^{I}=f^{i_{M-1}i_{M}}\circ f^{i_{M-2}i_{M-1}}\circ\cdots\circ f^{i_{0}i_{1}}:\mathcal{O}^{i_{0}}\to\mathcal{O}^{i_{M}} f I = f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 : O i 0 → O i M .
(c) For all i = 0 , 1 , … , N i=0,1,\ldots,N i = 0 , 1 , … , N and x , y ∈ O i x,y\in\mathcal{O}^{i} x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded R R R -module H i ( x , y ) = ⨁ d ∈ Z H d i ( x , y ) \mathcal{H}^{i}(x,y)=\bigoplus_{d\in\mathbb{Z}}\mathcal{H}^{i}_{d}(x,y) H i ( x , y ) = ⨁ d ∈ Z H d i ( x , y ) , thought of as the Hom space Hom A i ( x , y ) \operatorname{Hom}_{\mathcal{A}^{i}}(x,y) Hom A i ( x , y ) in the A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i . We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^{i}=(\mathcal{H}^{i}(x,y))_{x,y\in\mathcal{O}^{i}} H i = ( H i ( x , y ) ) x , y ∈ O i .
(d) For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M\ge0 M ≥ 0 , and for all k ≥ 0 k\ge0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}} x 0 , … , x k ∈ O i 0 , thought of as objects in A i 0 \mathcal{A}^{i_{0}} A i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) m^{I;x_{0},\ldots,x_{k}}_{k}:\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{i_{0}}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{i_{0}}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k})) m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k ))
graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k . When k = 0 k=0 k = 0 we require that m 0 I ; x 0 = 0 m^{I;x_{0}}_{0}=0 m 0 I ; x 0 = 0 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 .
We write m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 m^{I}_{k}=(m^{I;x_{0},\ldots,x_{k}}_{k})_{x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}}} m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 and m I = ( m k I ) k ≥ 0 m^{I}=(m^{I}_{k})_{k\ge0} m I = ( m k I ) k ≥ 0 .
We require these maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k to satisfy the following identities for all ∅ ≠ I = { i 0 < ⋯ < i M } ⊆ { 0 , 1 , … , N } \emptyset\ne I=\{i_{0}<\cdots<i_{M}\}\subseteq\{0,1,\ldots,N\} ∅ = I = { i 0 < ⋯ < i M } ⊆ { 0 , 1 , … , N } , n ≥ 0 n\ge0 n ≥ 0 , and x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 , firstly if $M=0
0 = ∑ k = 1 , … , n , y 0 , … , y k ∈ O i 0 m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n , (14) 0=\sum_{\substack{k=1,\ldots,n,\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m^{\{i_{0}\};y_{0},\ldots,y_{k}}_{k}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}, \tag*{(14)} 0 = k = 1 , … , n , y 0 , … , y k ∈ O i 0 ∑ m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n , (14)
and secondly if $M>0
0 = ∑ k = 1 , … , n y 0 , … , y k ∈ O i 0 m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M ∑ k = 1 , … , n + M − 1 , y 0 , … , y k ∈ O i M m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (15) 0=\sum_{\substack{k=1,\ldots,n\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m^{I;y_{0},\ldots,y_{k}}_{k}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}
+(-1)^{M}\sum_{\substack{k=1,\ldots,n+M-1,\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}}}}m^{\{i_{M}\};y_{0},\ldots,y_{k}}_{k}\circ d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}. \tag*{(15)} 0 = k = 1 , … , n y 0 , … , y k ∈ O i 0 ∑ m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M k = 1 , … , n + M − 1 , y 0 , … , y k ∈ O i M ∑ m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (15)
Here d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k I ; x 0 , … , x n is a graded R R R -linear map with a complicated definition in §6.1–§6.2 and §7.1, as a sum over labelled trees of multiple compositions of the m l J ; w 0 , … , w l m^{J;w_0,\ldots,w_l}_l m l J ; w 0 , … , w l . In both (1.14) and (1.15) there are only finitely many nonzero terms, and all nonzero terms map to H i M ( f 0 i ( x 0 ) , f i 0 i M ( x n ) ) \mathcal{H}^{iM}(f^{i}_0(x_0),f^{i0iM}(x_n)) H i M ( f 0 i ( x 0 ) , f i 0 i M ( x n )) , so (1.14)–(1.15) should be regarded as equations in this space.
(e) There should exist elements 1 x i ∈ H i ( x , x ) 0 1^i_x \in\mathcal{H}^i(x,x)^0 1 x i ∈ H i ( x , x ) 0 for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N and x ∈ O i x\in\mathcal{O}^i x ∈ O i satisfying analogues of (1.3)–(1.4). These make A i \mathcal{A}^i A i into a unital A ∞ A_\infty A ∞ -category.
(f) Parts (a)–(e) imply that using the data ( ( O i ) , ( f i j ) , ( H i ) , ( m I ) ) ((\mathcal{O}^i),(f^{ij}),(\mathcal{H}^i),(m^I)) (( O i ) , ( f ij ) , ( H i ) , ( m I )) , for all i = 0 , … , N i=0,\ldots,N i = 0 , … , N we can define a unital A ∞ A_\infty A ∞ -category A i \mathcal{A}^i A i , for all 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N we can define a unital A ∞ A_\infty A ∞ -functor F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^i\to\mathcal{A}^j F ij : A i → A j , and for all 0 ≤ i < j < k ≤ N 0\leq i<j<k\leq N 0 ≤ i < j < k ≤ N , using the data m k { i , j , k } ; x 0 , … , x k m^{\{i,j,k\};x_0,\ldots,x_k}_k m k { i , j , k } ; x 0 , … , x k we can define a unital A ∞ A_\infty A ∞ -natural transformation H i j k : F i j ⇒ F i k \mathcal{H}^{ijk}:\mathcal{F}^{ij}\Rightarrow\mathcal{F}^{ik} H ij k : F ij ⇒ F ik . We require this to be a unital A ∞ A_\infty A ∞ -natural isomorphism.
For the ∞ \infty ∞ -subcategories of ‘type K K K ’ in (1.13), we require that m 0 I = 0 m^I_0=0 m 0 I = 0 if ∣ I ∣ ≤ K |I|\leq K ∣ I ∣ ≤ K .
Actually proving that the simplicial sets in (1.13) are ∞ \infty ∞ -categories is very complicated, and among the main results of this book.
We write A ∞ Cat R u n , s u , ∞ ⊂ A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un,su},\infty}_R\subset A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , su , ∞ ⊂ A ∞ Cat R un , ∞ for the full ∞ \infty ∞ -subcategory whose objects have split units .
We write A ∞ Cat R u n , h s , ∞ ⊂ A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un,hs},\infty}_R\subset A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , hs , ∞ ⊂ A ∞ Cat R un , ∞ for the full ∞ \infty ∞ -subcategory whose objects are h h h -projective with split units .
We write A ∞ △ Cat R u n , ∞ ⊂ A ∞ Cat R u n , ∞ A^\triangle_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\subset A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ △ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ for the full ∞ \infty ∞ -subcategory whose objects are pretriangulated .
Remark 1.12. (a) The heart of the definition of A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is equations (1.14)–(1.15), which generalize the A ∞ A_\infty A ∞ -equation (1.1) to N N N -simplices. Their apparent simplicity is deceptive, as the definition of the operators d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k I ; x 0 , … , x n , which takes up §6.1, §6.2 and §7.1, is extraordinarily complicated. The simplest case when M = 0 M=0 M = 0 is
d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = ∑ 0 ≤ j < k , y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) . d^{\{i_0\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{\substack{0\leq j<k,\\ y_i=x_i,\ 0\leq i\leq j,\\ y_i=x_{n-k+i},\ j<i\leq k}}
(-1)^{j+(n-k+1)(k-j-1)}
\operatorname{id}_{\mathcal{H}^{i_0}(x_0,x_1)}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j-1},x_j)}
\otimes m^{\{i_0\};x_j,x_{j+1},\ldots,x_{j+n-k+1}}_{n-k+1}\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j+n-k+1},x_{j+n-k+2})}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{n-1},x_n)}. d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = 0 ≤ j < k , y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ∑ ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) .
Substituting this into (1.14) shows that the m k { i 0 } ; y 0 , … , y k m^{\{i_0\};y_0,\ldots,y_k}_k m k { i 0 } ; y 0 , … , y k satisfy (1.1), consistent with the N N N -simplex above including the data of an A ∞ A_\infty A ∞ -category A i \mathcal{A}^i A i at each vertex i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
Equations (1.14)–(1.15) have some very special properties, which are not obvious. From the full definition of A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ in §7.2, which is only sketched above, it is fairly straightforward to show that A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is a simplicial set. However, the crucial proof in Theorem 7.13 that A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is an ∞ \infty ∞ -category is very long and difficult, and relies on deep properties of (1.14)–(1.15).
(b) Equations (14) –(15) are used to define the ∞ \infty ∞ -categories in the first column of (13) . For the last three columns, they are replaced by
∑ 0 ≤ j ≤ M , 0 ≤ k ≤ max ( n , n + j − 1 ) , y 0 , … , y k ∈ O ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = ∑ j = 1 , … , M − 1 ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n (16) \sum_{\substack{0\le j\le M,\ 0\le k\le\max(n,n+j-1),\\ y_0,\ldots,y_k\in\mathcal O}}
(-1)^j m_k^{\{i_j,\ldots,i_M\};y_0,\ldots,y_k}\circ d_{n,k;y_0,\ldots,y_k}^{\{i_0,\ldots,i_j\};x_0,\ldots,x_n}
=
\sum_{j=1,\ldots,M-1}(-1)^j m_n^{I\setminus\{i_j\};x_0,\ldots,x_n} \tag*{(16)} 0 ≤ j ≤ M , 0 ≤ k ≤ m a x ( n , n + j − 1 ) , y 0 , … , y k ∈ O ∑ ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = j = 1 , … , M − 1 ∑ ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n (16)
This only makes sense if the maps f i j : O i → O j f^{ij}:\mathcal O^i\to\mathcal O^j f ij : O i → O j in (b) above satisfy the extra condition f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik for all 0 ≤ i < j < k ≤ N 0\le i<j<k\le N 0 ≤ i < j < k ≤ N , as otherwise different terms in (16) map to different spaces. This condition f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik is imposed in ∞ \infty ∞ -categories A ∞ C a t R , K ∗ , ∞ \mathcal A_{\infty}\mathrm{Cat}^{*,\infty}_{R,K} A ∞ Cat R , K ∗ , ∞ in (13) of type K K K for K ≥ 3 K\ge3 K ≥ 3 , but not K = 2 K=2 K = 2 . This is why the last three columns in (13) have no entry with K = 2 K=2 K = 2 .
Equation (16) also has non-obvious special properties needed to show that the last three columns of (13) are ∞ \infty ∞ -categories.
The definitions of ∞ \infty ∞ -categories using (14) –(15) requires the A i \mathcal A^i A i to be (strictly) unital, as the identities 1 x 1_x 1 x are used in the degeneracy maps in the simplicial set. The definitions using (16) work for non-unital A ∞ \mathcal A_\infty A ∞ -categories, but require f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik . The effect of this is that 2-morphisms in the last three columns of (13) are A ∞ \mathcal A_\infty A ∞ -homotopies, not unital A ∞ \mathcal A_\infty A ∞ -natural isomorphisms as in (f) above.
(c) It would have been very difficult to invent equations (14) –(15) or (16) from scratch. The author derived (16) from a construction in Symplectic Geometry, as part of a project with Guillem Cazassus and Alexander Ritter [16]. Very roughly, from a (for simplicity, say exact) symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) with a choice of compatible almost complex structure J J J , one can construct a Fukaya A ∞ \mathcal A_\infty A ∞ -category F ( X , ω ) J \mathcal F(X,\omega)_J F ( X , ω ) J . The author generalized this to define a categorical structure associated to a smooth family of almost complex structures J σ J_\sigma J σ parametrized by points σ \sigma σ in the topological N N N -simplex Δ t o p N \Delta^N_{\mathrm{top}} Δ top N . This was an example of an N N N -simplex in A ∞ C a t R c u , ∞ \mathcal A_{\infty}\mathrm{Cat}^{\mathrm{cu},\infty}_R A ∞ Cat R cu , ∞ , and led to its definition in §7.3. Later, the author transformed (16) to (14) –(15) , in order to define ∞ \infty ∞ -categories of A ∞ \mathcal A_\infty A ∞ -categories in which 2-morphisms were A ∞ \mathcal A_\infty A ∞ -natural transformations.
For filtered A ∞ \mathcal A_\infty A ∞ -categories over a complete Γ \Gamma Γ -filtered commutative ring R R R , as in §1.2, Chapter 8 defines a diagram (224) of ∞ \infty ∞ -categories of filtered A ∞ \mathcal A_\infty A ∞ categories over R R R similar to (1.13):
A ∞ f i Cat R u n , ∞ = A ∞ f i Cat R , 0 , 2 u n , ∞ A ∞ f i Cat R , 0 , 3 u n , ∞ = A ˚ ∞ f i Cat R , 0 , 3 u n , ∞ ⟶ A ∞ f i Cat R ∞ = A ∞ f i Cat R , 0 , 3 ∞ A ∞ f i Cat R , K ′ , L ′ u n , ∞ = A ˚ ∞ f i Cat R , K ′ , L ′ u n , ∞ ⟶ A ∞ f i Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) , K ≥ K ′ , L ≥ L ′ A ∞ f i Cat R , K , L u n , ∞ = A ˚ ∞ f i Cat R , K , L u n , ∞ ⟶ A ∞ f i Cat R , K , L ∞ ( 0 ≤ K ≤ L , L ≥ 3 ) , K ≥ K ′ , L ≥ L ′ A ∞ f i Cat R , K , ∞ u n , ∞ = A ˚ ∞ f i Cat R , K , ∞ u n , ∞ ⟶ A ∞ f i Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) , N ( A ∞ f i Cat R , K u n ) = N ( A ˚ ∞ f i Cat R , K u n ) ⟶ N ( A ∞ f i Cat R , K ) . (17) \begin{aligned}
&\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}
=\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,2}\\
&\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,3}
=\mathring{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,3}
\longrightarrow\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R}
=\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,0,3}\\
&\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K',L'}
=\mathring{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K',L'}
\longrightarrow\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K',L'}
\quad(0\leq K'\leq L',\ L'\geq3),\\
&\hspace{2cm}K\geq K',\ L\geq L'\\
&\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L}
=\mathring{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L}
\longrightarrow\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}
\quad(0\leq K\leq L,\ L\geq3),\\
&\hspace{2cm}K\geq K',\ L\geq L'\\
&\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty}
=\mathring{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty}
\longrightarrow\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,\infty}
\quad(K=0,1,\infty),\\
&\mathcal{N}\left(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K}\right)
=\mathcal{N}\left(\mathring{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K}\right)
\longrightarrow\mathcal{N}\left(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R,K}\right). \tag*{(17)}
\end{aligned} A ∞ fi Cat R un , ∞ = A ∞ fi Cat R , 0 , 2 un , ∞ A ∞ fi Cat R , 0 , 3 un , ∞ = A ˚ ∞ fi Cat R , 0 , 3 un , ∞ ⟶ A ∞ fi Cat R ∞ = A ∞ fi Cat R , 0 , 3 ∞ A ∞ fi Cat R , K ′ , L ′ un , ∞ = A ˚ ∞ fi Cat R , K ′ , L ′ un , ∞ ⟶ A ∞ fi Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) , K ≥ K ′ , L ≥ L ′ A ∞ fi Cat R , K , L un , ∞ = A ˚ ∞ fi Cat R , K , L un , ∞ ⟶ A ∞ fi Cat R , K , L ∞ ( 0 ≤ K ≤ L , L ≥ 3 ) , K ≥ K ′ , L ≥ L ′ A ∞ fi Cat R , K , ∞ un , ∞ = A ˚ ∞ fi Cat R , K , ∞ un , ∞ ⟶ A ∞ fi Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) , N ( A ∞ fi Cat R , K un ) = N ( A ˚ ∞ fi Cat R , K un ) ⟶ N ( A ∞ fi Cat R , K ) . (17)
Here the ∞ \infty ∞ -categories A ∞ f i Cat R , K , L ∗ , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K,L} A ∞ fi Cat R , K , L ∗ , ∞ are of type ( K , L ) (K,L) ( K , L ) for 0 ≤ K ≤ L ≤ ∞ 0\leq K\leq L\leq\infty 0 ≤ K ≤ L ≤ ∞ .
For N N N -simplices ( ( O i ) , ( f i j ) , ( H i ) , ( m I ) ) \left((\mathcal{O}^{i}),(f^{ij}),(\mathcal{H}^{i}),(m^{I})\right) ( ( O i ) , ( f ij ) , ( H i ) , ( m I ) ) in A ∞ f i Cat R , K , L ∗ , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K,L} A ∞ fi Cat R , K , L ∗ , ∞ , when ∣ I ∣ ≠ 3 |I|\neq3 ∣ I ∣ = 3 , the rule is that m 0 I ; x 0 = 0 m^{I;x_{0}}_{0}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\leq K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) m^{I;x_{0}}_{0}(1) m 0 I ; x 0 ( 1 ) should have strictly positive grading if K < ∣ I ∣ ≤ L K<|I|\leq L K < ∣ I ∣ ≤ L .
We also define a diagram of ∞ \infty ∞ -categories (8.2) identical to (1.17), but for ∞ \infty ∞ -categories A ∞ w f Cat R u n , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ , A ∞ w f Cat R , K , L u n , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ wf Cat R , K , L un , ∞ , … \ldots … that we call weakly filtered . These have the same objects and 1-morphisms as A ∞ f i Cat R u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , A ∞ f i Cat R , K , L u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ , … \ldots … , but have a weaker notion of N N N -morphism for N ≥ 2 N\geq2 N ≥ 2 : in A ∞ f i Cat R u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , … \ldots … the maps m I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}} m I ; x 0 , … , x k are required to preserve filtrations, but in A ∞ w f Cat R u n , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ , … \ldots … the m I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}} m I ; x 0 , … , x k are allowed to decrease filtrations by a bounded amount.
We will compare our ∞ \infty ∞ -categories with ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories defined by localization. See § 2.6 for more details on the next definition.
Definition 1.13. A category with weak equivalences ( C , W ) (C,W) ( C , W ) is a category C C C and a subset W ⊆ Mor ( C ) W\subseteq\operatorname{Mor}(C) W ⊆ Mor ( C ) of the morphisms of C C C , such that W W W contains all isomorphisms and satisfies the ‘two out of three’ property. A Dwyer–Kan localization of C C C along W W W is an ∞ \infty ∞ -category C [ W − 1 ] ∞ C[W^{-1}]^{\infty} C [ W − 1 ] ∞ and an ∞ \infty ∞ -functor λ : N ( C ) → C [ W − 1 ] ∞ \lambda:\mathcal{N}(C)\to C[W^{-1}]^{\infty} λ : N ( C ) → C [ W − 1 ] ∞ , such that if f ∈ W f\in W f ∈ W then λ ( f ) \lambda(f) λ ( f ) is an equivalence in C [ W − 1 ] ∞ C[W^{-1}]^{\infty} C [ W − 1 ] ∞ , and λ \lambda λ has the universal property that whenever F : N ( C ) → D F:\mathcal{N}(C)\to\mathcal{D} F : N ( C ) → D is an ∞ \infty ∞ -functor which maps W W W to equivalences in D \mathcal{D} D , then we can factor F F F as
N ( C ) → Π l o c ∞ C [ W − 1 ] ∞ F ~ ↓ ↓ G D = D \begin{CD}
\mathcal{N}(C) @>{\Pi^{\infty}_{\mathrm{loc}}}>> C[W^{-1}]^{\infty}\\
@V{\widetilde{F}}VV @VV{G}V\\
\mathcal{D} @= \mathcal{D}
\end{CD} N ( C ) F ↓ ⏐ D Π loc ∞ C [ W − 1 ] ∞ ↓ ⏐ G D
for some G : C [ W − 1 ] ∞ → D G:C[W^{-1}]^{\infty}\to\mathcal{D} G : C [ W − 1 ] ∞ → D unique up to a contractible choice. Dwyer–Kan localizations C [ W − 1 ] ∞ C[W^{-1}]_{\infty} C [ W − 1 ] ∞ always exist, and are unique up to Joyal equivalence. We can take C [ W − 1 ] ∞ C[W^{-1}]_{\infty} C [ W − 1 ] ∞ to have the same objects as C C C .
Starting from an ordinary category of A ∞ A_{\infty} A ∞ -categories such as A ∞ C a t R u n A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un or A ∞ C a t R c u A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R cu , the obvious notions of weak equivalence are homotopy equivalences or quasi-equivalences from Definition 1.8. If R R R is a field, and if we restrict to A ∞ A_{\infty} A ∞ -categories without zero objects in the unital case, then these are the same, but in general { homotopy equivalences } ⊊ { quasi-equivalences } \{\text{homotopy equivalences}\} \subsetneq\{\text{quasi-equivalences}\} { homotopy equivalences } ⊊ { quasi-equivalences } . We will write H E , Q E \mathit{HE},\mathit{QE} HE , QE to denote the classes of homotopy equivalences and quasi-equivalences in our category, with subscripts un, cu, … to indicate the category. Thus we could for example consider the ∞ \infty ∞ -categories
A ∞ C a t R u n [ H E u n − 1 ] ∞ , A ∞ C a t R u n [ Q E u n − 1 ] ∞ , A ∞ C a t R c u [ Q E c u − 1 ] ∞ , D G C a t R [ Q E d g − 1 ] ∞ . (18) \begin{aligned}
&A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathit{HE}_{\mathrm{un}}^{-1}]_{\infty},\quad A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathit{QE}_{\mathrm{un}}^{-1}]_{\infty},\\
&A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R}[\mathit{QE}_{\mathrm{cu}}^{-1}]_{\infty},\quad\mathrm{DGCat}_{R}[\mathit{QE}_{\mathrm{dg}}^{-1}]_{\infty}. \tag*{(18)}
\end{aligned} A ∞ Cat R un [ HE un − 1 ] ∞ , A ∞ Cat R un [ QE un − 1 ] ∞ , A ∞ Cat R cu [ QE cu − 1 ] ∞ , DGCat R [ QE dg − 1 ] ∞ . (18)
We do not consider localizations of A ∞ C a t R A_{\infty}\mathrm{Cat}_{R} A ∞ Cat R , as homotopy equivalences and quasi-equivalences are not well behaved for non-unital A ∞ A_{\infty} A ∞ -categories. The previous literature on ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories [11, 12, 24, 72, 73, 76, 88], discussed in §4.10, generally studies A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathit{QE}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ QE un − 1 ] ∞ and A ∞ C a t R c u [ Q E c u − 1 ] ∞ A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R}[\mathit{QE}_{\mathrm{cu}}^{-1}]_{\infty} A ∞ Cat R cu [ QE cu − 1 ] ∞ .
Some important differences between our ∞ \infty ∞ -categories (13) , and localized ∞ \infty ∞ -categories such as (18) , are that our ∞ \infty ∞ -categories are completely explicit, and it is a lot of work to define them and prove they are ∞ \infty ∞ -categories. For localized ∞ \infty ∞ -categories, their definition and existence is trivial (given the existence of Dwyer–Kan localizations), but they are non-explicit, and the goal is to prove something useful about them.
∞ \infty ∞ -bicategories of A ∞ A_{\infty} A ∞ -categories
Informally, an ( ∞ , n ) (\infty,n) ( ∞ , n ) -category is some kind of higher category with a notion of k k k -morphism for all k = 1 , 2 , … k=1,2,\ldots k = 1 , 2 , … , such that all k k k -morphisms are invertible for k > n k>n k > n . Our ∞ \infty ∞ -categories are a kind of ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category. In §2.8 we explain ∞ \infty ∞ -bicategories, following Lurie [62, §4], which are a model for ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories.
An ∞ \infty ∞ -bicategory ( S , T ) (S,T) ( S , T ) is a simplicial set S S S with a distinguished subset T ⊂ S ( [ 2 ] ) T \subset S([2]) T ⊂ S ([ 2 ]) of 2-simplices in S S S , called thin 2-simplices. These S , T S,T S , T satisfy more complicated horn-filling conditions than ∞ \infty ∞ -categories do. Roughly, we think of a 2-simplex t : Δ 2 → S t:\Delta^2 \to S t : Δ 2 → S as a diagram
Figure 2.
where f : X → Y f:X\to Y f : X → Y , g : Y → Z g:Y\to Z g : Y → Z and h : X → Z h:X\to Z h : X → Z are 1-morphisms in S S S and η : g ∘ f ⇒ h \eta:g\circ f\Rightarrow h η : g ∘ f ⇒ h is a 2-morphism. Then t t t is thin if η \eta η is a 2-isomorphism. We write Core ( S , T ) ⊂ S \operatorname{Core}(S,T)\subset S Core ( S , T ) ⊂ S for the simplicial subset of S S S consisting of N N N -simplices all of whose 2-simplices are thin. Then Core ( S , T ) \operatorname{Core}(S,T) Core ( S , T ) is an ∞ \infty ∞ -category.
In §7.2 we define our ∞ \infty ∞ -category A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ of unital A ∞ A_{\infty} A ∞ -categories over R R R by first defining a simplicial set A ^ ∞ C a t R u n , ∞ \widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ of unital A ∞ A_{\infty} A ∞ -categories, and then defining a simplicial subset A ∞ C a t R u n , ∞ ⊂ A ^ ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\subset\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ in which all 2-simplices satisfy an invertibility condition. Define T ⊂ A ^ ∞ C a t R u n , ∞ ( [ 2 ] ) T\subset\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}([2]) T ⊂ A ∞ Cat R un , ∞ ([ 2 ]) to be the subset of 2-simplices satisfying this condition. Then Theorem 7.17 shows that ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},T) ( A ∞ Cat R un , ∞ , T ) is an ∞ \infty ∞ -bicategory, and A ∞ C a t R u n , ∞ = Core ( A ^ ∞ C a t R u n , ∞ , T ) \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}=\operatorname{Core}(\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},T) A ∞ Cat R un , ∞ = Core ( A ∞ Cat R un , ∞ , T ) . The basic difference is that 2-morphisms in ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},T) ( A ∞ Cat R un , ∞ , T ) are unital A ∞ A_{\infty} A ∞ -natural transformations, but 2-morphisms in A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ are A ∞ A_{\infty} A ∞ -natural isomorphisms.
Theorems 6.22 and 8.10 give analogues of this for ∞ \infty ∞ -bicategories of A ∞ A_{\infty} A ∞ -algebras, and of (weakly) filtered A ∞ A_{\infty} A ∞ -categories. These are all interesting as examples of ∞ \infty ∞ -bicategories which appear ‘in nature’.
As we explain in §1.6.2 and §4.11, Faonte [24, §3.4] defines another kind of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category of unital A ∞ A_{\infty} A ∞ -categories over R R R . We expect, but do not prove, that our construction and Faonte’s are equivalent in a suitable sense.
Apart from proving the existence of these ∞ \infty ∞ -bicategories, we do not develop their theory in this book. Perhaps they will find some applications in future. For example, they may be useful in problems involving A ∞ A_{\infty} A ∞ -natural transformations which are not A ∞ A_{\infty} A ∞ -natural isomorphisms. Also, the full ∞ \infty ∞ -bisubcategory ( D G C a t ^ R u n , ∞ , T d g ) ⊂ ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathrm{DGCat}}_{R}^{\mathrm{un},\infty},T^{\mathrm{dg}})\subset(\widehat{\mathrm{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},T) ( DGCat R un , ∞ , T dg ) ⊂ ( A ∞ Cat R un , ∞ , T ) with objects dg-categories is an ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category of dg-categories over R R R , which may be of interest.
Theorems about our ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories
We now review some of the results of Chapter 9, which generalizes various results about A ∞ A_{\infty} A ∞ -categories to the ∞ \infty ∞ -categorical level.
Mapping spaces
If C \mathcal{C} C is an ∞ \infty ∞ -category and x , y x,y x , y are objects of C \mathcal{C} C , as in §2.3 we can define the space of left morphisms Hom C L ( x , y ) \operatorname{Hom}^{L}_{\mathcal{C}}(x,y) Hom C L ( x , y ) , a Kan complex which is a particular model for the mapping space Hom C ( x , y ) \operatorname{Hom}_{\mathcal{C}}(x,y) Hom C ( x , y ) up to homotopy equivalence. Being able to compute mapping spaces Hom C ( x , y ) \operatorname{Hom}_{\mathcal{C}}(x,y) Hom C ( x , y ) explicitly is useful in several ways, and a sign that we understand the ∞ \infty ∞ -category C \mathcal{C} C . Here is part of Theorem 9.1 .
Theorem 1.14. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R . Then there is a canonical strict isomorphism of Kan complexes
Hom A ∞ C a t R u n , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ u n ( A , B ) ) K a n . (19) \operatorname{Hom}^{L}_{\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}}(\mathcal{A},\mathcal{B})\cong N_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))^{\mathrm{Kan}}. \tag*{(19)} Hom A ∞ Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ un ( A , B ) ) Kan . (19)
Here Hom A ∞ C a t R u n , ∞ L ( A , B ) \operatorname{Hom}^{L}_{\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un , ∞ L ( A , B ) is the space of left morphisms in A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ from §2.3, one model for the mapping space Hom A ∞ C a t R u n , ∞ ( A , B ) \operatorname{Hom}_{\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un , ∞ ( A , B ) in the ∞ \infty ∞ -category A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , and N A ∞ ( Fun A ∞ u n ( A , B ) ) N_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})) N A ∞ ( Fun A ∞ un ( A , B )) is the simplicial nerve from Definition 1.9 of the A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) from Definition 1.7 , and ( ⋯ ) K a n (\cdots)^{\mathrm{Kan}} ( ⋯ ) Kan denotes the maximal Kan complex in an ∞ \infty ∞ -category. Analogues of (19) hold for A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , A ∞ C a t R w f , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{wf},\infty} A ∞ Cat R wf , ∞ in the filtered case.
The author views this as evidence A ∞ C a t R u n , ∞ \mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , A ∞ f i C a t R u n , ∞ \mathbf{A}_{\infty}^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ fi Cat R un , ∞ , A ∞ w f C a t R u n , ∞ \mathbf{A}_{\infty}^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ wf Cat R un , ∞ are the ‘correct’ definitions of ∞ \infty ∞ -categories generalizing A ∞ C a t R u n \mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , A ∞ f i C a t R u n \mathbf{A}_{\infty}^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ fi Cat R un . In particular, (19) holds not just up to homotopy equivalence, but on the nose.
The next theorem, explained in §4.10, is proved by Tanaka [88, Th. 3.37], building on work by Canonaco–Ornaghi–Stellari [12, Th. C] and Toën [90].
Theorem 1.15 (Tanaka [88]). Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R , and Φ : A ′ → A \Phi:\mathcal{A}^{\prime}\to\mathcal{A} Φ : A ′ → A be a quasi-equivalence with A ′ \mathcal{A}^{\prime} A ′ h h h -projective with split units (such Φ \Phi Φ exist). Then we have a homotopy equivalence
Hom A ∞ C a t R u n [ Q E u n − 1 ] ∞ L ( A , B ) ≃ N A ∞ ( Fun A ∞ u n ( A ′ , B ) ) K a n , \operatorname{Hom}^{L}_{\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathrm{QE}_{\mathrm{un}}^{-1}]^{\infty}}(\mathcal{A},\mathcal{B}) \simeq\mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}^{\prime},\mathcal{B})\right)^{\mathrm{Kan}}, Hom A ∞ Cat R un [ QE un − 1 ] ∞ L ( A , B ) ≃ N A ∞ ( Fun A ∞ un ( A ′ , B ) ) Kan , where Hom A ∞ C a t R u n [ Q E u n − 1 ] ∞ L ( A , B ) \operatorname{Hom}^{L}_{\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathrm{QE}_{\mathrm{un}}^{-1}]^{\infty}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un [ QE un − 1 ] ∞ L ( A , B ) is the mapping space in the localized ∞ \infty ∞ -category A ∞ C a t R u n [ Q E u n − 1 ] ∞ \mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathrm{QE}_{\mathrm{un}}^{-1}]^{\infty} A ∞ Cat R un [ QE un − 1 ] ∞ , for Q E u n \mathrm{QE}_{\mathrm{un}} QE un the class of unitary quasi-equivalences in A ∞ C a t R u n \mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , and N A ∞ ( Fun A ∞ u n ( A ′ , B ) ) K a n \mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}^{\prime},\mathcal{B})\right)^{\mathrm{Kan}} N A ∞ ( Fun A ∞ un ( A ′ , B ) ) Kan is the maximal Kan complex in the simplicial nerve of the A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( A ′ , B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}^{\prime},\mathcal{B}) Fun A ∞ un ( A ′ , B ) .
Here is Theorem 9.4(d) below, which is new.
Theorem 1.16. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R . Then we have a homotopy equivalence
Hom A ∞ C a t R u n [ H E u n − 1 ] ∞ L ( A , B ) ≃ N A ∞ ( Fun A ∞ u n ( A , B ) ) K a n , \operatorname{Hom}^{L}_{\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathrm{HE}_{\mathrm{un}}^{-1}]^{\infty}}(\mathcal{A},\mathcal{B}) \simeq\mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\right)^{\mathrm{Kan}}, Hom A ∞ Cat R un [ HE un − 1 ] ∞ L ( A , B ) ≃ N A ∞ ( Fun A ∞ un ( A , B ) ) Kan , where H E u n \mathrm{HE}_{\mathrm{un}} HE un is the class of unital homotopy equivalences in A ∞ C a t R u n \mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un .
Equivalences of ∞ \infty ∞ -categories
We consider two ∞ \infty ∞ -categories C , D \mathcal{C},\mathcal{D} C , D to be ‘the same’ if there is a (Joyal) equivalence F ~ : C → D \widetilde{F}:\mathcal{C}\to\mathcal{D} F : C → D . We summarize some of our results on equivalences between ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories. They are proved by combining the computations of mapping spaces in § 1.5.1 with the fact in Theorem 2.20 that F ~ : C → D \widetilde{F}:\mathcal{C}\to\mathcal{D} F : C → D is an equivalence if and only if F ~ \widetilde{F} F induces homotopy equivalences of mapping spaces Hom C ( x , y ) → Hom D ( F ~ ( x ) , F ~ ( y ) ) \operatorname{Hom}_{\mathcal{C}}(x,y)\to\operatorname{Hom}_{\mathcal{D}}(\widetilde{F}(x),\widetilde{F}(y)) Hom C ( x , y ) → Hom D ( F ( x ) , F ( y )) , and Ho ( F ~ ) : Ho ( C ) → Ho ( D ) \operatorname{Ho}(\widetilde{F}):\operatorname{Ho}(\mathcal{C})\to\operatorname{Ho}(\mathcal{D}) Ho ( F ) : Ho ( C ) → Ho ( D ) is an equivalence of ordinary categories. The next theorem combines parts of Theorems 9.6 and 9.7 and Corollary 9.8.
Theorem 1.18. Let R R R be a commutative ring.
(a) Consider the categories with weak equivalences ( A ∞ Cat R u n , h s , H E u n , h s ) (A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}},\mathcal{H}\mathcal{E}_{\mathrm{un,hs}}) ( A ∞ Cat R un , hs , H E un , hs ) and ( A ∞ Cat R u n , Q E u n ) (A_\infty\operatorname{Cat}_{R}^{\mathrm{un}},\mathcal{Q}\mathcal{E}_{\mathrm{un}}) ( A ∞ Cat R un , Q E un ) . The inclusion inc : A ∞ Cat R u n , h s ↪ A ∞ Cat R u n \operatorname{inc}:A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}}\hookrightarrow A_\infty\operatorname{Cat}_{R}^{\mathrm{un}} inc : A ∞ Cat R un , hs ↪ A ∞ Cat R un maps H E h s \mathcal{H}\mathcal{E}_{\mathrm{hs}} H E hs to Q E u n \mathcal{Q}\mathcal{E}_{\mathrm{un}} Q E un , and so induces an ∞ \infty ∞ -functor of localized ∞ \infty ∞ -categories
inc ∗ : A ∞ Cat R u n , h s [ H E u n , h s − 1 ] ∞ ⟶ A ∞ Cat R u n [ Q E u n − 1 ] ∞ . \operatorname{inc}_{*}:A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}}[\mathcal{H}\mathcal{E}_{\mathrm{un,hs}}^{-1}]_{\infty}\longrightarrow A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}. inc ∗ : A ∞ Cat R un , hs [ H E un , hs − 1 ] ∞ ⟶ A ∞ Cat R un [ Q E un − 1 ] ∞ . Then inc ∗ \operatorname{inc}_{*} inc ∗ is a Joyal equivalence.
(b) As in (1.13), in § 7.3 we define an explicit ∞ \infty ∞ -functor inc R u n : N ( A ∞ Cat R u n ) → A ∞ Cat R u n , ∞ \operatorname{inc}_{R}^{\mathrm{un}}:\mathcal{N}(A_\infty\operatorname{Cat}_{R}^{\mathrm{un}})\to A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} inc R un : N ( A ∞ Cat R un ) → A ∞ Cat R un , ∞ , which maps homotopy equivalences in A ∞ Cat R u n A_\infty\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un to equivalences in A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ . Thus by the universal property of ∞ \infty ∞ -category localizations, there exists an ∞ \infty ∞ -functor Φ \Phi Φ in the diagram
N ( A ∞ Cat R u n ) → Π ∞ l o c A ∞ Cat R u n [ H E u n − 1 ] ∞ inc R u n ↓ ↓ Φ A ∞ Cat R u n , ∞ = A ∞ Cat R u n , ∞ \begin{CD}
\mathcal{N}(A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}) @>{\Pi_{\infty}^{\mathrm{loc}}}>> A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}\\
@V{\operatorname{inc}_{R}^{\mathrm{un}}}VV @VV{\Phi}V\\
A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} @= A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ Π ∞ loc A ∞ Cat R un [ H E un − 1 ] ∞ ↓ ⏐ Φ A ∞ Cat R un , ∞ Then Φ : A ∞ Cat R u n [ H E u n − 1 ] ∞ → A ∞ Cat R u n , ∞ \Phi:A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}\to A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} Φ : A ∞ Cat R un [ H E un − 1 ] ∞ → A ∞ Cat R un , ∞ is a Joyal equivalence.
Furthermore, if C \mathcal{C} C is any of the full subcategories of A ∞ Cat R u n A_\infty\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un
D G C a t R A ∞ , D G C a t R A ∞ , s u , D G C a t R A ∞ , h p , D G C a t R A ∞ , h s , A ∞ Cat R u n , s u , \mathrm{DGCat}_{R}^{A_\infty},\ \mathrm{DGCat}_{R}^{A_\infty,\mathrm{su}},\ \mathrm{DGCat}_{R}^{A_\infty,\mathrm{hp}},\ \mathrm{DGCat}_{R}^{A_\infty,\mathrm{hs}},\ A_\infty\operatorname{Cat}_{R}^{\mathrm{un,su}}, DGCat R A ∞ , DGCat R A ∞ , su , DGCat R A ∞ , hp , DGCat R A ∞ , hs , A ∞ Cat R un , su , A ∞ Cat R u n , h p , A ∞ Cat R u n , h s , A ∞ △ Cat R u n , s u , A ∞ △ Cat R u n , h p , A ∞ △ Cat R u n , h s , A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hp}},\ A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}},\ A_\infty^{\triangle}\operatorname{Cat}_{R}^{\mathrm{un,su}},\ A_\infty^{\triangle}\operatorname{Cat}_{R}^{\mathrm{un,hp}},\ A_\infty^{\triangle}\operatorname{Cat}_{R}^{\mathrm{un,hs}}, A ∞ Cat R un , hp , A ∞ Cat R un , hs , A ∞ △ Cat R un , su , A ∞ △ Cat R un , hp , A ∞ △ Cat R un , hs , then the localization of C \mathcal{C} C at its homotopy equivalences is equivalent to the corresponding full ∞ \infty ∞ -subcategory of A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ .
(c) All the following ∞ \infty ∞ -categories are Joyal equivalent:
A ∞ Cat R u n , h s , ∞ , D G C a t R A ∞ , h s , ∞ , A ∞ Cat R u n [ Q E u n − 1 ] ∞ , A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs},\infty},\quad\mathrm{DGCat}_{R}^{A_\infty,\mathrm{hs},\infty},\quad A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}, A ∞ Cat R un , hs , ∞ , DGCat R A ∞ , hs , ∞ , A ∞ Cat R un [ Q E un − 1 ] ∞ , A ∞ Cat R c u [ Q E c u − 1 ] ∞ , D G C a t R [ Q E d g − 1 ] ∞ , A ∞ Cat R u n , h s [ H E u n , h s − 1 ] ∞ , A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}}[\mathcal{Q}\mathcal{E}_{\mathrm{cu}}^{-1}]_{\infty},\quad\mathrm{DGCat}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_{\infty},\quad A_\infty\operatorname{Cat}_{R}^{\mathrm{un,hs}}[\mathcal{H}\mathcal{E}_{\mathrm{un,hs}}^{-1}]_{\infty}, A ∞ Cat R cu [ Q E cu − 1 ] ∞ , DGCat R [ Q E dg − 1 ] ∞ , A ∞ Cat R un , hs [ H E un , hs − 1 ] ∞ , D G C a t R h s [ H E d g , h s − 1 ] ∞ , D G C a t R A ∞ , h s [ H E A ∞ , d g , h s − 1 ] ∞ . (20) \mathrm{DGCat}_{R}^{\mathrm{hs}}[\mathcal{H}\mathcal{E}_{\mathrm{dg,hs}}^{-1}]_{\infty},\quad\mathrm{DGCat}_{R}^{A_\infty,\mathrm{hs}}[\mathcal{H}\mathcal{E}_{A_\infty,\mathrm{dg,hs}}^{-1}]_{\infty}. \tag*{(20)} DGCat R hs [ H E dg , hs − 1 ] ∞ , DGCat R A ∞ , hs [ H E A ∞ , dg , hs − 1 ] ∞ . (20) They are all presentable ∞ \infty ∞ -categories with closed symmetric monoidal structure ⊗ ∞ \otimes_\infty ⊗ ∞ , as this is known to be true for D G C a t R [ Q E d g − 1 ] ∞ \mathrm{DGCat}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_{\infty} DGCat R [ Q E dg − 1 ] ∞ .
Remark 1.19. The ∞ \infty ∞ -categories A ∞ C a t R u n [ Q E u n − 1 ] ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ and A ∞ C a t R c u [ Q E c u − 1 ] ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{cu}}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{cu}}^{-1}]_{\infty} A ∞ Cat R cu [ Q E cu − 1 ] ∞ in (20) are considered in the literature [11, 12, 24, 72, 73, 76, 88] to be ‘the’ ∞ \infty ∞ -category of A ∞ A_{\infty} A ∞ -categories over R R R . Part (c) shows these can be identified with full ∞ \infty ∞ -subcategory A ∞ C a t R u n , h s , ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un,hs},\infty}_{R} A ∞ Cat R un , hs , ∞ of our explicit ∞ \infty ∞ -category A ∞ C a t R u n , ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ .
We do not prove any analogues of Theorem 1.18 for filtered A ∞ A_{\infty} A ∞ -categories, as some of the tools involved do not work, for example, neither cohomology categories H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) nor quasi-equivalences are defined in the filtered case.
The ∞ \infty ∞ -categorical simplicial nerve functor
Definition 1.9 defined the simplicial nerve functor N A ∞ : A ∞ C a t R u n → C a t ∞ 1 \mathcal{N}_{A_{\infty}}:\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}\to\mathop{\mathrm{Cat}}\nolimits^{1}_{\infty} N A ∞ : A ∞ Cat R un → Cat ∞ 1 , where C a t ∞ 1 ⊂ S S e t \mathop{\mathrm{Cat}}\nolimits^{1}_{\infty}\subset\mathop{\mathrm{SSet}} Cat ∞ 1 ⊂ SSet is the ordinary category of ∞ \infty ∞ -categories. To each unital A ∞ A_{\infty} A ∞ -category A \mathcal{A} A this associates an ∞ \infty ∞ -category N A ∞ ( A ) \mathcal{N}_{A_{\infty}}(\mathcal{A}) N A ∞ ( A ) with homotopy category Ho ( N A ∞ ( A ) ) ≅ H 0 ( A ) \operatorname{Ho}(\mathcal{N}_{A_{\infty}}(\mathcal{A}))\cong H^{0}(\mathcal{A}) Ho ( N A ∞ ( A )) ≅ H 0 ( A ) . That is, N A ∞ ( A ) \mathcal{N}_{A_{\infty}}(\mathcal{A}) N A ∞ ( A ) is an ∞ \infty ∞ -categorical enhancement of the cohomology category H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) .
Here is part of Theorem 9.12 and Corollary 9.13 .
Theorem 1.20. Let R R R be a commutative ring. By an explicit construction, we can define an ∞ \infty ∞ -functor N A ∞ ∞ : A ∞ C a t R u n , ∞ → C a t ∞ \mathcal{N}^{\infty}_{A_{\infty}}:\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R}\to\mathop{\mathrm{Cat}}\nolimits_{\infty} N A ∞ ∞ : A ∞ Cat R un , ∞ → Cat ∞ , where C a t ∞ \mathop{\mathrm{Cat}}\nolimits_{\infty} Cat ∞ is the ∞ \infty ∞ -category of ∞ \infty ∞ -categories, such that the following diagram of ∞ \infty ∞ -categories commutes up to homotopy:
N ( A ∞ C a t R u n ) → N ( N A ∞ ) N ( C a t ∞ 1 ) ↓ i n c R u n ⇑ ↓ ι C a t ∞ A ∞ C a t R u n , ∞ → N A ∞ ∞ C a t ∞ \begin{array}{ccc}
\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}) & \xrightarrow{\mathcal{N}(\mathcal{N}_{A_{\infty}})} & \mathcal{N}(\mathop{\mathrm{Cat}}\nolimits^{1}_{\infty}) \\
\downarrow{\mathrm{inc}^{\mathrm{un}}_{R}} & \Uparrow& \downarrow{\iota^{\mathrm{Cat}_{\infty}}} \\
\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} & \xrightarrow{\mathcal{N}^{\infty}_{A_{\infty}}} & \mathop{\mathrm{Cat}}\nolimits_{\infty}
\end{array} N ( A ∞ Cat R un ) ↓ inc R un A ∞ Cat R un , ∞ N ( N A ∞ ) ⇑ N A ∞ ∞ N ( Cat ∞ 1 ) ↓ ι Cat ∞ Cat ∞ We call N A ∞ ∞ \mathcal{N}^{\infty}_{A_{\infty}} N A ∞ ∞ the ∞ \infty ∞ -categorical simplicial nerve functor.
This N A ∞ ∞ \mathcal{N}^{\infty}_{A_{\infty}} N A ∞ ∞ restricts to an ∞ \infty ∞ -functor
N A ∞ ∞ : A ∞ △ C a t R u n , ∞ ⟶ C a t ∞ s t (21) \mathcal{N}^{\infty}_{A_{\infty}}:\mathbf{A}^{\triangle}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R}\longrightarrow\mathop{\mathrm{Cat}}\nolimits^{\mathrm{st}}_{\infty} \tag*{(21)} N A ∞ ∞ : A ∞ △ Cat R un , ∞ ⟶ Cat ∞ st (21) from the ∞ \infty ∞ -category A ∞ △ C a t R u n , ∞ \mathbf{A}^{\triangle}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} A ∞ △ Cat R un , ∞ of pretriangulated unital A ∞ A_{\infty} A ∞ -categories over R R R to the ∞ \infty ∞ -category C a t ∞ s t \mathop{\mathrm{Cat}}\nolimits^{\mathrm{st}}_{\infty} Cat ∞ st of stable ∞ \infty ∞ -categories.
‘Inner Hom’ functors
Using the equivalence A ∞ C a t R u n [ H E u n − 1 ] ∞ ≃ A ∞ C a t R u n , ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}\simeq\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} A ∞ Cat R un [ H E un − 1 ] ∞ ≃ A ∞ Cat R un , ∞ in Theorem 1.18 (b), in Theorem 9.15 we prove:
Theorem 1.21. Let R R R be a commutative ring, and Fun A ∞ u n \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}} Fun A ∞ un be as in Definition 1.7. Then there exists an ∞ \infty ∞ -functor Fun A ∞ u n , ∞ \operatorname{Fun}^{\mathrm{un},\infty}_{A_{\infty}} Fun A ∞ un , ∞ , unique up to homotopy, such that the following diagram of ∞ \infty ∞ -categories commutes up to homotopy:
N ( A ∞ C a t R u n ) o p × N ( A ∞ C a t R u n ) → N ( Fun A ∞ u n ) N ( A ∞ C a t R u n ) ↓ ( i n c R u n ) o p × i n c R u n ⇑ ↓ i n c R u n ( A ∞ C a t R u n , ∞ ) o p × A ∞ C a t R u n , ∞ → Fun A ∞ u n , ∞ A ∞ C a t R u n , ∞ (22) \begin{array}{ccc}
\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R})^{\mathrm{op}}\times\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}) & \xrightarrow{\mathcal{N}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}})} & \mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}) \\
\downarrow{(\mathrm{inc}^{\mathrm{un}}_{R})^{\mathrm{op}}\times\mathrm{inc}^{\mathrm{un}}_{R}} & \Uparrow& \downarrow{\mathrm{inc}^{\mathrm{un}}_{R}} \\
(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R})^{\mathrm{op}}\times\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} & \xrightarrow{\operatorname{Fun}^{\mathrm{un},\infty}_{A_{\infty}}} & \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} \tag*{(22)}
\end{array} N ( A ∞ Cat R un ) op × N ( A ∞ Cat R un ) ↓ ( inc R un ) op × inc R un ( A ∞ Cat R un , ∞ ) op × A ∞ Cat R un , ∞ N ( Fun A ∞ un ) ⇑ Fun A ∞ un , ∞ N ( A ∞ Cat R un ) ↓ inc R un A ∞ Cat R un , ∞ (22)
Since F u n A ∞ u n , ∞ \mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ maps between explicitly-defined ∞ \infty ∞ -categories, it is an obvious question whether we can give explicit formulae for it. The author has not managed to do this. However, in §9.4 we do define explicit ∞ \infty ∞ -functors
F u n ^ A ∞ u n , ∞ : N ( A ∞ C a t R u n ) o p × A ∞ C a t R u n , ∞ ⟶ A ∞ C a t R u n , ∞ , (23) \widehat{\mathrm{Fun}}_{A_\infty}^{\mathrm{un},\infty}:\mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}\longrightarrow A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}, \tag*{(23)} Fun A ∞ un , ∞ : N ( A ∞ Cat R un ) op × A ∞ Cat R un , ∞ ⟶ A ∞ Cat R un , ∞ , (23)
F u n ~ A ∞ u n , ∞ : ( A ∞ C a t R u n , ∞ ) o p × N ( A ∞ C a t R u n ) ⟶ A ∞ C a t R u n , ∞ . (24) \widetilde{\mathrm{Fun}}_{A_\infty}^{\mathrm{un},\infty}:(A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un}})\longrightarrow A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}. \tag*{(24)} Fun A ∞ un , ∞ : ( A ∞ Cat R un , ∞ ) op × N ( A ∞ Cat R un ) ⟶ A ∞ Cat R un , ∞ . (24)
which are homotopic to the restrictions of F u n A ∞ u n , ∞ \mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ to the domain ∞ \infty ∞ -subcategories. So for example, linking (1.22) and (1.23), in the following diagram of ∞ \infty ∞ -categories, the top rectangle commutes strictly, and the bottom commutes up to homotopy:
N ( A ∞ C a t R u n ) o p × N ( A ∞ C a t R u n ) → N ( F u n A ∞ u n , ∞ ) N ( A ∞ C a t R u n , ∞ ) i d × i n c R u n ↓ ↓ i n c R u n N ( A ∞ C a t R u n ) o p × A ∞ C a t R u n , ∞ → F u n ^ A ∞ u n , ∞ A ∞ C a t R u n , ∞ ( i n c R u n ) o p × i d ↓ ↓ i d ( A ∞ C a t R u n , ∞ ) o p × A ∞ C a t R u n , ∞ → F u n A ∞ u n , ∞ A ∞ C a t R u n , ∞ \begin{CD}
\mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un}}) @>{\mathcal{N}(\mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty})}>> \mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}) \\
@V{\mathrm{id}\times\mathrm{inc}_{R}^{\mathrm{un}}}VV @VV{\mathrm{inc}_{R}^{\mathrm{un}}}V \\
\mathcal{N}(A_\infty\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty} @>{\widehat{\mathrm{Fun}}_{A_\infty}^{\mathrm{un},\infty}}>> A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty} \\
@V{(\mathrm{inc}_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathrm{id}}VV @VV{\mathrm{id}}V \\
(A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty} @>{\mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty}}>> A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}
\end{CD} N ( A ∞ Cat R un ) op × N ( A ∞ Cat R un ) id × inc R un ↓ ⏐ N ( A ∞ Cat R un ) op × A ∞ Cat R un , ∞ ( inc R un ) op × id ↓ ⏐ ( A ∞ Cat R un , ∞ ) op × A ∞ Cat R un , ∞ N ( Fun A ∞ un , ∞ ) Fun A ∞ un , ∞ Fun A ∞ un , ∞ N ( A ∞ Cat R un , ∞ ) ↓ ⏐ inc R un A ∞ Cat R un , ∞ ↓ ⏐ id A ∞ Cat R un , ∞
For filtered A ∞ A_\infty A ∞ -categories A ∞ f i C a t R u n , ∞ A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ fi Cat R un , ∞ , A ∞ w f C a t R u n , ∞ A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ wf Cat R un , ∞ , we do not construct analogues of (1.22), as we do not have filtered analogues of the equivalence A ∞ C a t R u n [ H E u n − 1 ] ∞ ≃ A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}_{R}^{\mathrm{un}}[HE_{\mathrm{un}}^{-1}]_\infty\simeq A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un [ H E un − 1 ] ∞ ≃ A ∞ Cat R un , ∞ . However, we do define explicit analogues of (1.23)–(1.24) in the filtered case:
F u n ^ A ∞ f i u n , ∞ : N ( A ∞ f i C a t R u n ) o p × A ∞ f i C a t R u n , ∞ ⟶ A ∞ f i C a t R , 2 , 2 u n , ∞ , (25) \widehat{\mathrm{Fun}}_{A_\infty^{\mathrm{fi}}}^{\mathrm{un},\infty}:\mathcal{N}(A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\longrightarrow A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}, \tag*{(25)} Fun A ∞ fi un , ∞ : N ( A ∞ fi Cat R un ) op × A ∞ fi Cat R un , ∞ ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ , (25)
F u n ^ A ∞ w f u n , ∞ : N ( A ∞ f i C a t R u n ) o p × A ∞ w f C a t R u n , ∞ ⟶ A ∞ w f C a t R , 2 , 2 u n , ∞ , (26) \widehat{\mathrm{Fun}}_{A_\infty^{\mathrm{wf}}}^{\mathrm{un},\infty}:\mathcal{N}(A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\longrightarrow A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}, \tag*{(26)} Fun A ∞ wf un , ∞ : N ( A ∞ fi Cat R un ) op × A ∞ wf Cat R un , ∞ ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ , (26)
F u n ~ A ∞ f i u n , ∞ : ( A ∞ f i C a t R u n , ∞ ) o p × N ( A ∞ f i C a t R u n ) ⟶ A ∞ f i C a t R , 2 , 2 u n , ∞ , (27) \widetilde{\mathrm{Fun}}_{A_\infty^{\mathrm{fi}}}^{\mathrm{un},\infty}:(A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}})\longrightarrow A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}, \tag*{(27)} Fun A ∞ fi un , ∞ : ( A ∞ fi Cat R un , ∞ ) op × N ( A ∞ fi Cat R un ) ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ , (27)
F u n ~ A ∞ w f u n , ∞ : ( A ∞ w f C a t R u n , ∞ ) o p × N ( A ∞ f i C a t R u n ) ⟶ A ∞ w f C a t R , 2 , 2 u n , ∞ . (28) \widetilde{\mathrm{Fun}}_{A_\infty^{\mathrm{wf}}}^{\mathrm{un},\infty}:(A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}})\longrightarrow A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}. \tag*{(28)} Fun A ∞ wf un , ∞ : ( A ∞ wf Cat R un , ∞ ) op × N ( A ∞ fi Cat R un ) ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ . (28)
Some ∞ \infty ∞ -functors involving filtered A ∞ A_\infty A ∞ -categories
Let R R R be a complete Γ \Gamma Γ -filtered commutative ring. As in §1.3, we define ∞ \infty ∞ -categories A ∞ f i C a t R u n , ∞ A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ fi Cat R un , ∞ , A ∞ w f C a t R u n , ∞ A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ wf Cat R un , ∞ of unital filtered A ∞ A_\infty A ∞ -categories over R R R , and we define ∞ \infty ∞ -subcategories A ∞ f i C a t R , 2 , 2 u n , ∞ ⊂ A ∞ f i C a t R u n , ∞ A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}\subset A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ fi Cat R , 2 , 2 un , ∞ ⊂ A ∞ fi Cat R un , ∞ and A ∞ w f C a t R , 2 , 2 u n , ∞ ⊂ A ∞ w f C a t R u n , ∞ A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}\subset A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ wf Cat R , 2 , 2 un , ∞ ⊂ A ∞ wf Cat R un , ∞ of type ( 2 , 2 ) (2,2) ( 2 , 2 ) , in which the objects A \mathcal{A} A and 1-morphisms F \mathcal{F} F are strict (they have m 0 A = F ~ 0 = 0 m^{\mathcal{A}}_{0}=\widetilde{\mathcal{F}}_{0}=0 m 0 A = F 0 = 0 for all objects x x x ). In §8.1, we define ∞ \infty ∞ -functors
Π A ∞ f i A ∞ : A ∞ f i C a t R , 2 , 2 u n , ∞ ⟶ A ∞ C a t R u n , ∞ , (29) \Pi_{A_\infty^{\mathrm{fi}}}^{A_\infty}:A_\infty^{\mathrm{fi}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}\longrightarrow A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}, \tag*{(29)} Π A ∞ fi A ∞ : A ∞ fi Cat R , 2 , 2 un , ∞ ⟶ A ∞ Cat R un , ∞ , (29)
Π A ∞ w f A ∞ : A ∞ w f C a t R , 2 , 2 u n , ∞ ⟶ A ∞ C a t R u n , ∞ , (30) \Pi_{A_\infty^{\mathrm{wf}}}^{A_\infty}:A_\infty^{\mathrm{wf}}\mathrm{Cat}_{R,2,2}^{\mathrm{un},\infty}\longrightarrow A_\infty\mathrm{Cat}_{R}^{\mathrm{un},\infty}, \tag*{(30)} Π A ∞ wf A ∞ : A ∞ wf Cat R , 2 , 2 un , ∞ ⟶ A ∞ Cat R un , ∞ , (30)
which are analogues of (1.10). They are forgetful ∞ \infty ∞ -functors, which forget the filtration information.
In §9.4 we use {(xref)}(1.25)–(1.26) to define ∞ \infty ∞ -categorical analogues of {(xref)}(1.11):
Str ∞ = Fun ^ A ∞ f i u n , ∞ ( P u n , − ) : A f i Cat R u n , ∞ ⟶ A f i Cat R , 2 , 2 u n , ∞ , (31) \operatorname{Str}^{\infty}=\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}^{\mathrm{un},\infty}(\mathcal{P}^{\mathrm{un}},-):\mathcal{A}^{\mathrm{fi}}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\longrightarrow\mathcal{A}^{\mathrm{fi}}\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}, \tag*{(31)} Str ∞ = Fun A ∞ fi un , ∞ ( P un , − ) : A fi Cat R un , ∞ ⟶ A fi Cat R , 2 , 2 un , ∞ , (31)
Str w f ∞ = Fun ^ A ∞ w f u n , ∞ ( P u n , − ) : A w f Cat R u n , ∞ ⟶ A w f Cat R , 2 , 2 u n , ∞ . (32) \operatorname{Str}^{\infty}_{\mathrm{wf}}=\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{wf}}_{\infty}}^{\mathrm{un},\infty}(\mathcal{P}^{\mathrm{un}},-):\mathcal{A}^{\mathrm{wf}}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\longrightarrow\mathcal{A}^{\mathrm{wf}}\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}. \tag*{(32)} Str wf ∞ = Fun A ∞ wf un , ∞ ( P un , − ) : A wf Cat R un , ∞ ⟶ A wf Cat R , 2 , 2 un , ∞ . (32)
Here P u n \mathcal{P}^{\mathrm{un}} P un is the unital filtered A ∞ A_{\infty} A ∞ -category with one object ∗ \ast ∗ and Hom P u n ( ∗ , ∗ ) = R \operatorname{Hom}_{\mathcal{P}^{\mathrm{un}}}(\ast,\ast)=R Hom P un ( ∗ , ∗ ) = R . We can compose {(xref)}(1.29)–(1.30) and {(xref)}(1.31)–(1.32) as in {(xref)}(1.12).
The twist ∞ \infty ∞ -functor
Let R R R be a commutative ring. In §9.5 we define an explicit ∞ \infty ∞ -functor
Tw ∞ : A ∞ Cat R u n , ∞ ⟶ A ∞ Δ Cat R u n , ∞ , (33) \operatorname{Tw}^{\infty}:\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\longrightarrow\mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}, \tag*{(33)} Tw ∞ : A ∞ Cat R un , ∞ ⟶ A ∞ Δ Cat R un , ∞ , (33)
which is an ∞ \infty ∞ -categorical analogue of Tw : A ∞ Cat R u n → A ∞ Δ Cat R u n \operatorname{Tw}:\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}\to\mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} Tw : A ∞ Cat R un → A ∞ Δ Cat R un in Definition 1.10. The following diagram of ∞ \infty ∞ -functors strictly commutes:
N ( A ∞ Cat R u n ) → N ( Tw ) N ( A ∞ Δ Cat R u n ) inc R u n ↓ ↓ inc R u n A ∞ Cat R u n , ∞ → Tw ∞ A ∞ Δ Cat R u n , ∞ \begin{CD}
\mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) @>{\mathcal{N}(\operatorname{Tw})}>> \mathcal{N}(\mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) \\
@V{\operatorname{inc}^{\mathrm{un}}_{R}}VV @VV{\operatorname{inc}^{\mathrm{un}}_{R}}V \\
\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} @>{\operatorname{Tw}^{\infty}}>> \mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ N ( Tw ) Tw ∞ N ( A ∞ Δ Cat R un ) ↓ ⏐ inc R un A ∞ Δ Cat R un , ∞
Here is part of Theorem 9.25:
Theorem 1.22. (a) There is an ∞ \infty ∞ -natural isomorphism Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ \operatorname{Tw}^{\infty}\Rightarrow\operatorname{Tw}^{\infty}\circ\operatorname{Tw}^{\infty} Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ .
(b) Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ is left adjoint to the fully faithful inclusion ∞ \infty ∞ -functor
\operatorname{inc}:\mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\lhook\joinrel\longrightarrow\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}.
Observe that if R R R is a commutative ring we have a composition of ∞ \infty ∞ -functors
A ∞ Cat R u n , ∞ → Tw ∞ A ∞ Δ Cat R u n , ∞ → N ∞ A ∞ C ∞ s t , \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\operatorname{Tw}^{\infty}} \mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\mathcal{N}^{\mathcal{A}_{\infty}}_{\infty}} \mathcal{C}^{\mathrm{st}}_{\infty}, A ∞ Cat R un , ∞ Tw ∞ A ∞ Δ Cat R un , ∞ N ∞ A ∞ C ∞ st ,
and if R R R is a complete Γ \Gamma Γ -filtered commutative ring we have compositions
A ∞ f i Cat R u n , ∞ → Str ∞ A ∞ f i Cat R , 2 , 2 u n , ∞ → F ∞ A ∞ A ∞ Cat R u n , ∞ → Tw ∞ A ∞ Δ Cat R u n , ∞ → N ∞ A ∞ C ∞ s t , \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\operatorname{Str}^{\infty}} \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}\xrightarrow{\mathcal{F}^{\mathcal{A}_{\infty}}_{\infty}} \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\operatorname{Tw}^{\infty}} \mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\mathcal{N}^{\mathcal{A}_{\infty}}_{\infty}} \mathcal{C}^{\mathrm{st}}_{\infty}, A ∞ fi Cat R un , ∞ Str ∞ A ∞ fi Cat R , 2 , 2 un , ∞ F ∞ A ∞ A ∞ Cat R un , ∞ Tw ∞ A ∞ Δ Cat R un , ∞ N ∞ A ∞ C ∞ st ,
A ∞ w f Cat R u n , ∞ → Str w f ∞ A ∞ w f Cat R , 2 , 2 u n , ∞ → F w f A ∞ A ∞ Cat R u n , ∞ → Tw ∞ A ∞ Δ Cat R u n , ∞ → N ∞ A ∞ C ∞ s t . \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\operatorname{Str}^{\infty}_{\mathrm{wf}}} \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}\xrightarrow{\mathcal{F}^{\mathcal{A}_{\infty}}_{\mathrm{wf}}} \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\operatorname{Tw}^{\infty}} \mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\xrightarrow{\mathcal{N}^{\mathcal{A}_{\infty}}_{\infty}} \mathcal{C}^{\mathrm{st}}_{\infty}. A ∞ wf Cat R un , ∞ Str wf ∞ A ∞ wf Cat R , 2 , 2 un , ∞ F wf A ∞ A ∞ Cat R un , ∞ Tw ∞ A ∞ Δ Cat R un , ∞ N ∞ A ∞ C ∞ st .
These map (filtered) unital A ∞ A_{\infty} A ∞ -categories over R R R to stable ∞ \infty ∞ -categories, in an ∞ \infty ∞ -categorical way. For a symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) , these compositions should be used to transform the Fukaya category F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) (a unital A ∞ A_{\infty} A ∞ -category or unital filtered A ∞ A_{\infty} A ∞ -category), to the derived Fukaya category D b F ( X , ω ) D^{b}\mathcal{F}(X,\omega) D b F ( X , ω ) (a stable ∞ \infty ∞ -category, so that Ho ( D b F ( X , ω ) ) \operatorname{Ho}(D^{b}\mathcal{F}(X,\omega)) Ho ( D b F ( X , ω )) is a triangulated category).
Relation to other work in the literature
We now compare our ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories with work by other authors.
Comparison with localized ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories
As we explain in §4.10 below, Canonaco–Ornaghi–Stellari [11, 12], Oh–Tanaka [72], Ornaghi [73], Tanaka [88], and Pascaleff [76] study ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories defined by starting with the category A ∞ C a t R u n A_\infty\mathrm{Cat}^{\mathrm{un}}_R A ∞ Cat R un (or some variant), and localizing the quasi-equivalences Q E u n ⊂ Mor ( A ∞ C a t R u n ) \mathcal{QE}_{\mathrm{un}} \subset\operatorname{Mor}(A_\infty\mathrm{Cat}^{\mathrm{un}}_R) QE un ⊂ Mor ( A ∞ Cat R un ) from Definition 4.28, as in §2.6, to get an ∞ \infty ∞ -category A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_R[\mathcal{QE}_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ QE un − 1 ] ∞ .
Section 9.2, summarized in Theorem 1.18, explores the relation between our ∞ \infty ∞ -category A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ and this localized ∞ \infty ∞ -category A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_R[\mathcal{QE}_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ QE un − 1 ] ∞ , which is considered to be ‘the’ ∞ \infty ∞ -category of A ∞ A_\infty A ∞ -categories by the authors of [11, 12, 72, 73, 76, 88]. In particular, we show that A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_R[\mathcal{QE}_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ QE un − 1 ] ∞ is Joyal equivalent to the full ∞ \infty ∞ -subcategory A ∞ C a t R u n , h s , ∞ ⊂ A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\mathrm{hs},\infty}_R \subset A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , hs , ∞ ⊂ A ∞ Cat R un , ∞ with objects A ∞ A_\infty A ∞ -categories which are h-projective with split units.
Theorem 1.14 computes mapping spaces in A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ . As in Theorem 1.15, work in [12,13,24,88,90] computes mapping spaces in A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_R[\mathcal{QE}_{\mathrm{un}}^{-1}]^\infty A ∞ Cat R un [ QE un − 1 ] ∞ , and after cofibrant replacement gives the same answer.
The authors of [11,12,72,73,76,88] take the view that to form an ∞ \infty ∞ -category of unital A ∞ A_\infty A ∞ -categories by localization, one should invert quasi-equivalences , as in Definition 1.8. One reason may be the Tabuada model structure [87] on D G C a t R \mathbf{DGCat}_R DGCat R from Theorem 4.35, whose weak equivalences are quasi-equivalences. Our results suggest that instead inverting homotopy equivalences may be a fruitful thing to do. Inverting homotopy equivalences is a more gentle thing to do, as homotopy equivalences are already invertible up to A ∞ A_\infty A ∞ -natural isomorphism; so cofibrant replacement is unnecessary. As in Theorem 1.18(b), we prove an equivalence A ∞ C a t R u n [ H E u n − 1 ] ∞ ≃ A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_R[\mathcal{HE}_{\mathrm{un}}^{-1}]^\infty\simeq A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un [ HE un − 1 ] ∞ ≃ A ∞ Cat R un , ∞ .
Comparison with Faonte [24]
Section 4.11 summarizes work of Faonte [24, §3]. This is in one way the closest paper the author has found to our own work. For R R R a field of characteristic zero, Faonte defines two explicit simplicial categories of unital A ∞ A_\infty A ∞ -categories over R R R , the first a category enriched in Kan complexes, and the second a category enriched in ∞ \infty ∞ -categories (rather than just enriched in simplicial sets). These can be interpreted as ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) - and ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories of A ∞ A_\infty A ∞ -categories.
They should be compared with our ∞ \infty ∞ -category A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ from §1.3, and our ∞ \infty ∞ -bicategory A ∞ C a t ^ R u n , ∞ \widehat{A_\infty\mathrm{Cat}}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ from §1.4, respectively. In particular, Theorem 9.1(a) implies that the left mapping spaces in A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ , A ∞ C a t ^ R u n , ∞ \widehat{A_\infty\mathrm{Cat}}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ agree with those in Faonte’s simplicial categories not just up to homotopy equivalence, but by strict isomorphism.
Because of our Theorem 9.4(d), the author expects that if Faonte’s construction is done over a general ring R R R , it will yield an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category equivalent to A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{un}}]^{\infty} A ∞ Cat R un [ H E un − 1 ] ∞ rather than A ∞ C a t R u n [ Q E u n − 1 ] ∞ \mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{Q}\mathcal{E}^{-1}_{\mathrm{un}}]^{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ . It seems likely that under some suitable equivalences between different models of ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) - and ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories, Faonte’s constructions may be equivalent to our A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , A ^ ∞ C a t R u n , ∞ \widehat{\mathcal{A}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ . But we do not prove this.
Comparison with Cazassus [ref-15 ]
Section 3.7 briefly describes work by Cazassus [15, §5.2], who defines an explicit ∞ \infty ∞ -category f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R of (nonunital) A ∞ \mathcal{A}_{\infty} A ∞ -algebras over R R R . This is in a different way the closest paper the author has found to our own work since, like us in Chapter 6, Cazassus constructs an explicit simplicial set whose 0-simplices are A ∞ \mathcal{A}_{\infty} A ∞ -algebras over R R R , and shows it is an ∞ \infty ∞ -category.
The author does not yet understand the exact relation between f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R and our ∞ \infty ∞ -categories A ∞ A l g R u n , ∞ \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R} A ∞ Alg R un , ∞ , A ∞ A l g R ∞ , … \mathcal{A}_{\infty}\mathrm{Alg}^{\infty}_{R},\ldots A ∞ Alg R ∞ , … from Chapter 6. They are not equivalent. For N ≥ 2 N \ge2 N ≥ 2 , the extra data in an N N N -simplex in A ∞ A l g R u n , ∞ \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R} A ∞ Alg R un , ∞ which is not contained in lower-dimensional simplices is maps m k { 0 , … , N } : ( A 0 ) ⊗ k → A N m^{\{0,\ldots,N\}}_{k}:(A^{0})^{\otimes k}\to A^{N} m k { 0 , … , N } : ( A 0 ) ⊗ k → A N for k ≥ 0 k \ge0 k ≥ 0 , but the extra data in an N N N -simplex in f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R is finite families of maps D k , l α : ( A 0 ) ⊗ k → ( A N ) ⊗ l D^{\alpha}_{k,l}:(A^{0})^{\otimes k}\to(A^{N})^{\otimes l} D k , l α : ( A 0 ) ⊗ k → ( A N ) ⊗ l for k , l ≥ 0 k,l \ge0 k , l ≥ 0 and α ∈ F k , l \alpha\in F_{k,l} α ∈ F k , l . So it seems that f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R is ‘larger’ (has many more N N N -simplices) than A ∞ A l g R u n , ∞ , … \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R},\ldots A ∞ Alg R un , ∞ , … . Also f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R is a more linear object than A ∞ A l g R u n , ∞ \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R} A ∞ Alg R un , ∞ , as the defining equations of f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R are linear in the D k , l α D^{\alpha}_{k,l} D k , l α , but those of A ∞ A l g R u n , ∞ \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R} A ∞ Alg R un , ∞ are polynomial in the m k I m^{I}_{k} m k I .
The 2-morphisms in f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R seem not correspond to either A ∞ \mathcal{A}_{\infty} A ∞ -homotopies, or A ∞ \mathcal{A}_{\infty} A ∞ -natural isomorphisms. But they do have the property that if A ∞ \mathcal{A}_{\infty} A ∞ -functors f , g : ( A , m ) → ( B , n ) f,g:(A,m)\to(B,n) f , g : ( A , m ) → ( B , n ) are 2-isomorphic in f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R then H 0 ( f ) = H 0 ( g ) H^{0}(f)=H^{0}(g) H 0 ( f ) = H 0 ( g ) , which is necessary to use f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R to study functoriality of Morse cohomology.
The author expects that some of our ∞ \infty ∞ -categories A ∞ A l g R , K u n , ∞ , … \mathcal{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K},\ldots A ∞ Alg R , K un , ∞ , … from Chapter 6 can be embedded in f - A l g R \mathrm{f}\text{-}\mathrm{Alg}_{R} f - Alg R as ∞ \infty ∞ -subcategories. It would be interesting to know whether Cazassus’ ‘Morse complex’ ∞ \infty ∞ -functor C M ∙ : N ( M a n o p ) → f - A l g R CM^{\bullet}:N(\mathrm{Man}^{\mathrm{op}})\to\mathrm{f}\text{-}\mathrm{Alg}_{R} C M ∙ : N ( Man op ) → f - Alg R from [15, Th. C] factors via one of our ∞ \infty ∞ -categories.
On Fukaya categories, and the origin of this book
The author came to write this book as part of a long-term in-progress project [16], joint with Guillem Cazassus and Alexander Ritter, to develop a theory of (derived) Fukaya categories F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) , D b F ( X , ω ) D^{b}\mathcal{F}(X,\omega) D b F ( X , ω ) of semipositive symplectic manifolds ( X , ω ) (X,\omega) ( X , ω ) . The theory should have certain desirable characteristics we want for applications, for example, that all compact immersed Lagrangians L L L in ( X , ω ) (X,\omega) ( X , ω ) should appear as objects in F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) .
The reader should understand that although the idea of Fukaya categories has been around for 30 years, constructing a fully general theory of (derived) Fukaya categories with all the properties one would like is a hugely long and challenging task, which is far from complete. The theories currently available either involve restrictive assumptions, for example Seidel [84] for exact symplectic manifolds, or they are partial, for example Fukaya–Oh–Ohta–Ono [33] define Lagrangian Floer cohomology for embedded Lagrangians (roughly, the morphism spaces in Fukaya categories), but do not go as far as Fukaya categories.
The author hopes that this book will form part of the foundations of the fully general theory of Fukaya categories, when this emerges. We will now explain why our explicit ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories, such as A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , will be helpful in this project. See §9.6 below for more details. We will (over)simplify the discussion by not including filtered A ∞ A_\infty A ∞ -categories, as in Chapters 5 and 8.
Given a symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) , we want to construct its Fukaya category F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) , which is an A ∞ A_\infty A ∞ -category (let us say, strictly unital) over a commutative ring R R R (this could be Z \mathbb{Z} Z for ( X , ω ) (X,\omega) ( X , ω ) exact, or the Novikov ring Λ 0 n o v \Lambda^{\mathrm{nov}}_0 Λ 0 nov or Λ n o v \Lambda^{\mathrm{nov}} Λ nov for more general ( X , ω ) (X,\omega) ( X , ω ) ). Defining F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) involves difficult analysis of moduli spaces of J J J -holomorphic curves in X X X with boundary in Lagrangians L 0 , … , L k L_0,\ldots,L_k L 0 , … , L k . Then by a category-theoretic construction, starting from F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) we define the derived Fukaya category D b F ( X , ω ) D^b\mathcal{F}(X,\omega) D b F ( X , ω ) , which is a triangulated category, or (better) a stable ∞ \infty ∞ -category, and is what we really care about.
To define F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) we must choose some arbitrary auxiliary data D \mathcal{D} D , so we write it F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D . Here D \mathcal{D} D should include a choice of almost complex structure J J J on X X X compatible with ω \omega ω (and possibly domain-dependent). It could also include families of perturbations of Lagrangians as in Seidel [84], choices of Morse functions, . . . . The details depend on the set-up, and there are several different approaches in the literature.
We would like F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D and D b F ( X , ω ) D D^b\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D to be independent of the extra data D \mathcal{D} D , and thus be invariants of ( X , ω ) (X,\omega) ( X , ω ) . If D 1 , D 2 \mathcal{D}_1,\mathcal{D}_2 D 1 , D 2 are choices for D \mathcal{D} D , in general F ( X , ω ) D 1 \mathcal{F}(X,\omega)_{\mathcal{D}_1} F ( X , ω ) D 1 , D b F ( X , ω ) D 1 D^b\mathcal{F}(X,\omega)_{\mathcal{D}_1} D b F ( X , ω ) D 1 and F ( X , ω ) D 2 \mathcal{F}(X,\omega)_{\mathcal{D}_2} F ( X , ω ) D 2 , D b F ( X , ω ) D 2 D^b\mathcal{F}(X,\omega)_{\mathcal{D}_2} D b F ( X , ω ) D 2 are not isomorphic, but they are ‘homotopy equivalent’.
Our explicit ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories, such as A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , can be used to make precise how F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D depends on the auxiliary data D \mathcal{D} D . To do this, we should define a contractible Kan complex S X , ω \mathcal{S}_{X,\omega} S X , ω whose 0 0 0 -simplices are choices of data D \mathcal{D} D , and an ∞ \infty ∞ -functor F : S X , ω → A ∞ C a t R u n , ∞ F:\mathcal{S}_{X,\omega}\to A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R} F : S X , ω → A ∞ Cat R un , ∞ which maps F : D ↦ F ( X , ω ) D F:\mathcal{D}\mapsto\mathcal{F}(X,\omega)_{\mathcal{D}} F : D ↦ F ( X , ω ) D on 0 0 0 -simplices. As a contractible Kan complex is equivalent to a point in ∞ \infty ∞ -category theory, this means the F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D are independent of D \mathcal{D} D up to coherent homotopies, in an ∞ \infty ∞ -categorical sense.
It is important here that F F F maps to our explicit ∞ \infty ∞ -category A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , and not to a non-explicit localized ∞ \infty ∞ -category such as A ∞ C a t R u n [ H E u n − 1 ] ∞ A_\infty\mathrm{Cat}^{\mathrm{un}}_{R}[\mathrm{HE}^{-1}_{\mathrm{un}}]_\infty A ∞ Cat R un [ HE un − 1 ] ∞ . This is because the definition of F F F is also explicit: we generalize the construction of F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D to a structure depending on an N N N -simplex Δ N \Delta^N Δ N family of choices of D \mathcal{D} D , and this maps to an N N N -simplex in A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ .
Remarks on Set Theory issues, and conventions
Now in the theory of ∞ \infty ∞ -categories in Chapter 2, an ∞ \infty ∞ -category C ∞ \mathcal{C}_{\infty} C ∞ is a simplicial set, so its family of objects Obj C ∞ \operatorname{Obj}\mathcal{C}_{\infty} Obj C ∞ must be a set. Thus, to promote categories to ∞ \infty ∞ -categories, as we wish to do, we must restrict our attention to small categories. However, if for example we just defined A ∞ Alg R A_{\infty}\operatorname{Alg}_{R} A ∞ Alg R to be the category of all A ∞ A_{\infty} A ∞ -algebras over a ring R R R , it would not be a small category.
There is a standard way of dealing with this problem: as in Grothendieck–Verdier [40, §0–§1] and Bourbaki [8], we fix a ‘Grothendieck universe’ U \mathcal{U} U , a very large set in which all of mathematics can be performed, and we assume all sets in our theory are subsets of U \mathcal{U} U (that is, they are ‘U \mathcal{U} U -small’). See Lurie [60, §1.2.15, §5.4] for a discussion of these issues in the context of ∞ \infty ∞ -categories.
This book raises no new Set Theory problems not already discussed in [8, 40, 60] and many other places. So we will just fix a Grothendieck universe U \mathcal{U} U , assume implicitly from now on that all structures in this book are U \mathcal{U} U -small, and not worry about the Set Theory issues. (Actually, for some purposes this is an oversimplification: there are constructions which require one to pass from a universe U \mathcal{U} U to a larger one U ′ \mathcal{U}' U ′ . We will ignore this.)
Use of generative AI. This project began in 2022-23, and was 90% finished before AI began to challenge human mathematicians. In late 2026 I had completed proofs of all the major theorems except Theorems 9.6 and 9.7. I had several conversations with ChatGPT Edu about ∞ \infty ∞ -categories and issues in the proofs, in general terms, without uploading any part of this document. ChatGPT was really useful at helping me understand colimits in ∞ \infty ∞ -categories and Arakawa–Carranza–Kapulkin [1] better than I had before. It also suggested references which helped me finish the proofs: for Theorem 9.6 it pointed out Holstein [43], and for Theorem 9.7 it located results I wanted in Lurie [60, 63], including the ‘last vertex map’. Apart from these, all results, proofs, and mistakes in this book were written by a human.
Acknowledgements. I would like to thank Mohammed Abouzaid, Denis Auroux, Lukas Brantner, Alberto Canonaco, Guillem Cazassus, Jason Lotay, Mattia Ornaghi, Alexander Ritter, Paolo Stellari, Hiro Lee Tanaka, and ChatGPT Edu for helpful conversations. This research was supported by EPSRC grant EP/T012749/1. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript (AAM) version arising from this submission.
Simplicial sets and ∞ \infty ∞ -categories
Simplicial sets
For an elementary introduction to simplicial sets we recommend Friedman [26], and for foundational material see Goerss–Jardine [38, §I].
Definition 2.1. The simplex category Δ \Delta Δ has objects sets [ N ] = { 0 , 1 , … , N } [N] = \{0,1,\ldots,N\} [ N ] = { 0 , 1 , … , N } for N ≥ 0 N \ge0 N ≥ 0 and morphisms f : [ N ] → [ M ] f : [N] \to[M] f : [ N ] → [ M ] the order-preserving functions, that is, if $0 \le i \le j \le N$ then f ( i ) ≤ f ( j ) f(i) \le f(j) f ( i ) ≤ f ( j ) .
A simplicial set is a functor S : Δ o p → S e t S : \Delta^{\mathrm{op}} \to\mathrm{Set} S : Δ op → Set , where S e t \mathrm{Set} Set is the category of sets. A morphism of simplicial sets f : S → T f : S \to T f : S → T , or just a map of simplicial sets , is a natural transformation of functors f : S ⇒ T f : S \Rightarrow T f : S ⇒ T . This makes simplicial sets into a category S S e t \mathrm{SSet} SSet . It is complete and cocomplete, and (co)limits may be defined objectwise in Δ o p \Delta^{\mathrm{op}} Δ op by taking (co)limits in S e t \mathrm{Set} Set .
Define morphisms in Δ \Delta Δ , the face maps δ N , i : [ N − 1 ] → [ N ] \delta^{N,i} : [N-1] \to[N] δ N , i : [ N − 1 ] → [ N ] and degeneracy maps σ N , i : [ N + 1 ] → [ N ] \sigma^{N,i} : [N+1] \to[N] σ N , i : [ N + 1 ] → [ N ] for i = 0 , … , N i = 0,\ldots,N i = 0 , … , N by
δ N , i ( j ) = { j , j < i , j + 1 , j ≥ i , σ N , i ( j ) = { j , j ≤ i , j − 1 , j > i . \delta^{N,i}(j) =
\begin{cases}
j, & j < i,\\
j+1, & j \ge i,
\end{cases}
\qquad
\sigma^{N,i}(j) =
\begin{cases}
j, & j \le i,\\
j-1, & j > i.
\end{cases} δ N , i ( j ) = { j , j + 1 , j < i , j ≥ i , σ N , i ( j ) = { j , j − 1 , j ≤ i , j > i .
That is, δ N , i \delta^{N,i} δ N , i misses i i i , and σ N , i \sigma^{N,i} σ N , i repeats i i i . These satisfy the identities
δ N , j ∘ δ N − 1 , i = δ N , i ∘ δ N − 1 , j − 1 , 0 ≤ i < j ≤ N , σ N , j ∘ σ N + 1 , i = σ N , i ∘ σ N + 1 , j + 1 , 0 ≤ i ≤ j ≤ N , σ N , j ∘ δ N + 1 , i = δ N , i ∘ σ N − 1 , j − 1 , 0 ≤ i < j ≤ N , σ N − 1 , j ∘ δ N , i = δ N − 1 , i − 1 ∘ σ N − 2 , j , 0 ≤ j < j + 1 < i ≤ N , σ N , j ∘ δ N + 1 , i = i d , i = j or j + 1. (34) \begin{aligned}
\delta^{N,j} \circ\delta^{N-1,i} &= \delta^{N,i} \circ\delta^{N-1,j-1}, && 0 \le i < j \le N,\\
\sigma^{N,j} \circ\sigma^{N+1,i} &= \sigma^{N,i} \circ\sigma^{N+1,j+1}, && 0 \le i \le j \le N,\\
\sigma^{N,j} \circ\delta^{N+1,i} &= \delta^{N,i} \circ\sigma^{N-1,j-1}, && 0 \le i < j \le N,\\
\sigma^{N-1,j} \circ\delta^{N,i} &= \delta^{N-1,i-1} \circ\sigma^{N-2,j}, && 0 \le j < j+1 < i \le N,\\
\sigma^{N,j} \circ\delta^{N+1,i} &= \mathrm{id}, && i=j \text{ or } j+1. \tag*{(34)}
\end{aligned} δ N , j ∘ δ N − 1 , i σ N , j ∘ σ N + 1 , i σ N , j ∘ δ N + 1 , i σ N − 1 , j ∘ δ N , i σ N , j ∘ δ N + 1 , i = δ N , i ∘ δ N − 1 , j − 1 , = σ N , i ∘ σ N + 1 , j + 1 , = δ N , i ∘ σ N − 1 , j − 1 , = δ N − 1 , i − 1 ∘ σ N − 2 , j , = id , 0 ≤ i < j ≤ N , 0 ≤ i ≤ j ≤ N , 0 ≤ i < j ≤ N , 0 ≤ j < j + 1 < i ≤ N , i = j or j + 1. (34)
The category Δ \Delta Δ is generated by the δ N , i , σ N , i \delta^{N,i}, \sigma^{N,i} δ N , i , σ N , i subject only to the relations (2.1).
Now let S : Δ o p → S e t S:\Delta^{\mathrm{op}}\to\mathrm{Set} S : Δ op → Set be a simplicial set. Write S N = S ( [ N ] ) S_N=S([N]) S N = S ([ N ]) , and call S N S_N S N the set of N N N -simplices of S S S . Define face maps d N , i = S ( δ N , i ) : S N → S N − 1 d_{N,i}=S(\delta^{N,i}):S_N\to S_{N-1} d N , i = S ( δ N , i ) : S N → S N − 1 and degeneracy maps s N , i = S ( σ N , i ) : S N → S N + 1 s_{N,i}=S(\sigma^{N,i}):S_N\to S_{N+1} s N , i = S ( σ N , i ) : S N → S N + 1 . Reversing directions of morphisms, these satisfy
d N − 1 , i ∘ d N , j = d N − 1 , j − 1 ∘ d N , i , 0 ≤ i < j ≤ N , s N + 1 , i ∘ s N , j = s N + 1 , j + 1 ∘ s N , i , 0 ≤ i ≤ j ≤ N , d N + 1 , i ∘ s N , j = s N − 1 , j − 1 ∘ d N , i , 0 ≤ i < j ≤ N , d N , i ∘ s N − 1 , j = s N − 2 , j ∘ d N − 1 , i − 1 , 0 ≤ j < j + 1 < i ≤ N , d N + 1 , i ∘ s N , j = id , i = j or j + 1. (35) \begin{aligned}
d_{N-1,i}\circ d_{N,j}&=d_{N-1,j-1}\circ d_{N,i}, &&0\le i<j\le N,\\
s_{N+1,i}\circ s_{N,j}&=s_{N+1,j+1}\circ s_{N,i}, &&0\le i\le j\le N,\\
d_{N+1,i}\circ s_{N,j}&=s_{N-1,j-1}\circ d_{N,i}, &&0\le i<j\le N,\\
d_{N,i}\circ s_{N-1,j}&=s_{N-2,j}\circ d_{N-1,i-1}, &&0\le j<j+1<i\le N,\\
d_{N+1,i}\circ s_{N,j}&=\operatorname{id}, &&i=j\text{ or }j+1. \tag*{(35)}
\end{aligned} d N − 1 , i ∘ d N , j s N + 1 , i ∘ s N , j d N + 1 , i ∘ s N , j d N , i ∘ s N − 1 , j d N + 1 , i ∘ s N , j = d N − 1 , j − 1 ∘ d N , i , = s N + 1 , j + 1 ∘ s N , i , = s N − 1 , j − 1 ∘ d N , i , = s N − 2 , j ∘ d N − 1 , i − 1 , = id , 0 ≤ i < j ≤ N , 0 ≤ i ≤ j ≤ N , 0 ≤ i < j ≤ N , 0 ≤ j < j + 1 < i ≤ N , i = j or j + 1. (35)
As Δ \Delta Δ is generated by the δ N , i , σ N , i \delta^{N,i},\sigma^{N,i} δ N , i , σ N , i subject to (2.1), to define a simplicial set S S S it is enough to give sets S N = S ( Δ N ) S_N=S(\Delta_N) S N = S ( Δ N ) for N ≥ 0 N\ge0 N ≥ 0 and maps d N , i : S N → S N − 1 d_{N,i}:S_N\to S_{N-1} d N , i : S N → S N − 1 and s N , i : S N → S N + 1 s_{N,i}:S_N\to S_{N+1} s N , i : S N → S N + 1 for 0 ≤ i ≤ N 0\le i\le N 0 ≤ i ≤ N satisfying (2.2), and all the other morphisms S ( f ) S(f) S ( f ) in S S S for f : [ k ] → [ l ] f:[k]\to[l] f : [ k ] → [ l ] can be written as compositions of the d N , i , s N , i d_{N,i},s_{N,i} d N , i , s N , i . We also write d N , i S , s N , i S d^S_{N,i},s^S_{N,i} d N , i S , s N , i S for d N , i , s N , i d_{N,i},s_{N,i} d N , i , s N , i if we need to specify the simplicial set S S S .
This gives us a way to draw a picture of a simplicial set:
S 0 → s 0 , 0 S 1 → s 1 , 0 S 2 → s 2 , 0 S 3 → ⋯ d 1 , 0 ↑ d 2 , 0 ↑ d 3 , 0 ↑ d 1 , 1 ↓ d 2 , 1 ↓ d 3 , 1 ↓ s 1 , 1 ↑ s 2 , 1 ↑ d 3 , 2 ↑ d 2 , 2 ↓ s 2 , 2 ↓ s 3 , 3 ↓ (36) \begin{CD}
S_0 @>{s_{0,0}}>> S_1 @>{s_{1,0}}>> S_2 @>{s_{2,0}}>> S_3 @>>> \cdots\\
@A{d_{1,0}}AA @A{d_{2,0}}AA @A{d_{3,0}}AA\\
@V{d_{1,1}}VV @V{d_{2,1}}VV @V{d_{3,1}}VV\\
@A{s_{1,1}}AA @A{s_{2,1}}AA @A{d_{3,2}}AA\\
@V{d_{2,2}}VV @V{s_{2,2}}VV @V{s_{3,3}}VV \tag*{(36)}
\end{CD} S 0 d 1 , 0 ⏐ ↑ s 1 , 1 ⏐ ↑ s 0 , 0 d 1 , 1 ↓ ⏐ d 2 , 2 ↓ ⏐ S 1 d 2 , 0 ⏐ ↑ s 2 , 1 ⏐ ↑ s 1 , 0 d 2 , 1 ↓ ⏐ s 2 , 2 ↓ ⏐ S 2 d 3 , 0 ⏐ ↑ d 3 , 2 ⏐ ↑ s 2 , 0 d 3 , 1 ↓ ⏐ s 3 , 3 ↓ ⏐ S 3 ⋯ (36)
An N N N -simplex Σ ∈ S N \Sigma\in S_N Σ ∈ S N is called degenerate if it lies in the image of s N − 1 , i : S N − 1 → S N s_{N-1,i}:S_{N-1}\to S_N s N − 1 , i : S N − 1 → S N for some i = 0 , … , N − 1 i=0,\ldots,N-1 i = 0 , … , N − 1 , and nondegenerate otherwise.
For each Σ ∈ S N \Sigma\in S_N Σ ∈ S N there exist 0 ≤ M ≤ N 0\le M\le N 0 ≤ M ≤ N , a surjective morphism α : [ N ] → [ M ] \alpha:[N]\to[M] α : [ N ] → [ M ] in Δ \Delta Δ , and a nondegenerate M M M -simplex Σ ~ ∈ S M \widetilde{\Sigma}\in S_M Σ ∈ S M with Σ = S ( α ) ( Σ ~ ) \Sigma=S(\alpha)(\widetilde{\Sigma}) Σ = S ( α ) ( Σ ) , and M , α , Σ ~ M,\alpha,\widetilde{\Sigma} M , α , Σ are unique. Here Σ \Sigma Σ is nondegenerate if M ~ = N \widetilde{M}=N M = N , α = id [ N ] \alpha=\operatorname{id}_{[N]} α = id [ N ] and Σ = Σ ~ \Sigma=\widetilde{\Sigma} Σ = Σ .
A map of simplicial sets f : S → T f:S\to T f : S → T is equivalent to maps f N : S N → T N f_N:S_N\to T_N f N : S N → T N for N = 0 , 1 , 2 , … N=0,1,2,\ldots N = 0 , 1 , 2 , … commuting with the face and degeneracy maps d N , i , s N , i d_{N,i},s_{N,i} d N , i , s N , i .
If S , T S,T S , T are simplicial sets, we call S S S a simplicial subset of T T T , written S ⊆ T S\subseteq T S ⊆ T , if S N ⊆ T N S_N\subseteq T_N S N ⊆ T N for N = 0 , 1 , … N=0,1,\ldots N = 0 , 1 , … and the face and degeneracy maps satisfy d N , i S = d N , i T ∣ S N d^S_{N,i}=d^T_{N,i}|_{S_N} d N , i S = d N , i T ∣ S N and s N , i S = s N , i T ∣ S N s^S_{N,i}=s^T_{N,i}|_{S_N} s N , i S = s N , i T ∣ S N for all N , i N,i N , i . Inclusion inc : S ↪ T \operatorname{inc}:S\hookrightarrow T inc : S ↪ T is a map of simplicial sets.
Now let C \mathcal{C} C be a category. A simplicial object in C \mathcal{C} C is a functor F : Δ o p → C F:\Delta^{\mathrm{op}}\to\mathcal{C} F : Δ op → C . Thus a simplicial set is a simplicial object in S e t \mathrm{Set} Set . One often abbreviates ‘simplicial object in things’ to ‘simplicial thing’. To pass from classical Algebraic Geometry to Derived Algebraic Geometry, one often replaces objects by simplicial objects. For example, a simplicial commutative K K K -algebra (simplicial object in A l g K \mathrm{Alg}_K Alg K ) is a functor A : Δ o p → A l g K A:\Delta^{\mathrm{op}}\to\mathrm{Alg}_K A : Δ op → Alg K , and these are the appropriate notion of ‘derived commutative algebra’ used to define derived K K K -schemes.
We can draw a picture of a simplicial object in C \mathcal{C} C as a diagram (2.3), where now the S i S_i S i are objects in C \mathcal{C} C and the d N , i , s N , i d_{N,i},s_{N,i} d N , i , s N , i are morphisms in C \mathcal{C} C satisfying (2.2).
Definition 2.2. (i) The topological N N N -simplex Δ t o p N \Delta_{\mathrm{top}}^N Δ top N is
Δ t o p N = { ( x 0 , … , x N ) ∈ R N + 1 : x i ≥ 0 , x 0 + ⋯ + x N = 1 } . (37) \Delta_{\mathrm{top}}^N=\left\{(x_0,\ldots,x_N)\in\mathbb{R}^{N+1}:x_i\geq0,\ x_0+\cdots+x_N=1\right\}. \tag*{(37)} Δ top N = { ( x 0 , … , x N ) ∈ R N + 1 : x i ≥ 0 , x 0 + ⋯ + x N = 1 } . (37) If f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] is order-preserving we define f t o p : Δ t o p M → Δ t o p N f_{\mathrm{top}}:\Delta_{\mathrm{top}}^M\to\Delta_{\mathrm{top}}^N f top : Δ top M → Δ top N by
f t o p ( x 0 , … , x M ) = ( y 0 , … , y N ) , y i = ∑ j ∈ { 0 , … , M } : f ( j ) = i x j . f_{\mathrm{top}}(x_0,\ldots,x_M)=(y_0,\ldots,y_N),\qquad y_i=\sum_{j\in\{0,\ldots,M\}:f(j)=i}x_j. f top ( x 0 , … , x M ) = ( y 0 , … , y N ) , y i = j ∈ { 0 , … , M } : f ( j ) = i ∑ x j . This defines a functor G : Δ o p → C G H a u s G:\Delta^{\mathrm{op}}\to\mathbf{CGHaus} G : Δ op → CGHaus mapping [ N ] ↦ Δ t o p N [N]\mapsto\Delta_{\mathrm{top}}^N [ N ] ↦ Δ top N , f ↦ f t o p f\mapsto f_{\mathrm{top}} f ↦ f top , where C G H a u s ⊂ T o p \mathbf{CGHaus}\subset\mathbf{Top} CGHaus ⊂ Top is the category of compactly generated Hausdorff topological spaces.
(ii) Let X X X be a compactly-generated, Hausdorff topological space. A triangulation of X X X is data ( f i N : Δ t o p N → X ) i ∈ I N (f_i^N:\Delta_{\mathrm{top}}^N\to X)_{i\in I^N} ( f i N : Δ top N → X ) i ∈ I N for N ≥ 0 N\geq0 N ≥ 0 , where I N I^N I N is an indexing set, such that f i N f_i^N f i N is a homeomorphism with a closed subset Im f i N ⊂ X \operatorname{Im}f_i^N\subset X Im f i N ⊂ X , and for each j = 0 , … , N j=0,\ldots,N j = 0 , … , N the restriction of f i N f_i^N f i N to the face x j = 0 x_j=0 x j = 0 of Δ t o p N \Delta_{\mathrm{top}}^N Δ top N is f i ′ N − 1 f_{i'}^{N-1} f i ′ N − 1 for some unique i ′ ∈ I N − 1 i'\in I^{N-1} i ′ ∈ I N − 1 , that is,
f i N ( x 0 , … , x j − 1 , 0 , x j + 1 , … , x N ) = f i ′ N − 1 ( x 0 , … , x j − 1 , x j + 1 , … , x N ) , f_i^N(x_0,\ldots,x_{j-1},0,x_{j+1},\ldots,x_N)=f_{i'}^{N-1}(x_0,\ldots,x_{j-1},x_{j+1},\ldots,x_N), f i N ( x 0 , … , x j − 1 , 0 , x j + 1 , … , x N ) = f i ′ N − 1 ( x 0 , … , x j − 1 , x j + 1 , … , x N ) , and X = ∐ N ≥ 0 ∐ i ∈ I N f i N ( ( Δ t o p N ) ∘ ) X=\coprod_{N\geq0}\coprod_{i\in I^N}f_i^N((\Delta_{\mathrm{top}}^N)^\circ) X = ∐ N ≥ 0 ∐ i ∈ I N f i N (( Δ top N ) ∘ ) is a stratification of X X X into disjoint locally closed sets. For example, think of a topological surface divided into triangles.
(iii) Let S S S be a simplicial set. We define a compactly-generated, Hausdorff topological space X S X_S X S with a triangulation, the topological realization of S S S , by
X S = ( ∐ N ≥ 0 S ( [ N ] ) × Δ t o p N ) / ∼ , X_S=\left(\coprod_{N\geq0}S([N])\times\Delta_{\mathrm{top}}^N\right)/\sim, X S = ( N ≥ 0 ∐ S ([ N ]) × Δ top N ) / ∼ , where the equivalence relation ∼ \sim ∼ is generated by
( Σ , f t o p ( x 0 , … , x M ) ) ≃ ( S ( f ) Σ , ( x 0 , … , x M ) ) , (\Sigma,f_{\mathrm{top}}(x_0,\ldots,x_M))\simeq(S(f)\Sigma,(x_0,\ldots,x_M)), ( Σ , f top ( x 0 , … , x M )) ≃ ( S ( f ) Σ , ( x 0 , … , x M )) , for all f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] in Δ \Delta Δ , Σ ∈ S ( [ N ] ) \Sigma\in S([N]) Σ ∈ S ([ N ]) and ( x 0 , … , x M ) ∈ Δ t o p M (x_0,\ldots,x_M)\in\Delta_{\mathrm{top}}^M ( x 0 , … , x M ) ∈ Δ top M . The maps f i N : Δ t o p N → X f_i^N:\Delta_{\mathrm{top}}^N\to X f i N : Δ top N → X in the triangulation of X X X come from the inclusions Δ t o p N → { Σ } × Δ t o p N \Delta_{\mathrm{top}}^N\to\{\Sigma\}\times\Delta_{\mathrm{top}}^N Δ top N → { Σ } × Δ top N for nondegenerate N N N -simplices Σ ∈ S N \Sigma\in S_N Σ ∈ S N . In this way we can define a topological realization functor T R : S S e t → C G H a u s ⊂ T o p \mathbf{TR}:\mathbf{SSet}\to\mathbf{CGHaus}\subset\mathbf{Top} TR : SSet → CGHaus ⊂ Top . It is a right Kan extension
Δ o p → Yoneda embedding S S e t G ↓ ↓ T R C G H a u s = C G H a u s \begin{CD}
\Delta^{\mathrm{op}} @>{\text{Yoneda embedding}}>> \mathbf{SSet}\\
@V{G}VV @VV{\mathbf{TR}}V\\
\mathbf{CGHaus} @= \mathbf{CGHaus}
\end{CD} Δ op G ↓ ⏐ CGHaus Yoneda embedding SSet ↓ ⏐ TR CGHaus We use C G H a u s \mathbf{CGHaus} CGHaus rather than T o p \mathbf{Top} Top here as it has better categorical properties: both S S e t \mathbf{SSet} SSet and C G H a u s \mathbf{CGHaus} CGHaus are Cartesian closed, and T R \mathbf{TR} TR is product-preserving as a functor to C G H a u s \mathbf{CGHaus} CGHaus , but not as a functor to T o p \mathbf{Top} Top .
Example 2.3. Let Δ N \Delta^N Δ N be the simplicial set represented by the object [ N ] [N] [ N ] , so that by definition Δ N : Δ o p → S e t \Delta^N : \Delta^{\mathrm{op}} \to\mathbf{Set} Δ N : Δ op → Set has Δ N ( [ M ] ) = Hom Δ ( [ M ] , [ N ] ) \Delta^N([M]) = \operatorname{Hom}_{\Delta}([M],[N]) Δ N ([ M ]) = Hom Δ ([ M ] , [ N ]) . By the Yoneda Lemma, if S S S is a simplicial set the set of maps Δ N → S \Delta^N \to S Δ N → S (that is, morphisms of simplicial sets) is in natural bijection with S N S_N S N . The topological realization TR ( Δ N ) \operatorname{TR}(\Delta^N) TR ( Δ N ) is naturally homeomorphic to Δ t o p N \Delta^N_{\mathrm{top}} Δ top N in (37) .
Example 2.4. The next example will be important in §2.2. Let the simplicial set Δ N \Delta^N Δ N be as in Example 2.3, and let 0 ≤ i ≤ N 0 \le i \le N 0 ≤ i ≤ N . The i i i th N N N -horn is the simplicial subset Λ i N ⊂ Δ N \Lambda^N_i \subset\Delta^N Λ i N ⊂ Δ N such that Λ i N ( [ M ] ) \Lambda^N_i([M]) Λ i N ([ M ]) is the subset of order-preserving functions f : [ M ] → [ N ] f : [M] \to[N] f : [ M ] → [ N ] with [ N ] ∖ { i } ⊈ f ( [ M ] ) [N] \setminus\{i\} \nsubseteq f([M]) [ N ] ∖ { i } ⊈ f ([ M ]) . The topological realization TR ( Λ i N ) \operatorname{TR}(\Lambda^N_i) TR ( Λ i N ) is naturally homeomorphic to the subset of Δ t o p N \Delta^N_{\mathrm{top}} Δ top N in (37) obtained by deleting the interior of Δ t o p N \Delta^N_{\mathrm{top}} Δ top N , and the interior of the ( N − 1 ) (N-1) ( N − 1 ) -dimensional boundary face Δ i , t o p N − 1 \Delta^{N-1}_{i,\mathrm{top}} Δ i , top N − 1 opposite the i i i th vertex. So for example TR ( Λ 1 2 ) \operatorname{TR}(\Lambda^2_1) TR ( Λ 1 2 ) consists of two sides of the triangle Δ t o p 2 \Delta^2_{\mathrm{top}} Δ top 2 .
Example 2.5. There is a good notion of product of simplicial sets. Considering simplicial sets as functors S , T : Δ o p → S e t S,T : \Delta^{\mathrm{op}} \to\mathbf{Set} S , T : Δ op → Set , the product S × T S \times T S × T just takes products in S e t \mathbf{Set} Set , so that ( S × T ) ( [ N ] ) = S ( [ N ] ) × T ( [ N ] ) (S \times T)([N]) = S([N]) \times T([N]) ( S × T ) ([ N ]) = S ([ N ]) × T ([ N ]) . This satisfies the universal property of products in a category.
The picture of topological realizations of products in S S e t \mathbf{SSet} SSet is more interesting. The topological realization TR ( S × T ) \operatorname{TR}(S \times T) TR ( S × T ) is canonically homeomorphic to the topological product TR ( S ) × TR ( T ) \operatorname{TR}(S) \times\operatorname{TR}(T) TR ( S ) × TR ( T ) , but the triangulation of TR ( S × T ) \operatorname{TR}(S \times T) TR ( S × T ) is constructed from those of TR ( S ) \operatorname{TR}(S) TR ( S ) and TR ( T ) \operatorname{TR}(T) TR ( T ) by dividing products Δ t o p M × Δ t o p N \Delta^M_{\mathrm{top}} \times\Delta^N_{\mathrm{top}} Δ top M × Δ top N canonically into ( M + N M ) \binom{M+N}{M} ( M M + N ) copies of Δ t o p M + N \Delta^{M+N}_{\mathrm{top}} Δ top M + N .
Example 2.6. There is a functor Sing : T o p → S S e t \operatorname{Sing} : \mathbf{Top} \to\mathbf{SSet} Sing : Top → SSet which maps a space X X X to the functor F : Δ o p → S e t F : \Delta^{\mathrm{op}} \to\mathbf{Set} F : Δ op → Set with F ( [ N ] ) = Map C 0 ( Δ t o p N , X ) F([N]) = \operatorname{Map}_{C^0}(\Delta^N_{\mathrm{top}},X) F ([ N ]) = Map C 0 ( Δ top N , X ) for objects [ N ] [N] [ N ] and F ( f ) = − ∘ f t o p : Map C 0 ( Δ t o p N , X ) → Map C 0 ( Δ t o p M , X ) F(f) = - \circ f_{\mathrm{top}} : \operatorname{Map}_{C^0}(\Delta^N_{\mathrm{top}},X) \to\operatorname{Map}_{C^0}(\Delta^M_{\mathrm{top}},X) F ( f ) = − ∘ f top : Map C 0 ( Δ top N , X ) → Map C 0 ( Δ top M , X ) for morphisms f : [ M ] → [ N ] f : [M] \to[N] f : [ M ] → [ N ] , so that f t o p : Δ t o p M → Δ t o p N f_{\mathrm{top}} : \Delta^M_{\mathrm{top}} \to\Delta^N_{\mathrm{top}} f top : Δ top M → Δ top N . It is right adjoint to TR : S S e t → T o p \operatorname{TR} : \mathbf{SSet} \to\mathbf{Top} TR : SSet → Top .
If X X X is a topological space, there is a natural map η X : TR ∘ Sing ( X ) → X \eta_X : \operatorname{TR} \circ\operatorname{Sing}(X) \to X η X : TR ∘ Sing ( X ) → X coming from the adjunction. It turns out that η X \eta_X η X is a weak homotopy equivalence (i.e. it induces isomorphisms π n ( TR ∘ Sing ( X ) , x ′ ) → π n ( X , η X ( x ′ ) ) \pi_n(\operatorname{TR} \circ\operatorname{Sing}(X),x') \to\pi_n(X,\eta_X(x')) π n ( TR ∘ Sing ( X ) , x ′ ) → π n ( X , η X ( x ′ )) on homotopy groups for all n ≥ 0 n \ge0 n ≥ 0 and x ′ ∈ TR ∘ Sing ( X ) x' \in\operatorname{TR} \circ\operatorname{Sing}(X) x ′ ∈ TR ∘ Sing ( X ) ). Thus by the Whitehead Theorem, if X X X is a CW complex then η X \eta_X η X is a homotopy equivalence.
Theories of ∞ \infty ∞ -categories have their roots in homotopy theory of topological spaces. In an ∞ \infty ∞ -category (however one defines them, there are several approaches), one wants notions of homotopies and homotopy equivalences between morphisms, higher homotopies, and so on. Simplicial sets provide a model for homotopy theory of topological spaces which is very convenient for developing a theory of ∞ \infty ∞ -categories, as in Joyal [46] and Lurie [60, 61, 63].
The next definition follows Lurie [60, §1.2.1].
Definition 2.8. Define a functor Σ : Δ → Δ \Sigma: \Delta\to\Delta Σ : Δ → Δ by Σ ( [ N ] ) = [ N ] \Sigma([N]) = [N] Σ ([ N ]) = [ N ] on objects, and if f : [ N ] → [ M ] f : [N] \to[M] f : [ N ] → [ M ] is a morphism in Δ \Delta Δ then Σ ( f ) : [ N ] → [ M ] \Sigma(f) : [N] \to[M] Σ ( f ) : [ N ] → [ M ] is Σ ( f ) : i ↦ M − f ( N − i ) \Sigma(f) : i \mapsto M - f(N-i) Σ ( f ) : i ↦ M − f ( N − i ) . Then Σ ∘ Σ = id Δ \Sigma\circ\Sigma= \operatorname{id}_{\Delta} Σ ∘ Σ = id Δ , a strict equality of functors.
If S S S is a simplicial set, define the opposite simplicial set S o p S^{\mathrm{op}} S op by S o p = S ∘ Σ o p S^{\mathrm{op}} = S \circ\Sigma^{\mathrm{op}} S op = S ∘ Σ op in functors Δ o p → S e t \Delta^{\mathrm{op}} \to\mathbf{Set} Δ op → Set , that is, S o p S^{\mathrm{op}} S op is the composition Δ o p → Σ o p Δ o p → S S e t \Delta^{\mathrm{op}} \xrightarrow{\Sigma^{\mathrm{op}}} \Delta^{\mathrm{op}} \xrightarrow{S} \mathbf{Set} Δ op Σ op Δ op S Set . Clearly ( S o p ) o p = S (S^{\mathrm{op}})^{\mathrm{op}} = S ( S op ) op = S , a strict equality of simplicial sets. This extends to a functor ( − ) o p : S S e t → S S e t (-)^{\mathrm{op}} : \mathbf{SSet} \to\mathbf{SSet} ( − ) op : SSet → SSet by composing morphisms in S S e t \mathbf{SSet} SSet , which are natural transformations of functors Δ o p → S e t \Delta^{\mathrm{op}} \to\mathbf{Set} Δ op → Set , with Σ o p \Sigma^{\mathrm{op}} Σ op .
If we think of a simplicial set S S S as being built of N N N -simplices Δ N \Delta^N Δ N with vertices numbered 0 , 1 , … , N 0,1,\ldots,N 0 , 1 , … , N , then S o p S^{\mathrm{op}} S op is built of the same N N N -simplices, but with the numbering of the vertices changed, so that vertex i i i in S S S gets numbered N − i N-i N − i in S o p S^{\mathrm{op}} S op . In Definition 2.1, S o p S^{\mathrm{op}} S op has the same sets S N o p = S N S^{\mathrm{op}}_N = S_N S N op = S N , and face and degeneracy maps d N , i o p = d N , N − i d^{\mathrm{op}}_{N,i} = d_{N,N-i} d N , i op = d N , N − i , s N , i o p = s N , N − i s^{\mathrm{op}}_{N,i} = s_{N,N-i} s N , i op = s N , N − i .
Kan complexes and ∞ \infty ∞ -categories
Groth [39] and Joyal [45] give surveys of ∞ \infty ∞ -categories, and Land [55] is an introductory book. They were introduced by Boardman and Vogt [7] under the name weak Kan complexes . Joyal [46] and Lurie’s monumental works [60,61,63] develop the theory in detail.
Definition 2.9. Let the simplicial sets Δ N \Delta^N Δ N , Λ i N \Lambda^N_i Λ i N with inclusion morphism inc : Λ i N ↪ Δ N \operatorname{inc} : \Lambda^N_i \hookrightarrow\Delta^N inc : Λ i N ↪ Δ N be as in Examples 2.3–2.4. A simplicial set S S S is called a Kan complex if every map of simplicial sets f : Λ i N → S f : \Lambda^N_i \to S f : Λ i N → S can be extended to an N N N -simplex f ~ : Δ N → S \widetilde{f} : \Delta^N \to S f : Δ N → S for all 0 ≤ i ≤ N 0 \le i \le N 0 ≤ i ≤ N . That is, there should exist (not necessarily unique) f ~ \widetilde{f} f in the commutative diagram in S S e t \mathbf{SSet} SSet :
Λ i N → f S inc ↓ f ~ ↑ Δ N → f ~ S (38) \begin{CD}
\Lambda^N_i @>{f}>> S \\
@V{\operatorname{inc}}VV @A{\widetilde{f}}AA \\
\Delta^N @>>{\widetilde{f}}> S \tag*{(38)}
\end{CD} Λ i N inc ↓ ⏐ Δ N f f S f ⏐ ↑ S (38) A simplicial set S S S is called an ∞ \infty ∞ -category (or quasicategory, or inner Kan complex, or weak Kan complex) if every map of simplicial sets f : Λ i N → S f : \Lambda^N_i \to S f : Λ i N → S can be extended to an N N N -simplex f ~ : Δ N → S \widetilde{f} : \Delta^N \to S f : Δ N → S for all 0 < i < N 0 < i < N 0 < i < N .
We will write C a t ∞ 1 ⊂ S S e t \mathrm{Cat}^{1}_{\infty} \subset\mathbf{SSet} Cat ∞ 1 ⊂ SSet for the full subcategory of ∞ \infty ∞ -categories in S S e t \mathbf{SSet} SSet . This is an ordinary category. Note that it is different to the ∞ \infty ∞ -category of ∞ \infty ∞ -categories C a t ∞ \mathrm{Cat}_{\infty} Cat ∞ discussed in §2.4. If C \mathcal{C} C , D \mathcal{D} D are ∞ \infty ∞ -categories, a map of simplicial sets F : C → D F : \mathcal{C} \to\mathcal{D} F : C → D is called an ∞ \infty ∞ -functor, and these are the morphisms in C a t ∞ 1 \mathrm{Cat}^{1}_{\infty} Cat ∞ 1 .
Example 2.11. Let C \mathcal{C} C be a small category. We will define a simplicial set N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) called the nerve of C \mathcal{C} C . We define the N N N -simplices N ( C ) N \mathcal{N}(\mathcal{C})_N N ( C ) N for N ≥ 0 N \ge0 N ≥ 0 to be the set of N N N -tuples of composable morphisms C 0 → f 1 C 1 → f 2 ⋯ → f N C N C_0 \xrightarrow{f_1} C_1 \xrightarrow{f_2} \cdots\xrightarrow{f_N} C_N C 0 f 1 C 1 f 2 ⋯ f N C N in C \mathcal{C} C , that is, the set of ( 2 N + 1 ) (2N+1) ( 2 N + 1 ) -tuples ( C 0 , … , C N , f 1 , … , f N ) (C_0,\ldots,C_N,f_1,\ldots,f_N) ( C 0 , … , C N , f 1 , … , f N ) where f i : C i − 1 → C i f_i:C_{i-1}\to C_i f i : C i − 1 → C i is a morphism in C \mathcal{C} C for 1 ≤ i ≤ N 1\le i\le N 1 ≤ i ≤ N .
The face map d N , i : N ( C ) N → N ( C ) N − 1 d_{N,i}:\mathcal{N}(\mathcal{C})_N\to\mathcal{N}(\mathcal{C})_{N-1} d N , i : N ( C ) N → N ( C ) N − 1 for 0 < i < N 0<i<N 0 < i < N composes f i − 1 f_{i-1} f i − 1 and f i f_i f i , so that d N , i ( C 0 , … , f N ) = ( C 0 , … , C i − 1 , C i + 1 , … , C N , f 1 , … , f i ∘ f i − 1 , … , f N ) d_{N,i}(C_0,\ldots,f_N)=(C_0,\ldots,C_{i-1},C_{i+1},\ldots,C_N,f_1,\ldots,f_i\circ f_{i-1},\ldots,f_N) d N , i ( C 0 , … , f N ) = ( C 0 , … , C i − 1 , C i + 1 , … , C N , f 1 , … , f i ∘ f i − 1 , … , f N ) , and d N , 0 d_{N,0} d N , 0 omits C 0 , f 1 C_0,f_1 C 0 , f 1 , and d N , N d_{N,N} d N , N omits C N , f N C_N,f_N C N , f N .
The degeneracy map s N , i : N ( C ) N → N ( C ) N + 1 s_{N,i}:\mathcal{N}(\mathcal{C})_N\to\mathcal{N}(\mathcal{C})_{N+1} s N , i : N ( C ) N → N ( C ) N + 1 for 0 ≤ i ≤ N 0\le i\le N 0 ≤ i ≤ N acts on ( C 0 , … , f N ) (C_0,\ldots,f_N) ( C 0 , … , f N ) by inserting id C i : C i → C i \operatorname{id}_{C_i}:C_i\to C_i id C i : C i → C i between f i f_i f i and f i + 1 f_{i+1} f i + 1 .
Then N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) is a simplicial set, and an ∞ \infty ∞ -category. Also N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) is a Kan complex if and only if C \mathcal{C} C is a groupoid (i.e. all morphisms in C \mathcal{C} C are isomorphisms). We can recover the category C \mathcal{C} C from N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) (objects are 0-simplices, morphisms are 1-simplices, 2-simplices give composition, identities are given by s 0 , 0 s_{0,0} s 0 , 0 ).
In fact N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) has the extra property that the lifts f ~ \widetilde{f} f of f f f in (38) exist and are unique for all 0 < i < N 0<i<N 0 < i < N and f f f . A simplicial set S S S is isomorphic to a nerve N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) if and only if it is an ∞ \infty ∞ -category for which lifts f ~ \widetilde{f} f have this uniqueness property. Thus, we can see ∞ \infty ∞ -categories (or Kan complexes) as generalizations of small categories (or groupoids) in which the lifts f ~ \widetilde{f} f in (38) need not be unique.
The nerve defines a functor N : C a t 1 → C a t ∞ 1 \mathcal{N}:\mathrm{Cat}_1\to\mathrm{Cat}_{\infty}^{1} N : Cat 1 → Cat ∞ 1 , where C a t 1 \mathrm{Cat}_1 Cat 1 is the ordinary category of (small) ordinary categories, and C a t ∞ 1 \mathrm{Cat}_{\infty}^{1} Cat ∞ 1 the ordinary category of ∞ \infty ∞ -categories from Definition 2.9 .
Definition 2.12. Let C \mathcal{C} C be an ∞ \infty ∞ -category. An object X X X of C \mathcal{C} C is a 0-simplex in C \mathcal{C} C , or equivalently, a map of simplicial sets X : Δ 0 → C X:\Delta^0\to\mathcal{C} X : Δ 0 → C . A morphism , or 1-morphism, f : X → Y f:X\to Y f : X → Y of C \mathcal{C} C is a 1-simplex f f f in C \mathcal{C} C with d 1 , 0 ( f ) = X d_{1,0}(f)=X d 1 , 0 ( f ) = X and d 1 , 1 ( f ) = Y d_{1,1}(f)=Y d 1 , 1 ( f ) = Y , where d N , i d_{N,i} d N , i are the degeneracy maps in C \mathcal{C} C as in (36) , or equivalently, a map of simplicial sets f : Δ 1 → C f:\Delta^1\to\mathcal{C} f : Δ 1 → C with f ∣ Δ { 0 } = X f|_{\Delta^{\{0\}}}=X f ∣ Δ { 0 } = X and f ∣ Δ { 1 } = Y f|_{\Delta^{\{1\}}}=Y f ∣ Δ { 1 } = Y .
For an object X X X , the identity morphism is id X = s 0 , 0 ( X ) \operatorname{id}_X=s_{0,0}(X) id X = s 0 , 0 ( X ) , where s N , i s_{N,i} s N , i are the face maps in C \mathcal{C} C as in (36) .
Let f : X → Y f:X\to Y f : X → Y and g : Y → Z g:Y\to Z g : Y → Z be morphisms in C \mathcal{C} C . A morphism h : X → Z h:X\to Z h : X → Z is a composition g ∘ f g\circ f g ∘ f if there exists a 2-simplex η \eta η in C \mathcal{C} C of the form
X → f Y → g Z , X ↘ h = g ∘ f ↗ (39) \begin{aligned}
&X \xrightarrow{f} Y \xrightarrow{g} Z,\\
&\phantom{X}\searrow^{h=g\circ f}\quad\nearrow \tag*{(39)}
\end{aligned} X f Y g Z , X ↘ h = g ∘ f ↗ (39) The horn-filling condition for Λ 1 2 \Lambda^2_1 Λ 1 2 guarantees such h h h exist, but they are not unique . That is, in an ∞ \infty ∞ -category composition of morphisms is not uniquely defined, but there can be many candidates h h h for g ∘ f g\circ f g ∘ f . Because of this non-uniqueness, composition is not associative (though it is ‘up to homotopy’).
For fixed objects X , Y X,Y X , Y , define a binary relation ≃ \simeq ≃ on morphisms $f,g:X\to Y
by f ≈ g f \approx g f ≈ g if there exists a 2-simplex η \eta η in C \mathcal{C} C of the form
X → f Y i d X ↓ ↑ X → g Y \begin{CD}
X @>{f}>> Y \\
@V{\mathrm{id}_X}VV @AA{}A \\
X @>{g}>> Y
\end{CD} X id X ↓ ⏐ X f g Y ⏐ ↑ Y
The ∞ \infty ∞ -category axioms imply ≈ \approx ≈ is an equivalence relation (exercise), and also that f ≈ g f \approx g f ≈ g if and only if there exists a 2-simplex η \eta η in C \mathcal{C} C of the form
X → f Y g ↓ ↑ i d Y Y → Y \begin{CD}
X @>{f}>> Y \\
@V{g}VV @AA{\mathrm{id}_Y}A \\
Y @>>> Y
\end{CD} X g ↓ ⏐ Y f Y ⏐ ↑ id Y Y
As in [39, §1.1], [55, §1.2], the homotopy category Ho ( C ) \operatorname{Ho}(\mathcal{C}) Ho ( C ) is the ordinary category whose objects are objects of C \mathcal{C} C , and whose morphisms [ f ] : X → Y [f] : X \to Y [ f ] : X → Y are ≈ \approx ≈ -equivalence classes [ f ] [f] [ f ] of morphisms f : X → Y f : X \to Y f : X → Y in C \mathcal{C} C . Identity morphisms are i d X = [ i d X ] : X → X \mathrm{id}_X = [\mathrm{id}_X] : X \to X id X = [ id X ] : X → X , and for morphisms [ f ] : X → Y [f] : X \to Y [ f ] : X → Y , [ g ] : Y → Z [g] : Y \to Z [ g ] : Y → Z , the composition is [ g ] ∘ [ f ] = [ h ] [g] \circ[f] = [h] [ g ] ∘ [ f ] = [ h ] for h h h as in (39) . This is well defined.
A morphism f : X → Y f : X \to Y f : X → Y in C \mathcal{C} C is called an equivalence if [ f ] : X → Y [f] : X \to Y [ f ] : X → Y is an isomorphism in Ho ( C ) \operatorname{Ho}(\mathcal{C}) Ho ( C ) .
If C \mathcal{C} C , D \mathcal{D} D are ∞ \infty ∞ -categories then an ∞ \infty ∞ -functor F : C → D F : \mathcal{C} \to\mathcal{D} F : C → D induces an ordinary functor Ho ( F ) : Ho ( C ) → Ho ( D ) \operatorname{Ho}(F) : \operatorname{Ho}(\mathcal{C}) \to\operatorname{Ho}(\mathcal{D}) Ho ( F ) : Ho ( C ) → Ho ( D ) . As in Cisinski [17, Def. 3.9.3] and Land [55, Def. 2.3.3], an ∞ \infty ∞ -functor F : C → D F : \mathcal{C} \to\mathcal{D} F : C → D is called essentially surjective if Ho ( F ) : Ho ( C ) → Ho ( D ) \operatorname{Ho}(F) : \operatorname{Ho}(\mathcal{C}) \to\operatorname{Ho}(\mathcal{D}) Ho ( F ) : Ho ( C ) → Ho ( D ) is essentially surjective, that is, every object in Ho ( D ) \operatorname{Ho}(\mathcal{D}) Ho ( D ) is isomorphic to an object in the image of Ho ( F ) \operatorname{Ho}(F) Ho ( F ) .
Let C , D \mathcal{C}, \mathcal{D} C , D be ∞ \infty ∞ -categories, and F , G : C → D F, G : \mathcal{C} \to\mathcal{D} F , G : C → D be ∞ \infty ∞ -functors. An ∞ \infty ∞ -natural transformation η : F ⇒ G \eta: F \Rightarrow G η : F ⇒ G is a morphism of simplicial sets η : C × Δ 1 → D \eta: \mathcal{C} \times\Delta^1 \to\mathcal{D} η : C × Δ 1 → D with η ∣ C × { 0 } = F \eta|_{\mathcal{C} \times\{0\}} = F η ∣ C × { 0 } = F and η ∣ C × { 1 } = G \eta|_{\mathcal{C} \times\{1\}} = G η ∣ C × { 1 } = G . Then η \eta η induces a natural transformation Ho ( η ) : Ho ( F ) ⇒ Ho ( G ) \operatorname{Ho}(\eta) : \operatorname{Ho}(F) \Rightarrow\operatorname{Ho}(G) Ho ( η ) : Ho ( F ) ⇒ Ho ( G ) of ordinary functors Ho ( C ) → Ho ( D ) \operatorname{Ho}(\mathcal{C}) \to\operatorname{Ho}(\mathcal{D}) Ho ( C ) → Ho ( D ) . We call η \eta η an ∞ \infty ∞ -natural isomorphism if for each object x x x in C \mathcal{C} C , η ( { x } × Δ 1 ) \eta(\{x\} \times\Delta^1) η ({ x } × Δ 1 ) is an equivalence in D \mathcal{D} D . Then Ho ( η ) : Ho ( F ) ⇒ Ho ( G ) \operatorname{Ho}(\eta) : \operatorname{Ho}(F) \Rightarrow\operatorname{Ho}(G) Ho ( η ) : Ho ( F ) ⇒ Ho ( G ) is a natural isomorphism.
If C \mathcal{C} C is an ordinary category, so that its nerve N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) is an ∞ \infty ∞ -category, there is a canonical strict isomorphism of categories Ho ( N ( C ) ) ≅ C \operatorname{Ho}(\mathcal{N}(\mathcal{C})) \cong\mathcal{C} Ho ( N ( C )) ≅ C .
For any ∞ \infty ∞ -category C \mathcal{C} C , there is a unique ∞ \infty ∞ -functor
Π Ho : C ⟶ N ( Ho ( C ) ) (40) \Pi_{\operatorname{Ho}} : \mathcal{C} \longrightarrow\mathcal{N}(\operatorname{Ho}(\mathcal{C})) \tag*{(40)} Π Ho : C ⟶ N ( Ho ( C )) (40)
which maps X ↦ X X \mapsto X X ↦ X on objects and f ↦ [ f ] f \mapsto[f] f ↦ [ f ] on morphisms (1-simplices).
The next definition follows Lurie [60, §1.2.1].
Definition 2.13. Let C \mathcal{C} C be an ∞ \infty ∞ -category. The opposite ∞ \infty ∞ -category C o p \mathcal{C}^{\mathrm{op}} C op is the opposite simplicial set of C \mathcal{C} C in the sense of Definition 2.8. Then C o p \mathcal{C}^{\mathrm{op}} C op is an ∞ \infty ∞ -category, since taking opposites turns the horn-filling condition for Λ i N \Lambda^N_i Λ i N into the horn-filling condition for Λ N − i N \Lambda^N_{N-i} Λ N − i N . We have ( C o p ) o p = C (\mathcal{C}^{\mathrm{op}})^{\mathrm{op}} = \mathcal{C} ( C op ) op = C .
If f : X → Y f:X\to Y f : X → Y is a morphism in C \mathcal{C} C then f : Y → X f:Y\to X f : Y → X is a morphism in C o p \mathcal{C}^{\mathrm{op}} C op , as f : Δ 1 → C f:\Delta^1\to\mathcal{C} f : Δ 1 → C corresponds to f o p : ( Δ 1 ) o p → C o p f^{\mathrm{op}}:(\Delta^1)^{\mathrm{op}}\to\mathcal{C}^{\mathrm{op}} f op : ( Δ 1 ) op → C op , and changing Δ 1 ↦ ( Δ 1 ) o p \Delta^1\mapsto(\Delta^1)^{\mathrm{op}} Δ 1 ↦ ( Δ 1 ) op exchanges vertices 0 , 1 0,1 0 , 1 of Δ 1 \Delta^1 Δ 1 . Thus, opposite ∞ \infty ∞ -categories are the analogues of opposite categories in ordinary category theory.
If F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D is an ∞ \infty ∞ -functor then F o p : C o p → D o p F^{\mathrm{op}}:\mathcal{C}^{\mathrm{op}}\to\mathcal{D}^{\mathrm{op}} F op : C op → D op is an ∞ \infty ∞ -functor.
Definition 2.14. An ∞ \infty ∞ -category C \mathcal{C} C is called an ∞ \infty ∞ -groupoid if all morphisms in C \mathcal{C} C are equivalences, [55, Def. 1.2.24]. As in Land [55, Lem. 1.2.29 & Cor. 2.1.12], it is a nontrivial theorem that ∞ \infty ∞ -groupoids are the same as Kan complexes.
If C \mathcal{C} C is an ∞ \infty ∞ -category, define the maximal Kan complex C K a n \mathcal{C}^{\mathrm{Kan}} C Kan in C \mathcal{C} C to be the simplicial subset of C \mathcal{C} C such that an N N N -simplex S : Δ N → C S:\Delta^N\to\mathcal{C} S : Δ N → C in C \mathcal{C} C lies in C K a n \mathcal{C}^{\mathrm{Kan}} C Kan if and only if every 1 1 1 -simplex in S S S is an equivalence. Then C K a n \mathcal{C}^{\mathrm{Kan}} C Kan is an ∞ \infty ∞ -groupoid, and thus a Kan complex, and it contains every simplicial subset of C \mathcal{C} C which is a Kan complex, so it is the maximal Kan complex in C \mathcal{C} C .
If F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D is an ∞ \infty ∞ -functor of ∞ \infty ∞ -categories then F ( C K a n ) ⊆ D K a n ⊆ D F(\mathcal{C}^{\mathrm{Kan}})\subseteq\mathcal{D}^{\mathrm{Kan}}\subseteq\mathcal{D} F ( C Kan ) ⊆ D Kan ⊆ D , and we write F K a n = F ∣ C K a n : C K a n → D K a n F^{\mathrm{Kan}}=F|_{\mathcal{C}^{\mathrm{Kan}}}:\mathcal{C}^{\mathrm{Kan}}\to\mathcal{D}^{\mathrm{Kan}} F Kan = F ∣ C Kan : C Kan → D Kan .
Mapping spaces in ∞ \infty ∞ -categories
As one model for ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -categories C \mathcal{C} C is categories enhanced in topological spaces, one should expect mapping spaces Hom ( X , Y ) \operatorname{Hom}(X,Y) Hom ( X , Y ) in C \mathcal{C} C to have the structure not just of sets, but of topological spaces up to homotopy equivalence. In the world of simplicial sets, topological spaces correspond to Kan complexes. So in ∞ \infty ∞ -categories (quasicategories), we expect mapping spaces to be Kan complexes, natural up to homotopy equivalence. There are several different, but homotopy-equivalent, definitions of the Kan complex Hom ( X , Y ) \operatorname{Hom}(X,Y) Hom ( X , Y ) . The next definition and proposition follow Lurie [60, §1.2.2, Prop. 1.2.2.3 & Cor. 4.2.1.8] and [63, §4.6].
Definition 2.15. Let C \mathcal{C} C be an ∞ \infty ∞ -category, and let X , Y X,Y X , Y be objects (0 0 0 -simplices) in C \mathcal{C} C . Define a simplicial set Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) , the space of left morphisms from X X X to Y Y Y in C \mathcal{C} C , by taking Hom S S e t ( Δ N , Hom C L ( X , Y ) ) \operatorname{Hom}_{\mathrm{SSet}}(\Delta^N,\operatorname{Hom}^{L}_{\mathcal{C}}(X,Y)) Hom SSet ( Δ N , Hom C L ( X , Y )) to be the set of all morphisms Z : Δ N + 1 → C Z:\Delta^{N+1}\to\mathcal{C} Z : Δ N + 1 → C such that Z ∣ Δ { 0 } Z|_{\Delta^{\{0\}}} Z ∣ Δ { 0 } is the 0 0 0 -simplex X X X , and Z ∣ Δ { 1 , 2 , … , N + 1 } Z|_{\Delta^{\{1,2,\ldots,N+1\}}} Z ∣ Δ { 1 , 2 , … , N + 1 } is the constant N N N -simplex at Y Y Y . The face and degeneracy maps in Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) are induced from those in C \mathcal{C} C by extending an order-preserving function f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] to an order-preserving function f ~ : [ N + 1 ] → [ M + 1 ] \widetilde{f}:[N+1]\to[M+1] f : [ N + 1 ] → [ M + 1 ] with f ~ ( 0 ) = 0 \widetilde{f}(0)=0 f ( 0 ) = 0 and f ~ ( i ) = f ( i − 1 ) + 1 \widetilde{f}(i)=f(i-1)+1 f ( i ) = f ( i − 1 ) + 1 for i = 1 , … , N + 1 i=1,\ldots,N+1 i = 1 , … , N + 1 .
Similarly, define a simplicial set Hom C R ( X , Y ) \operatorname{Hom}^{R}_{\mathcal{C}}(X,Y) Hom C R ( X , Y ) , the space of right morphisms from X X X to Y Y Y in C \mathcal{C} C , by taking Hom S S e t ( Δ N , Hom C R ( X , Y ) ) \operatorname{Hom}_{\mathrm{SSet}}(\Delta^N,\operatorname{Hom}^{R}_{\mathcal{C}}(X,Y)) Hom SSet ( Δ N , Hom C R ( X , Y )) to be the set of all morphisms Z : Δ N + 1 → C Z:\Delta^{N+1}\to\mathcal{C} Z : Δ N + 1 → C such that Z ∣ Δ { 0 , 1 , … , N } Z|_{\Delta^{\{0,1,\ldots,N\}}} Z ∣ Δ { 0 , 1 , … , N } is a constant N N N -simplex at X X X , and Z ∣ Δ { N + 1 } Z|_{\Delta^{\{N+1\}}} Z ∣ Δ { N + 1 } is the 0 0 0 -simplex Y Y Y . The face and degeneracy maps in Hom C R ( X , Y ) \operatorname{Hom}^{R}_{\mathcal{C}}(X,Y) Hom C R ( X , Y ) are induced from those in C \mathcal{C} C by extending an order-preserving function f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] to an order-preserving function f ~ : [ N + 1 ] → [ M + 1 ] \widetilde{f}:[N+1]\to[M+1] f : [ N + 1 ] → [ M + 1 ] with f ~ ∣ [ N ] = f \widetilde{f}|_{[N]}=f f ∣ [ N ] = f and f ~ ( N + 1 ) = M + 1 \widetilde{f}(N+1)=M+1 f ( N + 1 ) = M + 1 .
Proposition 2.16. In Definition 2.15, Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) and Hom C R ( X , Y ) \operatorname{Hom}^{R}_{\mathcal{C}}(X,Y) Hom C R ( X , Y ) are Kan complexes, which are homotopy equivalent.
When we only care about mapping spaces up to homotopy equivalence, rather than the particular simplicial set, we may write Hom C ( X , Y ) \operatorname{Hom}_{\mathcal{C}}(X,Y) Hom C ( X , Y ) rather than Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) or Hom C R ( X , Y ) \operatorname{Hom}^{R}_{\mathcal{C}}(X,Y) Hom C R ( X , Y ) . The next definition follows Cisinski [17, Def. 3.9.3] and Land [55, Def. 2.3.2].
Definition 2.17. Let F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D be an ∞ \infty ∞ -functor of ∞ \infty ∞ -categories, and X , Y X,Y X , Y be objects in C \mathcal{C} C . Then composition of morphisms Δ N + 1 → C → F D \Delta^{N+1}\to\mathcal{C}\xrightarrow{F}\mathcal{D} Δ N + 1 → C F D induces a morphism of Kan complexes F X , Y : Hom C L ( X , Y ) → Hom D L ( F ( X ) , F ( Y ) ) F_{X,Y}:\operatorname{Hom}^{L}_{\mathcal{C}}(X,Y)\to\operatorname{Hom}^{L}_{\mathcal{D}}(F(X),F(Y)) F X , Y : Hom C L ( X , Y ) → Hom D L ( F ( X ) , F ( Y )) . We call F F F fully faithful if F X , Y F_{X,Y} F X , Y is a homotopy equivalence for all X , Y ∈ C X,Y\in\mathcal{C} X , Y ∈ C .
The ∞ \infty ∞ -category of ∞ \infty ∞ -categories
The ∞ \infty ∞ -category C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ is discussed in Lurie [60, §3] and Land [55, Def. 2.1.13].
Definition 2.18. In [60, Def. 3.0.0.1], Lurie defines the ∞ \infty ∞ -category of ∞ \infty ∞ -categories C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ . It is constructed as the simplicial nerve N ( C a t ∞ Δ ) N(\mathcal{C}\!at_{\infty}^{\Delta}) N ( C a t ∞ Δ ) of a simplicial category C a t ∞ Δ \mathcal{C}\!at_{\infty}^{\Delta} C a t ∞ Δ of ∞ \infty ∞ -categories. It has the properties:
(a) 0-simplices of C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ are (small) ∞ \infty ∞ -categories C , D \mathcal{C},\mathcal{D} C , D , and 1-simplices of C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ are ∞ \infty ∞ -functors F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D .
(b) The ordinary category of ∞ \infty ∞ -categories C a t ∞ 1 \mathcal{C}\!at^{1}_{\infty} C a t ∞ 1 from Definition 2.9 is a subcategory of C a t ∞ Δ \mathcal{C}\!at_{\infty}^{\Delta} C a t ∞ Δ (taking only 0-simplices in mapping spaces in C a t ∞ Δ \mathcal{C}\!at_{\infty}^{\Delta} C a t ∞ Δ ). The inclusion C a t ∞ 1 ↪ C a t ∞ Δ \mathcal{C}\!at^{1}_{\infty}\hookrightarrow\mathcal{C}\!at_{\infty}^{\Delta} C a t ∞ 1 ↪ C a t ∞ Δ induces an inclusion of ∞ \infty ∞ -subcategories N ( C a t ∞ 1 ) ↪ C a t ∞ N(\mathcal{C}\!at^{1}_{\infty})\hookrightarrow\mathcal{C}\!at_{\infty} N ( C a t ∞ 1 ) ↪ C a t ∞ .
(c) For ∞ \infty ∞ -categories C , D \mathcal{C},\mathcal{D} C , D , the right morphisms Hom C a t ∞ R ( C , D ) \operatorname{Hom}^{R}_{\mathcal{C}\!at_{\infty}}(\mathcal{C},\mathcal{D}) Hom C a t ∞ R ( C , D ) may be identified with the maximal Kan complex Fun ( C , D ) K a n \operatorname{Fun}(\mathcal{C},\mathcal{D})^{\mathrm{Kan}} Fun ( C , D ) Kan in the ∞ \infty ∞ -category Fun ( C , D ) \operatorname{Fun}(\mathcal{C},\mathcal{D}) Fun ( C , D ) of ∞ \infty ∞ -functors F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D , where N N N -simplices in Fun ( C , D ) \operatorname{Fun}(\mathcal{C},\mathcal{D}) Fun ( C , D ) correspond to maps of simplicial sets Δ N × C → D \Delta^{N}\times\mathcal{C}\to\mathcal{D} Δ N × C → D .
(d) As in [60, §3.3.3], C a t ∞ Δ \mathcal{C}\!at_{\infty}^{\Delta} C a t ∞ Δ may be identified with the subcategory of fibrant-cofibrant objects in a simplicial model category Set Δ + \operatorname{Set}^{+}_{\Delta} Set Δ + of ‘marked simplicial sets’ [60, §3.1]. This implies nice properties of C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ , for example, all small limits and colimits exist in C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ .
Definition 2.19. Let C , D \mathcal{C},\mathcal{D} C , D be ∞ \infty ∞ -categories. An ∞ \infty ∞ -functor F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D is called a Joyal equivalence [55, Def. 2.1.14] if the corresponding morphism [ F ] : C → D [F]:\mathcal{C}\to\mathcal{D} [ F ] : C → D in Ho ( C a t ∞ ) \operatorname{Ho}(\mathcal{C}\!at_{\infty}) Ho ( C a t ∞ ) is an isomorphism. This is the ∞ \infty ∞ -category analogue of equivalence of categories.
Here is Cisinski [17, Th. 3.9.7] and Land [55, Th. 2.3.20].
Theorem 2.20. Let F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D be an ∞ \infty ∞ -functor. Then F F F is a Joyal equivalence if and only if it is fully faithful and essentially surjective, in the sense of Definitions 2.12 and 2.17.
In §9.3 we will construct an ∞ \infty ∞ -functor A ∞ C a t R u n , ∞ → C a t ∞ \mathcal{A}_{\infty}\mathcal{C}\!at_{\mathrm{R}}^{\mathrm{un},\infty}\to\mathcal{C}\!at_{\infty} A ∞ C a t R un , ∞ → C a t ∞ from our ∞ \infty ∞ -category of A ∞ \mathcal{A}_{\infty} A ∞ -categories A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathcal{C}\!at_{\mathrm{R}}^{\mathrm{un},\infty} A ∞ C a t R un , ∞ . We do this using Lurie’s ‘straightening-unstraightening equivalence’ [60, §3.2], [55, §3], which for an ∞ \infty ∞ -category D \mathcal{D} D identifies ∞ \infty ∞ -functors D → C a t ∞ \mathcal{D}\to\mathcal{C}\!at_{\infty} D → C a t ∞ with ‘cocartesian fibrations’ p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D .
The next definition follows Lurie [60, §2.3–§2.4] and Land [55, §1.3 & §3.1], see also [63, §4.1, §4.2 & §5.1].
Definition 2.21. A map of simplicial sets p : S → T p:S\to T p : S → T is called an inner fibration if for all 0 < i < N 0<i<N 0 < i < N , any commutative square of maps of simplicial sets ‘→ \to → ’ as follows can be completed with a map ‘⇢ \dashrightarrow ⇢ ’ in a commutative diagram
Λ i N → S ↓ ↓ p Δ N → T (41) \begin{CD}
\Lambda_i^N @>>> S \\
@VVV @VVpV \\
\Delta^N @>>> T \tag*{(41)}
\end{CD} Λ i N ↓ ⏐ Δ N S ↓ ⏐ p T (41)
If T = Δ 0 T=\Delta^0 T = Δ 0 is a point, this is equivalent to S S S being an ∞ \infty ∞ -category. More generally it implies that the fibres of p : S → T p:S\to T p : S → T are ∞ \infty ∞ -categories. If T T T is an ∞ \infty ∞ -category, then S S S must be an ∞ \infty ∞ -category.
Similarly, we call p : S → T p:S\to T p : S → T a left fibration if we can complete (2.8) for all left horns Λ i N \Lambda_i^N Λ i N with 0 ≤ i < N 0\leq i<N 0 ≤ i < N , for N ≥ 1 N\geq1 N ≥ 1 , and a right fibration if we can complete (2.8) for all right horns Λ i N \Lambda_i^N Λ i N with 0 < i ≤ N 0<i\leq N 0 < i ≤ N , for N ≥ 1 N\geq1 N ≥ 1 .
Let p : S → T p:S\to T p : S → T be an inner fibration. A morphism f : Δ 1 → S f:\Delta^1\to S f : Δ 1 → S is called p p p -cocartesian if for all N ≥ 2 N\geq2 N ≥ 2 and all commutative diagrams of morphisms ‘→ \to → ’ as follows can be completed with a map ‘⇢ \dashrightarrow ⇢ ’ in a commutative diagram
Δ 1 → f S Δ { 0 , 1 } ↓ ↓ p Λ 0 N → Δ N (42) \begin{CD}
\Delta^1 @>{f}>> S \\
@V{\Delta^{\{0,1\}}}VV @VVpV \\
\Lambda_0^N @>>> \Delta^N \tag*{(42)}
\end{CD} Δ 1 Δ { 0 , 1 } ↓ ⏐ Λ 0 N f S ↓ ⏐ p Δ N (42)
Let p : S → T p:S\to T p : S → T be an inner fibration. We say that p p p is a cocartesian fibration if any commutative square of maps of simplicial sets ‘→ \to → ’ as follows can be completed with a map ‘⇢ \dashrightarrow ⇢ ’ in a commutative diagram
Δ 0 → f ~ S Δ { 0 } ↓ ↓ p Δ 1 → f T (43) \begin{CD}
\Delta^0 @>{\widetilde{f}}>> S \\
@V{\Delta^{\{0\}}}VV @VVpV \\
\Delta^1 @>{f}>> T \tag*{(43)}
\end{CD} Δ 0 Δ { 0 } ↓ ⏐ Δ 1 f f S ↓ ⏐ p T (43)
such that f ~ \widetilde{f} f is p p p -cocartesian.
If p p p is a cocartesian fibration and we are given an outer commutative square (2.10), then the space of p p p -cocartesian lifts f ~ \widetilde{f} f making (2.10) commute is a contractible Kan complex. That is, f ~ \widetilde{f} f is unique up to homotopy in a strong sense.
By Lurie [63, Prop. 5.1.4.15] and Land [55, Lem. 3.1.22], we have
Proposition 2.22. A map of simplicial sets p : S → T p:S\to T p : S → T is a left fibration if and only if it is cocartesian and every edge of S S S is p p p -cocartesian.
The next theorem comes from Lurie [60, §3.2–§3.3] and Land [55, §3.3].
Theorem 2.23. There is a universal cocartesian fibration u : F → C a t ∞ u:\mathcal{F}\to\mathcal{C}\mathrm{at}_{\infty} u : F → C at ∞ with the following properties:
(a) Every ∞ \infty ∞ -category C \mathcal{C} C is an object in C a t ∞ \mathcal{C}\mathrm{at}_{\infty} C at ∞ , and so corresponds to a morphism π C : Δ 0 → C a t ∞ \pi_{\mathcal{C}}:\Delta^{0}\to\mathcal{C}\mathrm{at}_{\infty} π C : Δ 0 → C at ∞ . There is a homotopy Cartesian square of ∞ \infty ∞ -categories
C → Φ C F ↓ ↓ u Δ 0 → π C C a t ∞ \begin{CD}
\mathcal{C} @>{\Phi_{\mathcal{C}}}>> \mathcal{F} \\
@V{}VV @VV{u}V \\
\Delta^{0} @>{\pi_{\mathcal{C}}}>> \mathcal{C}\mathrm{at}_{\infty}
\end{CD} C ↓ ⏐ Δ 0 Φ C π C F ↓ ⏐ u C at ∞ That is, the fibre F C \mathcal{F}_{\mathcal{C}} F C of u : F → C a t ∞ u:\mathcal{F}\to\mathcal{C}\mathrm{at}_{\infty} u : F → C at ∞ over each object C \mathcal{C} C in C a t ∞ \mathcal{C}\mathrm{at}_{\infty} C at ∞ is (Joyal equivalent to) C \mathcal{C} C .
(b) Let p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D be a cocartesian fibration of ∞ \infty ∞ -categories. Then there is an ∞ \infty ∞ -functor f : D → C a t ∞ f:\mathcal{D}\to\mathcal{C}\mathrm{at}_{\infty} f : D → C at ∞ , unique up to a contractible choice, which fits into a homotopy Cartesian square of ∞ \infty ∞ -categories
C → F p ↓ ↓ u D → f C a t ∞ (44) \begin{CD}
\mathcal{C} @>>> \mathcal{F} \\
@V{p}VV @VV{u}V \\
\mathcal{D} @>{f}>> \mathcal{C}\mathrm{at}_{\infty} \tag*{(44)}
\end{CD} C p ↓ ⏐ D f F ↓ ⏐ u C at ∞ (44) We call f f f the straightening of p p p .
(c) Let f : D → C a t ∞ f:\mathcal{D}\to\mathcal{C}\mathrm{at}_{\infty} f : D → C at ∞ be an ∞ \infty ∞ -functor of ∞ \infty ∞ -categories. Then there is a cocartesian fibration p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D , unique up to a contractible choice, which fits into a homotopy Cartesian square (2.11). We call p p p the unstraightening of f f f .
In (b),(c), part (a) implies that for all objects X ∈ D X\in\mathcal{D} X ∈ D , the image f ( X ) ∈ C a t ∞ f(X)\in\mathcal{C}\mathrm{at}_{\infty} f ( X ) ∈ C at ∞ , considered as an ∞ \infty ∞ -category, is Joyal equivalent to the fibre p − 1 ( X ) p^{-1}(X) p − 1 ( X ) .
The next example summarizes parts of Lurie [60, §3.2.5].
Example 2.24. Let D \mathcal{D} D be an ordinary category, and G : D → C a t ∞ 1 G:\mathcal{D}\to\mathrm{Cat}^{1}_{\infty} G : D → Cat ∞ 1 a functor, where C a t ∞ 1 ⊂ S S e t \mathrm{Cat}^{1}_{\infty}\subset\mathrm{SSet} Cat ∞ 1 ⊂ SSet is the ordinary category of ∞ \infty ∞ -categories from Definition 2.9. Definition 2.18(b) gives an ∞ \infty ∞ -functor I C a t ∞ : N ( C a t ∞ 1 ) ↪ C a t ∞ I^{\mathrm{Cat}_{\infty}}:\mathcal{N}(\mathrm{Cat}^{1}_{\infty})\hookrightarrow\mathcal{C}\mathrm{at}_{\infty} I Cat ∞ : N ( Cat ∞ 1 ) ↪ C at ∞ . So we can consider the composition ∞ \infty ∞ -functor I C a t ∞ ∘ N ( G ) : N ( D ) → C a t ∞ I^{\mathrm{Cat}_{\infty}}\circ\mathcal{N}(G):\mathcal{N}(\mathcal{D})\to\mathcal{C}\mathrm{at}_{\infty} I Cat ∞ ∘ N ( G ) : N ( D ) → C at ∞ . Theorem 2.23 then gives a cocartesian fibration p : C → N ( D ) p:\mathcal{C}\to\mathcal{N}(\mathcal{D}) p : C → N ( D ) , the unstraightening of I C a t ∞ ∘ N ( G ) I^{\mathrm{Cat}_{\infty}}\circ\mathcal{N}(G) I Cat ∞ ∘ N ( G ) , uniquely up to a contractible choice.
Lurie [60, Def. 3.2.5.2] writes down an explicit simplicial set N G ( D ) N_{G}(\mathcal{D}) N G ( D ) and an ∞ \infty ∞ -functor π : N G ( D ) → N ( D ) \pi:N_{G}(\mathcal{D})\to\mathcal{N}(\mathcal{D}) π : N G ( D ) → N ( D ) , the relative simplicial nerve of G G G , such that π : N G ( D ) → N ( D ) \pi:N_{G}(\mathcal{D})\to\mathcal{N}(\mathcal{D}) π : N G ( D ) → N ( D ) is a possible choice for the unstraightening p : C → N ( D ) p:\mathcal{C}\to\mathcal{N}(\mathcal{D}) p : C → N ( D ) .
If N ≥ 0 N\geq0 N ≥ 0 , an N N N -simplex ( S , ( τ ( I ) ) ∅ ≠ I ⊆ [ N ] ) (S,(\tau(I))_{\varnothing\neq I\subseteq[N]}) ( S , ( τ ( I ) ) ∅ = I ⊆ [ N ] ) of N G ( D ) N_{G}(\mathcal{D}) N G ( D ) consists of the data:
(i) an N N N -simplex S S S of N ( D ) \mathcal{N}(\mathcal{D}) N ( D ) . Here S S S is equivalent to objects X 0 , … , X N X_{0},\ldots,X_{N} X 0 , … , X N in D \mathcal{D} D and morphisms f i j : X i → X j f^{ij}:X_{i}\to X_{j} f ij : X i → X j in D \mathcal{D} D for 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , with f i i = id X i f^{ii}=\operatorname{id}_{X_{i}} f ii = id X i and f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik for 0 ≤ i ≤ j ≤ k ≤ N 0\leq i\leq j\leq k\leq N 0 ≤ i ≤ j ≤ k ≤ N .
(ii) For all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\neq I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } , a map of simplicial sets τ ( I ) : Δ I → G ( X max I ) \tau(I):\Delta^{I}\to G(X_{\max I}) τ ( I ) : Δ I → G ( X m a x I ) , where Δ I \Delta^{I} Δ I is the simplex with ordered vertices ( I , ≤ ) (I,\leq) ( I , ≤ ) , considered as a simplicial set, so that Δ I ≅ Δ ∣ I ∣ − 1 \Delta^{I}\cong\Delta^{|I|-1} Δ I ≅ Δ ∣ I ∣ − 1 . (iii) For all ∅ ≠ J ⊆ I ⊆ { 0 , … , N } \varnothing\ne J \subseteq I \subseteq\{0,\ldots,N\} ∅ = J ⊆ I ⊆ { 0 , … , N } , the following diagram of simplicial sets must commute:
Δ J → τ ( J ) G ( X max J ) Δ inc ( J ↪ I ) ↓ ↓ G ( f max I max J ) Δ I → τ ( I ) G ( X max I ) (45) \begin{CD}
\Delta^{J} @>{\tau(J)}>> G(X_{\max J}) \\
@V{\Delta^{\operatorname{inc}(J\hookrightarrow I)}}VV @VV{G(f^{\max J}_{\max I})}V \\
\Delta^{I} @>{\tau(I)}>> G(X_{\max I}) \tag*{(45)}
\end{CD} Δ J Δ inc ( J ↪ I ) ↓ ⏐ Δ I τ ( J ) τ ( I ) G ( X m a x J ) ↓ ⏐ G ( f m a x I m a x J ) G ( X m a x I ) (45)
with Δ inc ( J ↪ I ) : Δ J → Δ I \Delta^{\operatorname{inc}(J\hookrightarrow I)}:\Delta^{J}\to\Delta^{I} Δ inc ( J ↪ I ) : Δ J → Δ I the inclusion of simplices induced by J ↪ I J\hookrightarrow I J ↪ I .
These N N N -simplices have the obvious functoriality under morphisms α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] in Δ o p \Delta^{\mathrm{op}} Δ op , and so define a simplicial set N G ( D ) N_G(\mathcal{D}) N G ( D ) . The morphism π : N G ( D ) → N ( D ) \pi:N_G(\mathcal{D})\to\mathcal{N}(\mathcal{D}) π : N G ( D ) → N ( D ) acts on N N N -simplices by π : ( S , ( τ ( I ) ) ∅ ≠ I ⊆ [ N ] ) ↦ S \pi:(S,(\tau(I))_{\varnothing\ne I\subseteq[N]})\mapsto S π : ( S , ( τ ( I ) ) ∅ = I ⊆ [ N ] ) ↦ S .
If X X X is an object in D \mathcal{D} D , and thus a 0 0 0 -simplex in N ( D ) \mathcal{N}(\mathcal{D}) N ( D ) , consider the fibre π − 1 ( X ) ⊆ N G ( D ) \pi^{-1}(X)\subseteq N_G(\mathcal{D}) π − 1 ( X ) ⊆ N G ( D ) . An N N N -simplex ( S , ( τ ( I ) ) ∅ ≠ I ⊆ [ N ] ) (S,(\tau(I))_{\varnothing\ne I\subseteq[N]}) ( S , ( τ ( I ) ) ∅ = I ⊆ [ N ] ) in π − 1 ( X ) \pi^{-1}(X) π − 1 ( X ) has S S S the constant N N N -simplex at X X X , and the maps τ ( I ) : Δ I → G ( X ) \tau(I):\Delta^{I}\to G(X) τ ( I ) : Δ I → G ( X ) for I ⊆ [ N ] I\subseteq[N] I ⊆ [ N ] are uniquely determined by the N N N -simplex τ ( [ N ] ) \tau([N]) τ ([ N ]) in G ( X ) G(X) G ( X ) . Hence there is a canonical strict isomorphism of simplicial sets π − 1 ( X ) ≅ G ( X ) \pi^{-1}(X)\cong G(X) π − 1 ( X ) ≅ G ( X ) .
The ∞ \infty ∞ -category of spaces, mapping space functors, and Yoneda embedding
Definition 2.25. We will write S p c \mathcal{S}pc S p c for the ∞ \infty ∞ -category of spaces. Objects of S p c \mathcal{S}pc S p c are ‘spaces’ (roughly, topological spaces up to homotopy), or ∞ \infty ∞ -groupoids. It is a complete and cocomplete ∞ \infty ∞ -category.
There are several different ways to define S p c \mathcal{S}pc S p c in the literature, which yield equivalent ∞ \infty ∞ -categories. We follow Lurie [60, §1.2.16], [63, §3.1.6] and Land [55, Def. 1.3.41], and define S p c \mathcal{S}pc S p c to be the simplicial nerve N ( K a n △ ) \mathcal{N}(\mathrm{Kan}^{\triangle}) N ( Kan △ ) of the simplicial category K a n △ \mathrm{Kan}^{\triangle} Kan △ of Kan complexes, which is the full simplicial subcategory with objects Kan complexes in the simplicial category S S e t △ \mathrm{SSet}^{\triangle} SSet △ of simplicial sets. Then S p c \mathcal{S}pc S p c is a full ∞ \infty ∞ -subcategory of C a t ∞ \mathsf{Cat}_{\infty} Cat ∞ in Definition 2.18.
Thus, objects of S p c \mathcal{S}pc S p c are Kan complexes. A morphism f : X → Y f:X\to Y f : X → Y in S p c \mathcal{S}pc S p c is an equivalence in S p c \mathcal{S}pc S p c if and only if it is a homotopy equivalence of Kan complexes.
The ordinary category K a n \mathrm{Kan} Kan of Kan complexes, a full subcategory K a n ⊂ C a t ∞ 1 ⊂ S S e t \mathrm{Kan}\subset\mathrm{Cat}^{1}_{\infty}\subset\mathrm{SSet} Kan ⊂ Cat ∞ 1 ⊂ SSet , is a subcategory of K a n △ \mathrm{Kan}^{\triangle} Kan △ (taking only 0 0 0 -simplices in mapping spaces in K a n △ \mathrm{Kan}^{\triangle} Kan △ ). Thus as for I C a t ∞ I\mathrm{Cat}_{\infty} I Cat ∞ in Definition 2.18(b), the inclusion K a n ↪ K a n △ \mathrm{Kan}\hookrightarrow\mathrm{Kan}^{\triangle} Kan ↪ Kan △ induces an inclusion of ∞ \infty ∞ -subcategories I K a n : N ( K a n ) ↪ S p c I^{\mathrm{Kan}}:\mathcal{N}(\mathrm{Kan})\hookrightarrow\mathcal{S}pc I Kan : N ( Kan ) ↪ S p c .
Land [55, Def. 1.2.72] defines an ∞ \infty ∞ -category with objects CW-complexes (nice topological spaces), and shows [55, Obs. 1.3.43] it is equivalent to S p c \mathcal{S}pc S p c . Cisinski [17, Def. 5.2.3] gives an alternative definition, yielding an equivalent ∞ \infty ∞ -category [17, Rem.s 5.4.11 & 7.8.10].
Definition 2.26. Let C \mathcal{C} C be an ∞ \infty ∞ -category. Then there is an ∞ \infty ∞ -functor Hom C : C o p × C → S p c \operatorname{Hom}_{\mathcal{C}}:\mathcal{C}^{\mathrm{op}}\times\mathcal{C}\to\mathcal{S}pc Hom C : C op × C → S p c , with a difficult definition in Lurie [61, §5.2.1, Prop. 5.2.11] (see also [60, §5.1.3]), Cisinski [17, §5.8.1], and Land [55, Def. 4.2.5], such that for objects X , Y ∈ C X,Y\in\mathcal{C} X , Y ∈ C we have a homotopy equivalence Hom C ( X , Y ) ≃ Hom C L ( X , Y ) \operatorname{Hom}_{\mathcal{C}}(X,Y)\simeq\operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C ( X , Y ) ≃ Hom C L ( X , Y ) (i.e. an equivalence in S p c \mathcal{S}\mathrm{pc} S pc ), where Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) is the Kan complex in Definition 2.15. That is, Hom C \operatorname{Hom}_{\mathcal{C}} Hom C makes mapping spaces Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) (or just Hom C ( X , Y ) \operatorname{Hom}_{\mathcal{C}}(X,Y) Hom C ( X , Y ) ) in §2.3 functorial in X , Y X,Y X , Y .
Definition 2.27. Let C \mathcal{C} C be an ∞ \infty ∞ -category. As in Lurie [60, §5.1] define the ∞ \infty ∞ -category PSh ( C ) \operatorname{PSh}(\mathcal{C}) PSh ( C ) of presheaves on C \mathcal{C} C to be the ∞ \infty ∞ -category Fun ( C o p , S p c ) \operatorname{Fun}(\mathcal{C}^{\mathrm{op}},\mathcal{S}\mathrm{pc}) Fun ( C op , S pc ) of ∞ \infty ∞ -functors G : C o p → S p c G:\mathcal{C}^{\mathrm{op}}\to\mathcal{S}\mathrm{pc} G : C op → S pc . Then PSh ( C ) \operatorname{PSh}(\mathcal{C}) PSh ( C ) is complete and cocomplete [60, Cor. 5.1.2.4], and (co)limits in PSh ( C ) \operatorname{PSh}(\mathcal{C}) PSh ( C ) may be computed objectwise in C o p \mathcal{C}^{\mathrm{op}} C op .
If F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D is an ∞ \infty ∞ -functor there is a pullback ∞ \infty ∞ -functor F ∗ : PSh ( D ) → PSh ( C ) F^{*}:\operatorname{PSh}(\mathcal{D})\to\operatorname{PSh}(\mathcal{C}) F ∗ : PSh ( D ) → PSh ( C ) mapping G ↦ G ∘ F G\mapsto G\circ F G ↦ G ∘ F . This F ∗ F^{*} F ∗ has left and right adjoints, given by left and right Kan extension. We denote the left adjoint by F ! : PSh ( C ) → PSh ( D ) F_{!}:\operatorname{PSh}(\mathcal{C})\to\operatorname{PSh}(\mathcal{D}) F ! : PSh ( C ) → PSh ( D ) .
Definition 2.28. The Yoneda functor Y C : C → PSh ( C ) \mathcal{Y}_{\mathcal{C}}:\mathcal{C}\to\operatorname{PSh}(\mathcal{C}) Y C : C → PSh ( C ) maps Y C : X ↦ Hom C ( − , X ) \mathcal{Y}_{\mathcal{C}}:X\mapsto\operatorname{Hom}_{\mathcal{C}}(-,X) Y C : X ↦ Hom C ( − , X ) . As in Lurie [60, §5.1.3], Cisinski [17, Th. 5.8.13], and Land [55, Prop. 4.2.11], this is a full and faithful ∞ \infty ∞ -functor, so we call it the Yoneda embedding . Objects of PSh ( C ) \operatorname{PSh}(\mathcal{C}) PSh ( C ) in the essential image of Y C \mathcal{Y}_{\mathcal{C}} Y C are called representable . Write Rep ( C ) \operatorname{Rep}(\mathcal{C}) Rep ( C ) for the full ∞ \infty ∞ -subcategory of representable objects in PSh ( C ) \operatorname{PSh}(\mathcal{C}) PSh ( C ) . Then Y C : C → Rep ( C ) \mathcal{Y}_{\mathcal{C}}:\mathcal{C}\to\operatorname{Rep}(\mathcal{C}) Y C : C → Rep ( C ) is an equivalence of ∞ \infty ∞ -categories.
If F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D is an ∞ \infty ∞ -functor then by Lurie [60, Prop. 5.2.6.3] the following commutes up to natural equivalence:
C → Y C PSh ( C ) F ↓ ↓ F ! D → Y D PSh ( D ) . (46) \begin{CD}
\mathcal{C} @>{\mathcal{Y}_{\mathcal{C}}}>> \operatorname{PSh}(\mathcal{C}) \\
@V{F}VV @VV{F_{!}}V \\
\mathcal{D} @>{\mathcal{Y}_{\mathcal{D}}}>> \operatorname{PSh}(\mathcal{D}). \tag*{(46)}
\end{CD} C F ↓ ⏐ D Y C Y D PSh ( C ) ↓ ⏐ F ! PSh ( D ) . (46)
In Theorem 2.23 we explained Lurie’s ‘straightening/unstraightening correspondence’ between cocartesian fibrations of ∞ \infty ∞ -categories p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D and ∞ \infty ∞ -functors f : D → C a t ∞ f:\mathcal{D}\to\mathcal{C}\mathrm{at}_{\infty} f : D → C at ∞ . As above S p c \mathcal{S}\mathrm{pc} S pc is a full ∞ \infty ∞ -subcategory of C a t ∞ \mathcal{C}\mathrm{at}_{\infty} C at ∞ . Definition 2.21 defined left fibrations p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D , which are special examples of cocartesian fibrations by Proposition 2.22. If p p p is a cocartesian fibration then the fibres p − 1 ( X ) p^{-1}(X) p − 1 ( X ) for X ∈ D X\in\mathcal{D} X ∈ D are ∞ \infty ∞ -categories, but if p p p is a left fibration then the p − 1 ( X ) p^{-1}(X) p − 1 ( X ) are Kan complexes. Here is Land [55, Th. 3.3.16].
Theorem 2.29. The straightening/unstraightening correspondence in Theorem 2.23 (b),(c) restricts to a correspondence between left fibrations p : C → D p:\mathcal{C}\to\mathcal{D} p : C → D and ∞ \infty ∞ -functors f : D → S p c ⊂ C a t ∞ f:\mathcal{D}\to\mathcal{S}\mathrm{pc}\subset\mathcal{C}\mathrm{at}_{\infty} f : D → S pc ⊂ C at ∞ .
Example 2.30. Let C \mathcal{C} C be an ∞ \infty ∞ -category, and let X X X be an object (0-simplex) in C \mathcal{C} C . Define a simplicial set C X / \mathcal{C}_{X/} C X / by taking Hom S S e t ( Δ N , C X / ) \operatorname{Hom}_{\mathrm{SSet}}(\Delta^{N},\mathcal{C}_{X/}) Hom SSet ( Δ N , C X / ) to be the set of all morphisms Z : Δ N + 1 → C Z:\Delta^{N+1}\to\mathcal{C} Z : Δ N + 1 → C such that Z ∣ Δ { 0 } Z|_{\Delta^{\{0\}}} Z ∣ Δ { 0 } is the 0-simplex X X X . The face and degeneracy maps in Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) are induced from those in C \mathcal{C} C by extending an order-preserving function f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] to an order-preserving function f ~ : [ N + 1 ] → [ M + 1 ] \widetilde{f}:[N+1]\to[M+1] f : [ N + 1 ] → [ M + 1 ] with f ~ ( 0 ) = 0 \widetilde{f}(0)=0 f ( 0 ) = 0 and f ~ ( i ) = f ( i − 1 ) + 1 \widetilde{f}(i)=f(i-1)+1 f ( i ) = f ( i − 1 ) + 1 for i = 1 , … , N + 1 i=1,\ldots,N+1 i = 1 , … , N + 1 . Define a map ρ : C X / → C \rho:\mathcal{C}_{X/}\to\mathcal{C} ρ : C X / → C to map an N N N -simplex Z ∘ Δ { 1 , 2 , … , N + 1 } Z\circ\Delta^{\{1,2,\ldots,N+1\}} Z ∘ Δ { 1 , 2 , … , N + 1 } in C X / \mathcal{C}_{X/} C X / , thought of as an ( N + 1 ) (N+1) ( N + 1 ) -simplex Z : Δ N + 1 → C Z:\Delta^{N+1}\to\mathcal{C} Z : Δ N + 1 → C with Z ∘ Δ { 0 } = X Z\circ\Delta^{\{0\}}=X Z ∘ Δ { 0 } = X , to the N N N -simplex Z ∘ Δ { 1 , 2 , … , N + 1 } Z\circ\Delta^{\{1,2,\ldots,N+1\}} Z ∘ Δ { 1 , 2 , … , N + 1 } in C \mathcal{C} C .
By Lurie [60, Cor. 2.1.2.2], [63, Prop. 4.3.6.1], C X / \mathcal{C}_{X/} C X / is an ∞ \infty ∞ -category, called the undercategory of X X X , and p C : C X / → C p_{\mathcal{C}}:\mathcal{C}_{X/}\to\mathcal{C} p C : C X / → C is a left fibration. From Definition 2.15, we see that the fibre of p C p_{\mathcal{C}} p C over Y ∈ C Y\in\mathcal{C} Y ∈ C is the left morphisms Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) , a Kan complex. Theorem 2.29 says that p C p_{\mathcal{C}} p C corresponds to an ∞ \infty ∞ -functor F X : C → S p c F_X:\mathcal{C}\to\mathcal{S}_{\mathrm{pc}} F X : C → S pc , the straightening of p C p_{\mathcal{C}} p C , such that F X ( Y ) F_X(Y) F X ( Y ) is homotopy equivalent to Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) . As observed by Lurie [60, §2.1.2], this is a way to make mapping spaces Hom C ( X , − ) \operatorname{Hom}_{\mathcal{C}}(X,-) Hom C ( X , − ) into an ∞ \infty ∞ -functor C → S p c \mathcal{C}\to\mathcal{S}_{\mathrm{pc}} C → S pc .
Defining ∞ \infty ∞ -categories via localization
There are many situations in mathematics where one is given a category C \mathcal{C} C and a class of morphisms W \mathcal{W} W in C \mathcal{C} C (usually called weak equivalences ), and one wants to define a localized category C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] by inverting the morphisms in W \mathcal{W} W . That is, we should have a functor Π : C → C [ W − 1 ] \Pi:\mathcal{C}\to\mathcal{C}[\mathcal{W}^{-1}] Π : C → C [ W − 1 ] which maps morphisms in W \mathcal{W} W to isomorphisms, and this should be universal amongst functors Π : C → D \Pi:\mathcal{C}\to\mathcal{D} Π : C → D mapping W \mathcal{W} W to isomorphisms.
For example, the derived category D b A D^{b}\mathcal{A} D b A of an abelian category A \mathcal{A} A is obtained by inverting quasi-isomorphisms in a category of complexes Com b ( A ) \operatorname{Com}^{b}(\mathcal{A}) Com b ( A ) . Homotopy theory is the study of the ‘homotopy category’ obtained by localizing topological spaces T o p \mathbf{Top} Top along weak homotopy equivalences.
It turns out that there is a natural ∞ \infty ∞ -category C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ , whose homotopy category Ho ( C [ W − 1 ] ∞ ) \operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}) Ho ( C [ W − 1 ] ∞ ) is C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] . Often it is helpful to work in C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ rather than in C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] . For example, derived categories D b A D^{b}\mathcal{A} D b A suffer from ‘nonfunctoriality of the cone’, as in Gelfand and Manin [35, §IV.1.7], which is repaired at the ∞ \infty ∞ -category level. This is an important reason for studying ∞ \infty ∞ -categories.
We will discuss localizations of ordinary categories with weak equivalences. In fact localization works more generally (one can localize any ∞ \infty ∞ -category at any set of morphisms [61, Rem. 1.3.4.2], [63, §6.3.2], [55, Prop. 2.4.8]), but this is all we need. See Lurie [61, §1.3.4], [60, §5.2.7] and Land [55, §2.4].
Definition 2.31. A category with weak equivalences ( C , W ) (\mathcal{C},\mathcal{W}) ( C , W ) is an ordinary category C \mathcal{C} C and a subset W ⊆ Mor ( C ) \mathcal{W}\subseteq\operatorname{Mor}(\mathcal{C}) W ⊆ Mor ( C ) of the morphisms of C \mathcal{C} C , satisfying
(a) All isomorphisms in C \mathcal{C} C lie in W \mathcal{W} W .
(b) (The ‘two out of three’ property.) If f : X → Y f:X\to Y f : X → Y , g : Y → Z g:Y\to Z g : Y → Z are morphisms in C \mathcal{C} C and two out of three of f f f , g g g , g ∘ f g\circ f g ∘ f lie in W \mathcal{W} W , then the third lies in W \mathcal{W} W .
Let ( C , W ) (\mathcal{C},\mathcal{W}) ( C , W ) be a category with weak equivalences. A Dwyer–Kan localization of C \mathcal{C} C along W \mathcal{W} W is an ∞ \infty ∞ -category L \mathcal{L} L and an ∞ \infty ∞ -functor λ : N ( C ) → L \lambda:N(\mathcal{C})\to\mathcal{L} λ : N ( C ) → L , such that if f ∈ W f\in\mathcal{W} f ∈ W then λ ( f ) \lambda(f) λ ( f ) is an equivalence in L \mathcal{L} L (that is, on homotopy categories Ho ( λ ) : C ≃ Ho ( N ( C ) ) → Ho ( L ) \operatorname{Ho}(\lambda):\mathcal{C}\simeq\operatorname{Ho}(N(\mathcal{C}))\to\operatorname{Ho}(\mathcal{L}) Ho ( λ ) : C ≃ Ho ( N ( C )) → Ho ( L ) maps morphisms in W \mathcal{W} W to isomorphisms), and λ \lambda λ has the universal property that whenever F ~ : N ( C ) → D \widetilde{\mathcal{F}}:N(\mathcal{C})\to\mathcal{D} F : N ( C ) → D is an ∞ \infty ∞ -functor which maps W \mathcal{W} W to equivalences in D \mathcal{D} D , then we can factor F ~ \widetilde{\mathcal{F}} F as
N ( C ) → λ L F ~ ↓ ↓ G D = D (47) \begin{CD}
N(\mathcal{C}) @>{\lambda}>> \mathcal{L} \\
@V{\widetilde{\mathcal{F}}}VV @VV{\mathcal{G}}V \\
\mathcal{D} @= \mathcal{D} \tag*{(47)}
\end{CD} N ( C ) F ↓ ⏐ D λ L ↓ ⏐ G D (47)
for some G : L → D \mathcal{G}:\mathcal{L}\to\mathcal{D} G : L → D unique up to a contractible choice. (See [55, §2.4] for a precise statement.)
Dwyer–Kan localizations L \mathcal{L} L always exist, and are unique up to Joyal equivalence. There is an explicit construction, by first taking the ‘Dwyer–Kan hammock localization’ to give a simplicial category, and then taking the simplicial nerve of this. We will use the notation C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ for this explicit localization, and Π l o c ∞ : N ( C ) → C [ W − 1 ] ∞ \Pi_{\mathrm{loc}}^{\infty}:\mathcal{N}(\mathcal{C})\to\mathcal{C}[\mathcal{W}^{-1}]^{\infty} Π loc ∞ : N ( C ) → C [ W − 1 ] ∞ for the localization ∞ \infty ∞ -functor.
As in Definition 2.27 we have a pullback ∞ \infty ∞ -functor
( Π l o c ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) ⟶ PSh ( N ( C ) ) . (\Pi_{\mathrm{loc}}^{\infty})^{*}:\operatorname{PSh}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty})\longrightarrow\operatorname{PSh}(\mathcal{N}(\mathcal{C})). ( Π loc ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) ⟶ PSh ( N ( C )) .
The universal property of C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ implies that ( Π l o c ∞ ) ∗ (\Pi_{\mathrm{loc}}^{\infty})^{*} ( Π loc ∞ ) ∗ is full and faithful, with essential image those ∞ \infty ∞ -functors N ( C ) o p → S p c \mathcal{N}(\mathcal{C})^{\mathrm{op}}\to\mathcal{S}\mathrm{pc} N ( C ) op → S pc which map morphisms in W o p \mathcal{W}^{\mathrm{op}} W op to equivalences in S p c \mathcal{S}\mathrm{pc} S pc . Therefore C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ is equivalent to a full ∞ \infty ∞ -subcategory of PSh ( N ( C ) ) \operatorname{PSh}(\mathcal{N}(\mathcal{C})) PSh ( N ( C )) , the essential image of Rep ( C [ W − 1 ] ∞ ) \operatorname{Rep}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}) Rep ( C [ W − 1 ] ∞ ) under ( Π l o c ∞ ) ∗ (\Pi_{\mathrm{loc}}^{\infty})^{*} ( Π loc ∞ ) ∗ .
Remark 2.32. In Definition 2.31, we can also consider the localization C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] of C \mathcal{C} C along W \mathcal{W} W just in ordinary categories, not in ∞ \infty ∞ -categories. It turns out that there is an equivalence of categories C [ W − 1 ] ≃ Ho ( C [ W − 1 ] ∞ ) \mathcal{C}[\mathcal{W}^{-1}]\simeq\operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}) C [ W − 1 ] ≃ Ho ( C [ W − 1 ] ∞ ) , canonical up to natural isomorphism. To see why, suppose that f : C → D f:\mathcal{C}\to\mathcal{D} f : C → D is a functor of ordinary categories which maps W \mathcal{W} W to isomorphisms in D \mathcal{D} D . Then f f f should factor via the localization C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] as in
C → Π l o c C [ W − 1 ] → g D . \mathcal{C}\xrightarrow{\Pi_{\mathrm{loc}}}\mathcal{C}[\mathcal{W}^{-1}]\xrightarrow{g}\mathcal{D}. C Π loc C [ W − 1 ] g D .
where g g g is natural up to natural isomorphism.
Now consider the homotopy commutative ∞ \infty ∞ -category diagram
N ( C ) → N ( λ ) N ( Ho ( C [ W − 1 ] ∞ ) ) ↓ Π l o c ∞ ↓ N ( g ) C [ W − 1 ] ∞ → Π H o N ( D ) \begin{aligned}
\mathcal{N}(\mathcal{C})&\xrightarrow{\mathcal{N}(\lambda)}\mathcal{N}(\operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}))\\
\downarrow{\Pi_{\mathrm{loc}}^{\infty}}&\qquad\qquad\qquad\qquad\downarrow{\mathcal{N}(g)}\\
\mathcal{C}[\mathcal{W}^{-1}]^{\infty}&\xrightarrow{\Pi_{\mathrm{Ho}}}\mathcal{N}(\mathcal{D})\\
\end{aligned} N ( C ) ↓ Π loc ∞ C [ W − 1 ] ∞ N ( λ ) N ( Ho ( C [ W − 1 ] ∞ )) ↓ N ( g ) Π Ho N ( D )
where Π H o \Pi_{\mathrm{Ho}} Π Ho is as in (2.7). Here N ( f ) \mathcal{N}(f) N ( f ) factors via G \mathcal{G} G , uniquely up to a contractible choice, by the universal property of C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ . But as the target of G \mathcal{G} G is N ( D ) \mathcal{N}(\mathcal{D}) N ( D ) , G \mathcal{G} G factors via Π H o \Pi_{\mathrm{Ho}} Π Ho , and any ∞ \infty ∞ -functor N ( Ho ( C [ W − 1 ] ∞ ) ) → N ( D ) \mathcal{N}(\operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}))\to\mathcal{N}(\mathcal{D}) N ( Ho ( C [ W − 1 ] ∞ )) → N ( D ) is N ( g ) \mathcal{N}(g) N ( g ) for some functor g : Ho ( C [ W − 1 ] ∞ ) → D g:\operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty})\to\mathcal{D} g : Ho ( C [ W − 1 ] ∞ ) → D . It follows that Ho ( C [ W − 1 ] ∞ ) \operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}) Ho ( C [ W − 1 ] ∞ ) satisfies the universal property for C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] , and C [ W − 1 ] ≃ Ho ( C [ W − 1 ] ∞ ) \mathcal{C}[\mathcal{W}^{-1}]\simeq\operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]^{\infty}) C [ W − 1 ] ≃ Ho ( C [ W − 1 ] ∞ ) .
Thus, we can consider C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]^{\infty} C [ W − 1 ] ∞ as an ∞ \infty ∞ -categorical enhancement of the ordinary category localization C [ W − 1 ] \mathcal{C}[\mathcal{W}^{-1}] C [ W − 1 ] .
The next proposition, which is easy to prove, is used in [12, 76] to prove Theorem 4.64 below.
Proposition 2.33. Suppose ( C , W ) (C,W) ( C , W ) and ( C ′ , W ′ ) (C',W') ( C ′ , W ′ ) are categories with weak equivalences, and we are given adjoint functors
C → F C ′ G ↑ \begin{CD}
C @>F>> C'\\
@A G AA
\end{CD} C G ⏐ ↑ F C ′ satisfying
(i) F ( W ) ⊆ W ′ F(W) \subseteq W' F ( W ) ⊆ W ′ and G ( W ′ ) ⊆ W G(W') \subseteq W G ( W ′ ) ⊆ W .
(ii) Write ζ : Id C ⇒ G ∘ F \zeta:\operatorname{Id}_C \Rightarrow G\circ F ζ : Id C ⇒ G ∘ F and η : F ∘ G ⇒ Id C ′ \eta:F\circ G \Rightarrow\operatorname{Id}_{C'} η : F ∘ G ⇒ Id C ′ for the unit and counit of the adjunction. Then ζ ( X ) ∈ W \zeta(X)\in W ζ ( X ) ∈ W and η ( X ′ ) ∈ W ′ \eta(X')\in W' η ( X ′ ) ∈ W ′ for all X ∈ C X\in C X ∈ C and X ′ ∈ C ′ X'\in C' X ′ ∈ C ′ .
Then F , G F,G F , G induce quasi-inverse Joyal equivalences of localized ∞ \infty ∞ -categories:
C [ W − 1 ] ∞ ⇄ F ∗ G ∗ C ′ [ W ′ − 1 ] ∞ C[W^{-1}]_{\infty} \underset{G_*}{\overset{F_*}{\rightleftarrows}} C'[W'^{-1}]_{\infty} C [ W − 1 ] ∞ G ∗ ⇄ F ∗ C ′ [ W ′ − 1 ] ∞
The next theorem follows from Land [55, Lem. 2.4.9 & Th. 2.4.10]. It describes C a t ∞ , S p c \mathcal{C}\!at_{\infty},\mathcal{S}pc C a t ∞ , S p c in Definitions 2.18 and 2.25 as localizations.
Theorem 2.34. (a) Let I C a t ∞ : N ( C a t ∞ 1 ) → C a t ∞ I^{\mathcal{C}\!at_{\infty}}:\mathcal{N}(\mathcal{C}\!at_{\infty}^{1})\to\mathcal{C}\!at_{\infty} I C a t ∞ : N ( C a t ∞ 1 ) → C a t ∞ be as in Definition 2.18(b). Then I C a t ∞ I^{\mathcal{C}\!at_{\infty}} I C a t ∞ maps Joyal equivalences to equivalences in C a t ∞ \mathcal{C}\!at_{\infty} C a t ∞ by Definition 2.19. Hence as in (2.14) we have a canonical factorization
N ( C a t ∞ 1 ) → Π l o c ∞ C a t ∞ 1 [ J o y − 1 ] ∞ \mathcal{N}(\mathcal{C}\!at_{\infty}^{1}) \xrightarrow{\Pi_{\mathrm{loc}}^{\infty}} \mathcal{C}\!at_{\infty}^{1}[\mathrm{Joy}^{-1}]_{\infty} N ( C a t ∞ 1 ) Π loc ∞ C a t ∞ 1 [ Joy − 1 ] ∞ where J o y ⊂ Mor ( C a t ∞ 1 ) \mathrm{Joy}\subset\operatorname{Mor}(\mathcal{C}\!at_{\infty}^{1}) Joy ⊂ Mor ( C a t ∞ 1 ) is the set of Joyal equivalences. Then Π \Pi Π is a Joyal equivalence of ∞ \infty ∞ -categories.
(b) The category of simplicial sets SSet has the Quillen model category structure , in which the weak equivalences are weak homotopy equivalences (morphisms of simplicial sets which induce weak homotopy equivalences of the topological realizations). All objects are cofibrant, and fibrant objects are Kan complexes. Write Kan for the full subcategory of Kan complexes in SSet . Then weak homotopy equivalences in Kan are homotopy equivalences.
We have Joyal equivalences of ∞ \infty ∞ -categories
S S e t [ W E − 1 ] ∞ ≃ K a n [ H E − 1 ] ∞ ≃ S p c , (48) \mathbf{SSet}[\mathcal{W}\mathcal{E}^{-1}]_{\infty}\simeq\mathbf{Kan}[\mathcal{H}\mathcal{E}^{-1}]_{\infty}\simeq\mathcal{S}pc, \tag*{(48)} SSet [ W E − 1 ] ∞ ≃ Kan [ H E − 1 ] ∞ ≃ S p c , (48) where W E \mathcal{W}\mathcal{E} W E are the weak homotopy equivalences in SSet , and H E \mathcal{H}\mathcal{E} HE are the homotopy equivalences in Kan . Thus as for (2.15) we have a canonical factorization
N ( K a n ) → Π l o c ∞ K a n [ H E − 1 ] ∞ \mathcal{N}(\mathbf{Kan}) \xrightarrow{\Pi_{\mathrm{loc}}^{\infty}} \mathbf{Kan}[\mathcal{H}\mathcal{E}^{-1}]_{\infty} N ( Kan ) Π loc ∞ Kan [ H E − 1 ] ∞ where I K a n I^{\mathbf{Kan}} I Kan is as in Definition 2.25, and Π \Pi Π is a Joyal equivalence.
In several important situations there are tools for describing the morphisms in C [ W − 1 ] C[W^{-1}] C [ W − 1 ] , and the homotopy types of the mapping spaces in C [ W − 1 ] ∞ C[W^{-1}]^{\infty} C [ W − 1 ] ∞ . One such situation is the theory of model categories (or simplicial model categories).
full subcategory of objects with V = Y V=Y V = Y and j = id Y j=\operatorname{id}_Y j = id Y , and dually, if X X X is cofibrant then Hom C [ W − 1 ] ∞ L ( X , Y ) ≃ N ( M ( X , Y ) i = id ) \operatorname{Hom}^{L}_{C[W^{-1}]_{\infty}}(X,Y)\simeq\mathcal{N}\bigl(\mathcal{M}(X,Y)_{i=\operatorname{id}}\bigr) Hom C [ W − 1 ] ∞ L ( X , Y ) ≃ N ( M ( X , Y ) i = id ) , where M ( X , Y ) i = id ⊂ M ( X , Y ) \mathcal{M}(X,Y)_{i=\operatorname{id}}\subset\mathcal{M}(X,Y) M ( X , Y ) i = id ⊂ M ( X , Y ) is the full subcategory of objects with U = X U=X U = X and i = id X i=\operatorname{id}_X i = id X .
(c) A second way to compute mapping spaces Hom C [ W − 1 ] ∞ L ( X , Y ) \operatorname{Hom}^{L}_{C[W^{-1}]_{\infty}}(X,Y) Hom C [ W − 1 ] ∞ L ( X , Y ) is the theory of framings in Hirschhorn [42, §16] and Hovey [44, §5]. As in [42, Def. 16.1.2], a simplicial resolution of an object Y Y Y in C C C is a functor S : Δ op → C S:\Delta^{\operatorname{op}}\to C S : Δ op → C (that is, a simplicial object in C C C ) with a morphism i : Y → ∼ S ( [ 0 ] ) i:Y\xrightarrow{\sim}S([0]) i : Y ∼ S ([ 0 ]) in W W W such that S ( f ) ∈ W S(f)\in W S ( f ) ∈ W for all morphisms f f f in Δ op \Delta^{\operatorname{op}} Δ op , and S S S is a fibrant replacement for the constant functor Δ op → { Y } ⊂ C \Delta^{\operatorname{op}}\to\{Y\}\subset C Δ op → { Y } ⊂ C in the Reedy model structure on Fun ( Δ op , C ) \operatorname{Fun}(\Delta^{\operatorname{op}},C) Fun ( Δ op , C ) .
Let X X X be another object in C C C , and j : X ′ → X j:X'\to X j : X ′ → X a cofibrant replacement. Then as in [42, §17.2], [44, Th. 5.4.9], the mapping space Hom C [ W − 1 ] ∞ L ( X , Y ) \operatorname{Hom}^{L}_{C[W^{-1}]_{\infty}}(X,Y) Hom C [ W − 1 ] ∞ L ( X , Y ) is homotopy equivalent to the simplicial set Hom C ( X ′ , S ( − ) ) : Δ op → Set \operatorname{Hom}_C(X',S(-)):\Delta^{\operatorname{op}}\to\operatorname{Set} Hom C ( X ′ , S ( − )) : Δ op → Set , which is a Kan complex by [42, Cor. 16.5.3(2)]. This Kan complex is called the right homotopy function complex in [42, §17.2].
(d) Write C c f ⊂ C C_{\mathrm{cf}}\subset C C cf ⊂ C for the full subcategory of cofibrant objects, and W c f W_{\mathrm{cf}} W cf for the class of weak equivalences between cofibrant objects. Then the inclusion C c f ↪ C C_{\mathrm{cf}}\hookrightarrow C C cf ↪ C induces an ∞ \infty ∞ -functor C c f [ W c f − 1 ] ∞ → C [ W − 1 ] ∞ C_{\mathrm{cf}}[W_{\mathrm{cf}}^{-1}]_{\infty}\to C[W^{-1}]_{\infty} C cf [ W cf − 1 ] ∞ → C [ W − 1 ] ∞ . Cisinski [17, Th. 7.5.18] implies that this is an equivalence of ∞ \infty ∞ -categories.
Thus for some purposes, it is helpful to restrict to the subcategory C c f C_{\mathrm{cf}} C cf , which makes cofibrant replacements X ′ X' X ′ above unnecessary. In many interesting cases, all objects are fibrant, and then fibrant replacements are also unnecessary.
The next definition and theorem generalize Remark 2.35 (c) to a case when C C C need not be a model category. We will prove the theorem using the ‘Derived Mapping Space Lemma’ of Arakawa–Carranza–Kapulkin [1], which is based on a theorem on ‘left calculi of fractions’ in Cisinski [17, Def. 7.2.6 & Th. 7.2.8].
To compare Definition 2.36 and Theorem 2.37 to the model category case, we should imagine C C C , W W W below to be C c f C_{\mathrm{cf}} C cf , W c f W_{\mathrm{cf}} W cf in Remark 2.35 (d).
Definition 2.36. Let ( C , W ) (C,W) ( C , W ) be a category with weak equivalences, as in Definition 2.31. Suppose S : Δ op → C S:\Delta^{\operatorname{op}}\to C S : Δ op → C is a functor, that is, S S S is a simplicial object in C C C . Then there is a natural functor F S : C op → SSet F_S:C^{\operatorname{op}}\to\operatorname{SSet} F S : C op → SSet mapping X ↦ F S ( X ) = Hom C ( X , S ( − ) ) : Δ op → Set X\mapsto F_S(X)=\operatorname{Hom}_C(X,S(-)):\Delta^{\operatorname{op}}\to\operatorname{Set} X ↦ F S ( X ) = Hom C ( X , S ( − )) : Δ op → Set for objects X X X in C C C , and mapping γ ↦ − ∘ γ : F S ( X 2 ) → F S ( X 1 ) \gamma\mapsto-\circ\gamma:F_S(X_2)\to F_S(X_1) γ ↦ − ∘ γ : F S ( X 2 ) → F S ( X 1 ) for morphisms γ : X 1 → X 2 \gamma:X_1\to X_2 γ : X 1 → X 2 in C C C .
Let Y Y Y be an object of C C C . A simplicial resolution ( S , i ) (S,i) ( S , i ) of Y Y Y in ( C , W ) (C,W) ( C , W ) is a functor S : Δ op → C S:\Delta^{\operatorname{op}}\to C S : Δ op → C and a morphism i : Y → S ( [ 0 ] ) i:Y\to S([0]) i : Y → S ([ 0 ]) in W W W , satisfying
If f : [ N ] → [ M ] f:[N]\to[M] f : [ N ] → [ M ] is a morphism in Δ \Delta Δ then S ( f ) ∈ W S(f)\in W S ( f ) ∈ W .
F S ( γ ) : F S ( X 2 ) → F S ( X 1 ) F_S(\gamma):F_S(X_2)\to F_S(X_1) F S ( γ ) : F S ( X 2 ) → F S ( X 1 ) is a weak homotopy equivalence of simplicial sets for all morphisms γ : X 1 → X 2 \gamma:X_1\to X_2 γ : X 1 → X 2 in W W W .
Using the notation of Example 2.11 and Definition 2.28 we have an ∞ \infty ∞ -functor
Y N ( C ) ∘ N ( S ) : N ( Δ op ) ⟶ PSh ( N ( C ) ) . \mathscr{Y}_{\mathcal{N}(C)}\circ\mathcal{N}(S):\mathcal{N}(\Delta^{\operatorname{op}})\longrightarrow\operatorname{PSh}(\mathcal{N}(C)). Y N ( C ) ∘ N ( S ) : N ( Δ op ) ⟶ PSh ( N ( C )) .
Since PSh ( N ( C ) ) \operatorname{PSh}(\mathcal{N}(C)) PSh ( N ( C )) is cocomplete, we may form the ∞ \infty ∞ -categorical colimit
Γ C , S : = colim N ( Δ op ) Y N ( C ) ∘ N ( S ) ∈ PSh ( N ( C ) ) . (49) \Gamma_{C,S}:=\underset{\mathcal{N}(\Delta^{\operatorname{op}})}{\operatorname{colim}}\mathscr{Y}_{\mathcal{N}(C)}\circ\mathcal{N}(S)\in\operatorname{PSh}(\mathcal{N}(C)). \tag*{(49)} Γ C , S := N ( Δ op ) colim Y N ( C ) ∘ N ( S ) ∈ PSh ( N ( C )) . (49)
For colimits in ∞ \infty ∞ -categories see Lurie [60, §4.2], [63, §7] and Land [55, §4.3]. Here Γ C , S \Gamma_{C,S} Γ C , S is an ∞ \infty ∞ -functor N ( C ) op → S p c \mathcal{N}(C)^{\operatorname{op}} \to\mathcal{S}pc N ( C ) op → S p c . Since colimits in PSh ( N ( C ) ) \operatorname{PSh}(\mathcal{N}(C)) PSh ( N ( C )) can be computed objectwise, for each X ∈ C X \in C X ∈ C we have an equivalence in S p c \mathcal{S}pc S p c
Γ C , S ( X ) ≃ colim N ( Δ op ) Hom C ( X , N ( S ) ) . (50) \Gamma_{C,S}(X) \simeq\operatorname{colim}_{\mathcal{N}(\Delta^{\operatorname{op}})} \operatorname{Hom}_{C}(X,\mathcal{N}(S)). \tag*{(50)} Γ C , S ( X ) ≃ colim N ( Δ op ) Hom C ( X , N ( S )) . (50)
Now the functor Hom C ( X , N ( S ) ) \operatorname{Hom}_{C}(X,\mathcal{N}(S)) Hom C ( X , N ( S )) in (50) is equivalent to the composition
N ( Δ op ) → N ( F S ( X ) ) N ( Set ) → N ( inc ) N ( SSet ) → Π loc ∞ SSet [ W E − 1 ] ∞ → ≃ ( 2.16 ) S p c , \mathcal{N}(\Delta^{\operatorname{op}}) \xrightarrow{\mathcal{N}(F_{S}(X))} \mathcal{N}(\operatorname{Set}) \xrightarrow{\mathcal{N}(\operatorname{inc})} \mathcal{N}(\operatorname{SSet}) \xrightarrow{\Pi^{\infty}_{\operatorname{loc}}} \operatorname{SSet}[\mathit{WE}^{-1}]_{\infty} \xrightarrow[\simeq]{(2.16)} \mathcal{S}pc, N ( Δ op ) N ( F S ( X )) N ( Set ) N ( inc ) N ( SSet ) Π loc ∞ SSet [ WE − 1 ] ∞ ( 2.16 ) ≃ S p c ,
where inc : Set ↪ SSet \operatorname{inc}:\operatorname{Set}\hookrightarrow\operatorname{SSet} inc : Set ↪ SSet maps a set S S S to the constant simplicial set Δ op → S ∈ Set \Delta^{\operatorname{op}} \to S \in\operatorname{Set} Δ op → S ∈ Set . It follows from Lurie [60, Ex. A.2.9.31] that the ∞ \infty ∞ -categorical colimit of this composition is the simplicial set F S ( X ) F_{S}(X) F S ( X ) (up to weak homotopy equivalence), considered as an object of SSet [ W E − 1 ] ∞ \operatorname{SSet}[\mathit{WE}^{-1}]_{\infty} SSet [ WE − 1 ] ∞ . That is, we have
Γ C , S ( X ) ≃ F S ( X ) . (51) \Gamma_{C,S}(X) \simeq F_{S}(X). \tag*{(51)} Γ C , S ( X ) ≃ F S ( X ) . (51)
Similarly, if γ : X 1 → X 2 \gamma:X_{1}\to X_{2} γ : X 1 → X 2 is a morphism in C C C then Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) \Gamma_{C,S}(f):\Gamma_{C,S}(X_{2})\to\Gamma_{C,S}(X_{1}) Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) is F S ( γ ) : F S ( X 2 ) → F S ( X 1 ) F_{S}(\gamma):F_{S}(X_{2})\to F_{S}(X_{1}) F S ( γ ) : F S ( X 2 ) → F S ( X 1 ) , modulo weak homotopy equivalences of F S ( X 1 ) F_{S}(X_{1}) F S ( X 1 ) and F S ( X 2 ) F_{S}(X_{2}) F S ( X 2 ) . In particular, if γ : X 1 → X 2 \gamma:X_{1}\to X_{2} γ : X 1 → X 2 lies in W W W then Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) \Gamma_{C,S}(f):\Gamma_{C,S}(X_{2})\to\Gamma_{C,S}(X_{1}) Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) is a weak homotopy equivalence by (ii) above, that is, Γ C , S ( f ) \Gamma_{C,S}(f) Γ C , S ( f ) is an equivalence in S p c \mathcal{S}pc S p c .
The rest of the definition follows Arakawa–Carranza–Kapulkin [1, Def. 2.1] in our special case. Let [ n ] [n] [ n ] be an object in Δ \Delta Δ , and π n : [ n ] → [ 0 ] \pi_{n}:[n]\to[0] π n : [ n ] → [ 0 ] the unique morphism. Write Φ [ n ] \Phi_{[n]} Φ [ n ] for the composition of morphisms in PSh ( N ( C ) ) \operatorname{PSh}(\mathcal{N}(C)) PSh ( N ( C ))
Y N ( C ) ( Y ) → Y N ( C ) ( i ∘ S ( π n ) ) Y N ( C ) ( S ( [ n ] ) ) → colim [ n ] Γ C , S , \mathcal{Y}_{\mathcal{N}(C)}(Y) \xrightarrow{\mathcal{Y}_{\mathcal{N}(C)}(i\circ S(\pi_{n}))} \mathcal{Y}_{\mathcal{N}(C)}(S([n])) \xrightarrow{\operatorname{colim}_{[n]}} \Gamma_{C,S}, Y N ( C ) ( Y ) Y N ( C ) ( i ∘ S ( π n )) Y N ( C ) ( S ([ n ])) colim [ n ] Γ C , S ,
where colim [ n ] \operatorname{colim}_{[n]} colim [ n ] is the morphism to the colimit in (49) , which is natural up to homotopy. Then Φ [ n ] \Phi_{[n]} Φ [ n ] is independent of [ n ] [n] [ n ] up to homotopy, by properties of colimits.
Using the ∞ \infty ∞ -functor Π loc ∞ : N ( C ) → C [ W − 1 ] ∞ \Pi^{\infty}_{\operatorname{loc}}:\mathcal{N}(C)\to C[W^{-1}]_{\infty} Π loc ∞ : N ( C ) → C [ W − 1 ] ∞ , from Definition 2.27 we have ∞ \infty ∞ -functors ( Π loc ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) → PSh ( N ( C ) ) (\Pi^{\infty}_{\operatorname{loc}})^{*}:\operatorname{PSh}(C[W^{-1}]_{\infty})\to\operatorname{PSh}(\mathcal{N}(C)) ( Π loc ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) → PSh ( N ( C )) and its left adjoint ( Π loc ∞ ) ! : PSh ( N ( C ) ) → PSh ( C [ W − 1 ] ∞ ) (\Pi^{\infty}_{\operatorname{loc}})_{!}:\operatorname{PSh}(\mathcal{N}(C))\to\operatorname{PSh}(C[W^{-1}]_{\infty}) ( Π loc ∞ ) ! : PSh ( N ( C )) → PSh ( C [ W − 1 ] ∞ ) . Thus we have a 2-commutative diagram
Y N ( C ) ( Y ) → Φ [ n ] Γ C , S ↓ η ↓ [ ≃ ] η ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Y N ( C ) ( Y ) ) → ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Φ [ n ] ) ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Γ C , S ) , (52) \begin{aligned}
\mathcal{Y}_{\mathcal{N}(C)}(Y) &\xrightarrow{\Phi_{[n]}} \Gamma_{C,S} \\
\downarrow\eta&& \downarrow[\simeq]{\eta} \\
(\Pi^{\infty}_{\operatorname{loc}})^{*}\circ(\Pi^{\infty}_{\operatorname{loc}})_{!}(\mathcal{Y}_{\mathcal{N}(C)}(Y)) &\xrightarrow{(\Pi^{\infty}_{\operatorname{loc}})^{*}\circ(\Pi^{\infty}_{\operatorname{loc}})_{!}(\Phi_{[n]})} (\Pi^{\infty}_{\operatorname{loc}})^{*}\circ(\Pi^{\infty}_{\operatorname{loc}})_{!}(\Gamma_{C,S}), \tag*{(52)}
\end{aligned} Y N ( C ) ( Y ) ↓ η ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Y N ( C ) ( Y )) Φ [ n ] Γ C , S ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Φ [ n ] ) ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! ( Γ C , S ) , ↓ [ ≃ ] η (52)
where η : Id ⇒ ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! \eta:\operatorname{Id}\Rightarrow(\Pi^{\infty}_{\operatorname{loc}})^{*}\circ(\Pi^{\infty}_{\operatorname{loc}})_{!} η : Id ⇒ ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ ) ! is the unit of the adjunction.
We showed above that if γ : X 1 → X 2 \gamma:X_{1}\to X_{2} γ : X 1 → X 2 lies in W W W then Γ C , S ( f ) \Gamma_{C,S}(f) Γ C , S ( f ) is an equivalence in S p c \mathcal{S}pc S p c . From Definition 2.31, this implies that Γ C , S ( f ) \Gamma_{C,S}(f) Γ C , S ( f ) lies in the essential image of ( Π loc ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) → PSh ( N ( C ) ) (\Pi^{\infty}_{\operatorname{loc}})^{*}:\operatorname{PSh}(C[W^{-1}]_{\infty})\to\operatorname{PSh}(\mathcal{N}(C)) ( Π loc ∞ ) ∗ : PSh ( C [ W − 1 ] ∞ ) → PSh ( N ( C )) , which is full and faithful. So by properties of adjoint functors we see that the right hand column of (52) is an equivalence in $\operatorname{PSh}(\mathcal{N}(C))`.
Applying equation (2.13) to F = Π l o c ∞ F = \Pi^\infty_{\mathrm{loc}} F = Π loc ∞ , and noting that Π l o c ∞ ( Y ) = Y \Pi^\infty_{\mathrm{loc}}(Y) = Y Π loc ∞ ( Y ) = Y as N ( C ) \mathcal{N}(\mathcal{C}) N ( C ) and C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]_\infty C [ W − 1 ] ∞ have the same objects, gives an equivalence in PSh ( C [ W − 1 ] ∞ ) \operatorname{PSh}(\mathcal{C}[\mathcal{W}^{-1}]_\infty) PSh ( C [ W − 1 ] ∞ )
Y C [ W − 1 ] ∞ ( Y ) ≃ ( Π l o c ∞ ) ! ( Y N ( C ) ( Y ) ) . (53) \mathcal{Y}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(Y) \simeq(\Pi^\infty_{\mathrm{loc}})!(\mathcal{Y}_{\mathcal{N}(\mathcal{C})}(Y)). \tag*{(53)} Y C [ W − 1 ] ∞ ( Y ) ≃ ( Π loc ∞ )! ( Y N ( C ) ( Y )) . (53)
Combining (2.21)–(2.22) and the fact that the right hand column of (2.21) is an equivalence, we see that there exists a morphism Ψ \Psi Ψ unique up to equivalence which makes the following commute up to equivalence:
( Π l o c ∞ ) ∗ ∘ Y C [ W − 1 ] ∞ ( Y ) → Ψ Γ C , S ≃ ( Π l o c ∞ ) ∗ ( 2.22 ) ↓ ↓ ≃ η ( Π l o c ∞ ) ∗ ∘ ( Π l o c ∞ ) ! ( Y N ( C ) ( Y ) ) → ( Π l o c ∞ ) ∗ ∘ ( Π l o c ∞ ) ! ( Φ [ n ] ) ( Π l o c ∞ ) ∗ ∘ ( Π l o c ∞ ) ! ( Γ C , S ) . (54) \begin{CD}
(\Pi^\infty_{\mathrm{loc}})^* \circ\mathcal{Y}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(Y) @>{\Psi}>> \Gamma_{\mathcal{C},S} \\
@V{\simeq\;(\Pi^\infty_{\mathrm{loc}})^*(2.22)}VV @VV{\simeq\;\eta}V \\
(\Pi^\infty_{\mathrm{loc}})^* \circ(\Pi^\infty_{\mathrm{loc}})! (\mathcal{Y}_{\mathcal{N}(\mathcal{C})}(Y)) @>{(\Pi^\infty_{\mathrm{loc}})^* \circ(\Pi^\infty_{\mathrm{loc}})! (\Phi_{[n]})}>> (\Pi^\infty_{\mathrm{loc}})^* \circ(\Pi^\infty_{\mathrm{loc}})! (\Gamma_{\mathcal{C},S}). \tag*{(54)}
\end{CD} ( Π loc ∞ ) ∗ ∘ Y C [ W − 1 ] ∞ ( Y ) ≃ ( Π loc ∞ ) ∗ ( 2.22 ) ↓ ⏐ ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ )! ( Y N ( C ) ( Y )) Ψ ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ )! ( Φ [ n ] ) Γ C , S ↓ ⏐ ≃ η ( Π loc ∞ ) ∗ ∘ ( Π loc ∞ )! ( Γ C , S ) . (54)
Evaluating Ψ \Psi Ψ at an object X X X in C \mathcal{C} C and noting (2.21) and
( ( Π l o c ∞ ) ∗ ∘ Y C [ W − 1 ] ∞ ( Y ) ) ( X ) ≃ ( Y C [ W − 1 ] ∞ ( Y ) ) ( X ) = Hom C [ W − 1 ] ∞ ( X , Y ) ≃ Hom C [ W − 1 ] ∞ L ( X , Y ) , ((\Pi^\infty_{\mathrm{loc}})^* \circ\mathcal{Y}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(Y))(X) \simeq(\mathcal{Y}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(Y))(X) = \operatorname{Hom}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(X,Y)
\simeq\operatorname{Hom}^{L}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(X,Y), (( Π loc ∞ ) ∗ ∘ Y C [ W − 1 ] ∞ ( Y )) ( X ) ≃ ( Y C [ W − 1 ] ∞ ( Y )) ( X ) = Hom C [ W − 1 ] ∞ ( X , Y ) ≃ Hom C [ W − 1 ] ∞ L ( X , Y ) ,
gives a morphism in S p c \mathcal{S}_{\mathrm{pc}} S pc
Ψ ( X ) : Hom C [ W − 1 ] ∞ L ( X , Y ) ⟶ F S ( X ) . (55) \Psi(X) : \operatorname{Hom}^{L}_{\mathcal{C}[\mathcal{W}^{-1}]_\infty}(X,Y) \longrightarrow F_S(X). \tag*{(55)} Ψ ( X ) : Hom C [ W − 1 ] ∞ L ( X , Y ) ⟶ F S ( X ) . (55)
Theorem 2.37. In Definition 2.36, the morphism Ψ \Psi Ψ in (2.23) is an equivalence in PSh ( N ( C ) ) \operatorname{PSh}(\mathcal{N}(\mathcal{C})) PSh ( N ( C )) , so that Ψ ( X ) \Psi(X) Ψ ( X ) in (2.24) is an equivalence in S p c \mathcal{S}_{\mathrm{pc}} S pc for all X ∈ C X \in\mathcal{C} X ∈ C .
Proof. We apply Arakawa–Carranza–Kapulkin [1, Def. 2.1 & Th. 2.2] in a special case, which is similar to their proof of Remark 2.35(c) above in [1, Cor. 3.1].
They begin with an ∞ \infty ∞ -category C \mathcal{C} C with a suitable ∞ \infty ∞ -subcategory W ⊂ C \mathcal{W} \subset\mathcal{C} W ⊂ C to be localized, an object Y ∈ C Y \in\mathcal{C} Y ∈ C , a small ∞ \infty ∞ -category I \mathcal{I} I , and an ∞ \infty ∞ -functor Y ∙ : I → C ↓ Y Y_\bullet: \mathcal{I} \to\mathcal{C}_{\downarrow Y} Y ∙ : I → C ↓ Y to the undercategory of Y Y Y in C \mathcal{C} C . We take C = N ( C ) \mathcal{C} = \mathcal{N}(\mathcal{C}) C = N ( C ) , W = N ( W ) \mathcal{W} = \mathcal{N}(\mathcal{W}) W = N ( W ) , I = N ( Δ o p ) \mathcal{I} = \mathcal{N}(\Delta^{\mathrm{op}}) I = N ( Δ op ) , and Y ∙ Y_\bullet Y ∙ to be the lift of N ( S ) : N ( Δ o p ) → N ( C ) \mathcal{N}(S) : \mathcal{N}(\Delta^{\mathrm{op}}) \to\mathcal{N}(\mathcal{C}) N ( S ) : N ( Δ op ) → N ( C ) to N ( C ) ↓ Y \mathcal{N}(\mathcal{C}) \downarrow Y N ( C ) ↓ Y using the morphisms i ∘ S ( π ) : Y → S ( [ n ] ) i \circ S(\pi) : Y \to S([n]) i ∘ S ( π ) : Y → S ([ n ]) for all [ n ] [n] [ n ] in Δ o p \Delta^{\mathrm{op}} Δ op .
They impose conditions [1, Def. 2.1(R1)–(R3)] for S S S to be an ‘I \mathcal{I} I -resolution’ of Y Y Y . Here (R1) is automatic as Δ o p \Delta^{\mathrm{op}} Δ op has an initial object, (R2) follows from Definition 2.36(i), and (R3) follows from Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) \Gamma_{\mathcal{C},S}(f) : \Gamma_{\mathcal{C},S}(X_2) \to\Gamma_{\mathcal{C},S}(X_1) Γ C , S ( f ) : Γ C , S ( X 2 ) → Γ C , S ( X 1 ) an equivalence in S p c \mathcal{S}_{\mathrm{pc}} S pc if γ : X 1 → X 2 \gamma: X_1 \to X_2 γ : X 1 → X 2 is a morphism in W \mathcal{W} W , as proved in Definition 2.36. Thus [1, Th. 2.2] says that Ψ \Psi Ψ in (2.23) is an equivalence.
Stable ∞ \infty ∞ -categories
Stable ∞ \infty ∞ -categories were introduced by Lurie [59], [61, §1], see Groth [39, §5.1].
Definition 2.38. Let C \mathcal{C} C be an ∞ \infty ∞ -category. A zero object in C \mathcal{C} C is an object 0 ∈ C 0 \in\mathcal{C} 0 ∈ C which is both initial and terminal (in the ∞ \infty ∞ -categorical sense).
Suppose C \mathcal{C} C has a zero object. A triangle in C \mathcal{C} C is a 2-commutative diagram
X → f Y ↓ ↓ g 0 → Z (56) \begin{CD}
X @>{f}>> Y \\
@VVV @VV{g}V \\
0 @>>> Z \tag*{(56)}
\end{CD} X ↓ ⏐ 0 f Y ↓ ⏐ g Z (56)
where X , Y , Z X,Y,Z X , Y , Z are objects in C \mathcal{C} C and f , g f,g f , g are morphisms. We call (2.25) an exact triangle if it is Cartesian (in the ∞ \infty ∞ -categorical sense), and then we say that X X X is the kernel of g g g . We call (2.25) a coexact triangle if it is co-Cartesian, and then we say that Z Z Z is the cokernel of f f f .
We call C \mathcal{C} C stable if every morphism in C \mathcal{C} C has a kernel and a cokernel, and a triangle (2.25) is exact if and only if it is coexact. Write C a t ∞ s t ⊂ C a t ∞ \mathsf{Cat}_{\infty}^{\mathrm{st}} \subset\mathsf{Cat}_{\infty} Cat ∞ st ⊂ Cat ∞ for the full ∞ \infty ∞ -subcategory with objects stable ∞ \infty ∞ -categories.
If C \mathcal{C} C is a stable ∞ \infty ∞ -category then its homotopy category Ho ( C ) \operatorname{Ho}(\mathcal{C}) Ho ( C ) has the structure of a triangulated category, as in Gelfand–Manin [35]. Thus stable ∞ \infty ∞ -categories are an ∞ \infty ∞ -categorical version of triangulated categories. Note however that to be triangulated is an extra structure on a category T \mathcal{T} T , but to be stable is a property of an ∞ \infty ∞ -category C \mathcal{C} C .
It has been apparent since the introduction of triangulated categories by Grothendieck and Verdier that there were problems with the theory – for example, the ‘nonfunctoriality of the cone’ in triangulated categories, as in Gelfand and Manin [35, §IV.1.7] – and several attempts have been made to fix these problems by somehow enhancing the triangulated category structure.
There is a strong case for believing that stable ∞ \infty ∞ -categories are the ‘true’ theory to which triangulated categories are an imperfect approximation. Stable ∞ \infty ∞ -categories have many wonderful properties.
∞ \infty ∞ -bicategories
Informally, an ( ∞ , n ) (\infty,n) ( ∞ , n ) -category is some kind of higher category with a notion of k k k -morphism for all k = 1 , 2 , … k=1,2,\ldots k = 1 , 2 , … , such that all k k k -morphisms are invertible for k > n k>n k > n . This is not a precise definition, and there are many different models for ( ∞ , n ) (\infty,n) ( ∞ , n ) -categories. Our ∞ \infty ∞ -categories in §2.2–§2.7 are a kind of ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category, and Kan complexes are a kind of ( ∞ , 0 ) (\infty,0) ( ∞ , 0 ) -category.
We will explain ∞ \infty ∞ -bicategories, which are a particularly nice model for ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories, following Lurie [62, §4] and Gagna–Harpaz–Lanari [34]. A slightly different notion appears in Lurie [63, §5.4]. There it is called ∞ \infty ∞ -bicategory, conflicting with the notation of [62], but we will call them strong ∞ \infty ∞ -bicategories. The next definition follows Lurie [62, §3.1].
Definition 2.39. A scaled simplicial set is a pair ( S , T ) (S,T) ( S , T ) where S S S is a simplicial set, and T ⊆ S ( [ 2 ] ) T \subseteq S([2]) T ⊆ S ([ 2 ]) is a subset of the 2-simplices of S S S which contains every degenerate 2-simplex in S S S . A 2-simplex t : Δ 2 → S t:\Delta^2 \to S t : Δ 2 → S is called thin if it lies in T T T .
If ( S , T ) (S,T) ( S , T ) , ( S ′ , T ′ ) (S',T') ( S ′ , T ′ ) are scaled simplicial sets, a morphism f : ( S , T ) → ( S ′ , T ′ ) f:(S,T)\to(S',T') f : ( S , T ) → ( S ′ , T ′ ) is a morphism of simplicial sets f : S → S ′ f:S\to S' f : S → S ′ such that f f f maps 2-simplices in T T T to 2-simplices in T ′ T' T ′ . Then scaled simplicial sets form a category S S e t s c \mathsf{SSet}^{\mathrm{sc}} SSet sc .
Lurie [62, Th. 4.2.7] defines a combinatorial model category structure on S S e t s c \mathsf{SSet}^{\mathrm{sc}} SSet sc , in which every object is cofibrant. In [62, Th. 0.0.3] he shows that the model category S S e t s c \mathsf{SSet}^{\mathrm{sc}} SSet sc is part of a diagram of Quillen equivalences with other model categories that are agreed to be models for ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories (see Gagna–Harpaz–Lanari [34, Fig. 1] for a larger diagram).
Lurie [62, Def. 4.1.1] defines a class of scaled simplicial sets satisfying horn-filling conditions that he calls weak ∞ \infty ∞ -bicategories. In [62, Def. 4.2.8] he defines a scaled simplicial set to be an ∞ \infty ∞ -bicategory if it is a fibrant object in his model structure. As in [34, Prop. 1.30], one can show using [62, Prop. 3.1.13] that every ∞ \infty ∞ -bicategory is a weak ∞ \infty ∞ -bicategory. Gagna–Harpaz–Lanari [34, Th. 5.1] prove the converse, that every weak ∞ \infty ∞ -bicategory is an ∞ \infty ∞ -bicategory. The next definition follows [62, Def. 4.1.1], but using the name ∞ \infty ∞ -bicategory.
Definition 2.40. A scaled simplicial set ( S , T ) (S,T) ( S , T ) is called an ∞ \infty ∞ -bicategory if:
(a) For all N ≥ 3 N \ge3 N ≥ 3 and 0 < i < N 0 < i < N 0 < i < N , a commutative diagram of morphisms → \to → in which t t t is a thin 2-simplex, as follows:
Λ i N ← f Δ 2 i n c ↓ ↓ t Δ N → f ~ S (57) \begin{CD}
\Lambda_i^N @<{f}<< \Delta^2 \\
@V{\mathrm{inc}}VV @VV{t}V \\
\Delta^N @>{\widetilde{f}}>> S \tag*{(57)}
\end{CD} Λ i N inc ↓ ⏐ Δ N f f Δ 2 ↓ ⏐ t S (57) can be completed with a morphism f ~ \widetilde{f} f as shown. Here i n c : Λ i N ↪ Δ N \mathrm{inc}:\Lambda_i^N \hookrightarrow\Delta^N inc : Λ i N ↪ Δ N is as in Examples 2.3–2.4, and Δ { i − 1 , i , i + 1 } : Δ 2 → Δ N \Delta^{\{i-1,i,i+1\}}:\Delta^2 \to\Delta^N Δ { i − 1 , i , i + 1 } : Δ 2 → Δ N is the 2-simplex in Δ N \Delta^N Δ N with vertices i − 1 , i , i + 1 i-1,i,i+1 i − 1 , i , i + 1 . This lies in Λ i N ⊂ Δ N \Lambda_i^N \subset\Delta^N Λ i N ⊂ Δ N as N ≥ 3 N \ge3 N ≥ 3 .
(b) Every commutative diagram of morphisms → \to →
Λ i 2 → f S i n c ↓ Δ 2 → f ~ S (58) \begin{CD}
\Lambda_i^2 @>{f}>> S \\
@V{\mathrm{inc}}VV @. \\
\Delta^2 @>{\widetilde{f}}>> S \tag*{(58)}
\end{CD} Λ i 2 inc ↓ ⏐ Δ 2 f f S S (58) can be completed with a morphism f ~ \widetilde{f} f as shown, with f ~ \widetilde{f} f a thin 2-simplex.
(c) Suppose f : Δ 4 → S f:\Delta^4 \to S f : Δ 4 → S is a morphism and f ∘ Δ { i , j , k } : Δ 2 → S f\circ\Delta^{\{i,j,k\}}:\Delta^2\to S f ∘ Δ { i , j , k } : Δ 2 → S is thin for { i , j , k } = { 0 , 2 , 4 } , { 1 , 2 , 3 } , { 0 , 1 , 3 } , { 1 , 3 , 4 } , { 0 , 1 , 2 } \{i,j,k\}=\{0,2,4\},\{1,2,3\},\{0,1,3\},\{1,3,4\},\{0,1,2\} { i , j , k } = { 0 , 2 , 4 } , { 1 , 2 , 3 } , { 0 , 1 , 3 } , { 1 , 3 , 4 } , { 0 , 1 , 2 } . Then f ∘ Δ { i , j , k } : Δ 2 → S f\circ\Delta^{\{i,j,k\}}:\Delta^2\to S f ∘ Δ { i , j , k } : Δ 2 → S is also thin for { i , j , k } = { 0 , 3 , 4 } , { 0 , 1 , 4 } \{i,j,k\}=\{0,3,4\},\{0,1,4\} { i , j , k } = { 0 , 3 , 4 } , { 0 , 1 , 4 } .
(d) For N ≥ 3 N\ge3 N ≥ 3 , every commutative diagram of morphisms → \to →
Λ 0 N → Δ { 0 , 1 , N } Δ 2 → s 1 , 0 Δ 1 i n c ↓ ↓ g Δ N → f ~ S (59) \begin{CD}
\Lambda_0^N @>{\Delta^{\{0,1,N\}}}>> \Delta^2 @>{s_{1,0}}>> \Delta^1 \\
@V{\mathrm{inc}}VV @. @VV{g}V \\
\Delta^N @>{\widetilde{f}}>> S \tag*{(59)}
\end{CD} Λ 0 N inc ↓ ⏐ Δ N Δ { 0 , 1 , N } f Δ 2 S s 1 , 0 Δ 1 ↓ ⏐ g (59) can be completed with a morphism f ~ \widetilde{f} f as shown. Here s 1 , 0 : Δ 2 → Δ 1 s_{1,0}:\Delta^2\to\Delta^1 s 1 , 0 : Δ 2 → Δ 1 is the degeneracy map acting on vertices by 0 ↦ 0 0\mapsto0 0 ↦ 0 , 1 ↦ 0 1\mapsto0 1 ↦ 0 , 2 ↦ 1 2\mapsto1 2 ↦ 1 .
This is an outer horn-filling condition : it says that an outer horn f : Λ 0 N → S f:\Lambda_0^N\to S f : Λ 0 N → S can be filled provided the 2-simplex f ∘ Δ { 0 , 1 , N } : Δ 2 → S f\circ\Delta^{\{0,1,N\}}:\Delta^2\to S f ∘ Δ { 0 , 1 , N } : Δ 2 → S is left-degenerate (factors via s 1 , 0 s_{1,0} s 1 , 0 ).
We can think of ∞ \infty ∞ -bicategory ( S , T ) (S,T) ( S , T ) as a kind of category in a similar way to Definition 2.12 . An object X X X of ( S , T ) (S,T) ( S , T ) is a 0 0 0 -simplex in S S S , or equivalently, a map of simplicial sets X : Δ 0 → S X:\Delta^0 \to S X : Δ 0 → S . A morphism , or 1 1 1 -morphism, f : X → Y f:X\to Y f : X → Y of ( S , T ) (S,T) ( S , T ) is a 1 1 1 -simplex f f f in S S S with f ∣ Δ { 0 } = X f|_{\Delta^{\{0\}}}=X f ∣ Δ { 0 } = X and f ∣ Δ { 1 } = Y f|_{\Delta^{\{1\}}}=Y f ∣ Δ { 1 } = Y . If f : X → Y f:X\to Y f : X → Y and g : Y → Z g:Y\to Z g : Y → Z are 1 1 1 -morphisms, to define the ‘composition’ g ∘ f : X → Z g\circ f:X\to Z g ∘ f : X → Z we choose a thin 2 2 2 -simplex t : Δ 2 → S t:\Delta^2\to S t : Δ 2 → S with t ∘ Δ { 0 , 1 } = f t\circ\Delta^{\{0,1\}}=f t ∘ Δ { 0 , 1 } = f and t ∘ Δ { 1 , 2 } = g t\circ\Delta^{\{1,2\}}=g t ∘ Δ { 1 , 2 } = g , which exists by (b). Then we set g ∘ f = t ∘ Δ { 0 , 2 } g\circ f=t\circ\Delta^{\{0,2\}} g ∘ f = t ∘ Δ { 0 , 2 } . As in ∞ \infty ∞ -categories, composition is not uniquely defined, as t t t need not be unique.
Let η : Δ 2 → S \eta:\Delta^2\to S η : Δ 2 → S be a morphism, and set X = η ∘ Δ { 0 } X=\eta\circ\Delta^{\{0\}} X = η ∘ Δ { 0 } , Y = η ∘ Δ { 1 } Y=\eta\circ\Delta^{\{1\}} Y = η ∘ Δ { 1 } , Z = η ∘ Δ { 2 } Z=\eta\circ\Delta^{\{2\}} Z = η ∘ Δ { 2 } , f = η ∘ Δ { 0 , 1 } f=\eta\circ\Delta^{\{0,1\}} f = η ∘ Δ { 0 , 1 } , g = η ∘ Δ { 1 , 2 } g=\eta\circ\Delta^{\{1,2\}} g = η ∘ Δ { 1 , 2 } , h = η ∘ Δ { 0 , 2 } h=\eta\circ\Delta^{\{0,2\}} h = η ∘ Δ { 0 , 2 } , so that f : X → Y f:X\to Y f : X → Y , g : Y → Z g:Y\to Z g : Y → Z , h : X → X h:X\to X h : X → X are 1 1 1 -morphisms. We think of η \eta η as giving a 2 2 2 -morphism g ∘ f ⇒ h g\circ f\Rightarrow h g ∘ f ⇒ h in ( S , T ) (S,T) ( S , T ) , which is a 2 2 2 -isomorphism if η \eta η is thin.
Although part (d) above is not symmetric under taking opposite simplicial sets S ↦ S op S\mapsto S^{\operatorname{op}} S ↦ S op , Gagna–Harpaz–Lanari [34, Cor. 5.5] show that ( S , T ) (S,T) ( S , T ) is an ∞ \infty ∞ -bicategory if and only if ( S op , T ) (S^{\operatorname{op}},T) ( S op , T ) is an ∞ \infty ∞ -bicategory.
Definition 2.41. Let ( S , T ) (S,T) ( S , T ) be an ∞ \infty ∞ -bicategory. Define the core Core ( S , T ) \operatorname{Core}(S,T) Core ( S , T ) of ( S , T ) (S,T) ( S , T ) to be the simplicial subset Core ( S , T ) ⊆ S \operatorname{Core}(S,T)\subseteq S Core ( S , T ) ⊆ S such that f : Δ N → S f:\Delta^N\to S f : Δ N → S is an N N N -simplex of Core ( S , T ) \operatorname{Core}(S,T) Core ( S , T ) if and only if all compositions Δ 2 → g Δ N → f S \Delta^2\xrightarrow{g}\Delta^N\xrightarrow{f}S Δ 2 g Δ N f S are thin 2 2 2 -simplices in S S S . It follows from Definition 2.40 (a),(b) that Core ( S , T ) \operatorname{Core}(S,T) Core ( S , T ) is an ∞ \infty ∞ -category. Here our notation ‘core’ comes from Gagna–Harpaz–Lanari [34, Def. 1.22]. It is called the underlying ∞ \infty ∞ -category of ( S , T ) (S,T) ( S , T ) in Lurie [62, Rem. 4.1.3], and the pith of ( S , T ) (S,T) ( S , T ) in Lurie [63, §5.4.5].
In effect, Core ( S , T ) \operatorname{Core}(S,T) Core ( S , T ) is the ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -subcategory of ( S , T ) (S,T) ( S , T ) that only includes 2 2 2 -morphisms which are 2 2 2 -isomorphisms, and is thus an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -subcategory.
Example 2.42. Let ( S , T ) (S,T) ( S , T ) be a scaled simplicial set in which every 2 2 2 -simplex in S S S is thin. Then by Lurie [62, Rem. 4.1.2], ( S , T ) (S,T) ( S , T ) is an ∞ \infty ∞ -bicategory if and only if S S S is an ∞ \infty ∞ -category, and then Core ( S , T ) = S \operatorname{Core}(S,T)=S Core ( S , T ) = S .
Definition 2.43. An ∞ \infty ∞ -bicategory ( S , T ) (S,T) ( S , T ) is strong if T T T is the set of all 2 2 2 -simplices t : Δ 2 → S t:\Delta^2\to S t : Δ 2 → S for which Definition 2.40 (a) holds. Then T T T is determined by S S S , so we can just refer to S S S as a strong ∞ \infty ∞ -bicategory. Lurie [63, §5.4] develops a theory of ‘∞ \infty ∞ -bicategories’, which are equivalent to our strong ∞ \infty ∞ -bicategories.
If C C C is an ordinary category, Example 2.11 defined an ∞ \infty ∞ -category N ( C ) \mathcal{N}(C) N ( C ) , the nerve of C C C . In a similar way, if C C C is a weak 2 2 2 -category (bicategory), Lurie [63, §2.3] defines the Duskin nerve N D ( C ) \mathcal{N}^{D}(C) N D ( C ) , a simplicial set, and shows it is a strong ∞ \infty ∞ -bicategory [63, Prop. 5.4.1].
In Theorems 6.23 , 7.17 and 8.10 below we will construct ∞ \infty ∞ -bicategories of A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories. As in Remark 7.18(f), the author expects that these are not strong ∞ \infty ∞ -bicategories. This is our reason for choosing the ∞ \infty ∞ -bicategory theory from [62, §4] rather than from [63, §5.4].
The classical theory of A ∞ A_{\infty} A ∞ -algebras
A ∞ A_{\infty} A ∞ -algebras were introduced by Stasheff [85, 86]. Some good references are Fukaya et al. [33, §3.2–§3.3], Seidel [84], and two survey papers by Keller [51,52]. Kontsevich and Soibelman [54] give a geometric point of view on A ∞ A_{\infty} A ∞ -algebras. We follow the grading and sign conventions of Keller [51].
Sign conventions
We will use the following sign convention for tensor products of graded R R R -linear and R R R -multilinear operators throughout the book.
Convention 3.1. Suppose R R R is a commutative ring, n ≥ 1 n \ge1 n ≥ 1 , and p 1 , … , p n ≥ 0 p_1,\ldots,p_n \ge0 p 1 , … , p n ≥ 0 , and that B i , j = ⨁ d ∈ Z B i , j d B_{i,j} = \bigoplus_{d\in\mathbb{Z}} B_{i,j}^{d} B i , j = ⨁ d ∈ Z B i , j d and C i = ⨁ d ∈ Z C i d C_i = \bigoplus_{d\in\mathbb{Z}} C_i^{d} C i = ⨁ d ∈ Z C i d are Z \mathbb{Z} Z -graded R R R -modules for i = 1 , … , n i=1,\ldots,n i = 1 , … , n and j = 1 , … , p i j=1,\ldots,p_i j = 1 , … , p i , and that
g i : B i , 1 × B i , 2 × ⋯ × B i , p i ⟶ C i g_i:B_{i,1}\times B_{i,2}\times\cdots\times B_{i,p_i}\longrightarrow C_i g i : B i , 1 × B i , 2 × ⋯ × B i , p i ⟶ C i is an R R R -multilinear graded map of degree m i m_i m i for i = 1 , … , n i=1,\ldots,n i = 1 , … , n , that is, g i g_i g i maps B i , 1 d 1 × ⋯ × B i , p i d p i ⟶ C i m i + d 1 + ⋯ + d p i B_{i,1}^{d_1}\times\cdots\times B_{i,p_i}^{d_{p_i}}\longrightarrow C_i^{m_i+d_1+\cdots+d_{p_i}} B i , 1 d 1 × ⋯ × B i , p i d p i ⟶ C i m i + d 1 + ⋯ + d p i for d 1 , … , d p i ∈ Z d_1,\ldots,d_{p_i}\in\mathbb{Z} d 1 , … , d p i ∈ Z . Equivalently, we can regard g i g_i g i as an R R R -linear map graded of degree m i m_i m i
g i : B i , 1 ⊗ R B i , 2 ⊗ R ⋯ ⊗ R B i , p i ⟶ C i , g_i:B_{i,1}\otimes_R B_{i,2}\otimes_R\cdots\otimes_R B_{i,p_i}\longrightarrow C_i, g i : B i , 1 ⊗ R B i , 2 ⊗ R ⋯ ⊗ R B i , p i ⟶ C i , by identifying g i ( b i , 1 , … , b i , p i ) = g i ( b i , 1 ⊗ ⋯ ⊗ b i , p i ) g_i(b_{i,1},\ldots,b_{i,p_i})=g_i(b_{i,1}\otimes\cdots\otimes b_{i,p_i}) g i ( b i , 1 , … , b i , p i ) = g i ( b i , 1 ⊗ ⋯ ⊗ b i , p i ) . We often use the same symbol for R R R -multilinear maps, and R R R -linear maps of tensor products, in this way.
When p i = 0 p_i=0 p i = 0 , we consider g i g_i g i to be an element of C i m i C_i^{m_i} C i m i , or equivalently, an R R R -linear map R ⟶ C i m i R\longrightarrow C_i^{m_i} R ⟶ C i m i which recovers the element of C i m i C_i^{m_i} C i m i as g i ( 1 ) g_i(1) g i ( 1 ) .
Define
g 1 ⊗ ⋯ ⊗ g n : B 1 , 1 ⊗ R ⋯ ⊗ R B 1 , p 1 ⊗ R B 2 , 1 ⊗ R ⋯ ⊗ R B n , p n ⟶ C 1 ⊗ R ⋯ ⊗ R C n g_1\otimes\cdots\otimes g_n:B_{1,1}\otimes_R\cdots\otimes_R B_{1,p_1}\otimes_R B_{2,1}\otimes_R\cdots\otimes_R B_{n,p_n}\longrightarrow C_1\otimes_R\cdots\otimes_R C_n g 1 ⊗ ⋯ ⊗ g n : B 1 , 1 ⊗ R ⋯ ⊗ R B 1 , p 1 ⊗ R B 2 , 1 ⊗ R ⋯ ⊗ R B n , p n ⟶ C 1 ⊗ R ⋯ ⊗ R C n to be the R R R -linear map graded of degree m 1 + ⋯ + m n m_1+\cdots+m_n m 1 + ⋯ + m n satisfying
( g 1 ⊗ ⋯ ⊗ g n ) ( b 1 , 1 ⊗ ⋯ ⊗ b 1 , p 1 ⊗ b 2 , 1 ⊗ ⋯ ⊗ b n , 1 ⋯ ⊗ b n , p n ) = ( − 1 ) ∑ 1 ≤ i < k ≤ n , j = 1 , … , p i d i , j m k g 1 ( b 1 , 1 , … , b 1 , p 1 ) ⊗ ⋯ ⊗ g n ( b n , 1 , … , b n , p n ) (60) \begin{aligned}
&(g_1\otimes\cdots\otimes g_n)(b_{1,1}\otimes\cdots\otimes b_{1,p_1}\otimes b_{2,1}\otimes\cdots\otimes b_{n,1}\cdots\otimes b_{n,p_n})\\
&\qquad=(-1)^{\sum_{1\le i<k\le n,\;j=1,\ldots,p_i} d_{i,j}m_k}
g_1(b_{1,1},\ldots,b_{1,p_1})\otimes\cdots\otimes g_n(b_{n,1},\ldots,b_{n,p_n}) \tag*{(60)}
\end{aligned} ( g 1 ⊗ ⋯ ⊗ g n ) ( b 1 , 1 ⊗ ⋯ ⊗ b 1 , p 1 ⊗ b 2 , 1 ⊗ ⋯ ⊗ b n , 1 ⋯ ⊗ b n , p n ) = ( − 1 ) ∑ 1 ≤ i < k ≤ n , j = 1 , … , p i d i , j m k g 1 ( b 1 , 1 , … , b 1 , p 1 ) ⊗ ⋯ ⊗ g n ( b n , 1 , … , b n , p n ) (60) whenever b i , j ∈ B i , j b_{i,j}\in B_{i,j} b i , j ∈ B i , j is pure of degree d i , j d_{i,j} d i , j , that is, b i , j ∈ B i , j d i , j b_{i,j}\in B_{i,j}^{d_{i,j}} b i , j ∈ B i , j d i , j , for d i , j ∈ Z d_{i,j}\in\mathbb{Z} d i , j ∈ Z .
The important point in (60) is the sign ( − 1 ) ∑ i < k d i , j m k (-1)^{\sum_{i<k}d_{i,j}m_k} ( − 1 ) ∑ i < k d i , j m k , which we consider to be part of the definition of g 1 ⊗ ⋯ ⊗ g n g_1\otimes\cdots\otimes g_n g 1 ⊗ ⋯ ⊗ g n . Here is how to think about it: we split B i , j B_{i,j} B i , j as a direct sum B i , j = B i , j e v e n ⊕ B i , j o d d B_{i,j}=B_{i,j}^{\mathrm{even}}\oplus B_{i,j}^{\mathrm{odd}} B i , j = B i , j even ⊕ B i , j odd of ‘even’ and ‘odd’ parts, where B i , j e v e n = ⨁ d ∈ 2 Z B i , j d B_{i,j}^{\mathrm{even}}=\bigoplus_{d\in2\mathbb{Z}}B_{i,j}^{d} B i , j even = ⨁ d ∈ 2 Z B i , j d and B i , j o d d = ⨁ d ∈ 2 Z + 1 B i , j d B_{i,j}^{\mathrm{odd}}=\bigoplus_{d\in2\mathbb{Z}+1}B_{i,j}^{d} B i , j odd = ⨁ d ∈ 2 Z + 1 B i , j d . We also call g i g_i g i ‘even’ or ‘odd’ according to whether m i = deg g i m_i=\deg g_i m i = deg g i is even or odd.
The rule, known as the Koszul sign convention , is that whenever we exchange the order of two ‘odd’ elements or operators, we should change the sign by − 1 -1 − 1 . In equation (60) , the g i , b i , j g_i,b_{i,j} g i , b i , j are ordered g 1 , … , g n , b 1 , 1 , … , b n , p n g_1,\ldots,g_n,b_{1,1},\ldots,b_{n,p_n} g 1 , … , g n , b 1 , 1 , … , b n , p n on the
left hand side, but ordered g 1 , b 1 , 1 , … , b 1 , p 1 , g 2 , b 2 , 1 , … , g n , b n , 1 , … , b n , p n g_1,b_{1,1},\ldots,b_{1,p_1},g_2,b_{2,1},\ldots,g_n,b_{n,1},\ldots,b_{n,p_n} g 1 , b 1 , 1 , … , b 1 , p 1 , g 2 , b 2 , 1 , … , g n , b n , 1 , … , b n , p n on the right hand side. So the order of b i , j b_{i,j} b i , j and g k g_k g k is different on the two sides if i < k i<k i < k , which gives the sign change ( − 1 ) d i , j m k (-1)^{d_{i,j}m_k} ( − 1 ) d i , j m k , and the overall sign ( − 1 ) ∑ i < k d i , j m k (-1)^{\sum_{i<k}d_{i,j}m_k} ( − 1 ) ∑ i < k d i , j m k .
Now suppose A i , j , k A_{i,j,k} A i , j , k is a Z \mathbb{Z} Z -graded R R R -module for i = 1 , … , n i=1,\ldots,n i = 1 , … , n , j = 1 , … , p i j=1,\ldots,p_i j = 1 , … , p i and k = 1 , … , q i , j k=1,\ldots,q_{i,j} k = 1 , … , q i , j , and we are given R R R -multilinear maps
f i , j : A i , j , 1 × ⋯ × A i , j , q i , j ⟶ B i , j , f_{i,j}: A_{i,j,1}\times\cdots\times A_{i,j,q_{i,j}}\longrightarrow B_{i,j}, f i , j : A i , j , 1 × ⋯ × A i , j , q i , j ⟶ B i , j ,
graded of degree l i , j l_{i,j} l i , j for i = 1 , … , n i=1,\ldots,n i = 1 , … , n , j = 1 , … , p i j=1,\ldots,p_i j = 1 , … , p i . Equivalently, f i , j f_{i,j} f i , j is an R R R -linear map graded of degree l i , j l_{i,j} l i , j
f i , j : A i , j , 1 ⊗ ⋯ ⊗ A i , j , q i , j ⟶ B i , j . f_{i,j}: A_{i,j,1}\otimes\cdots\otimes A_{i,j,q_{i,j}}\longrightarrow B_{i,j}. f i , j : A i , j , 1 ⊗ ⋯ ⊗ A i , j , q i , j ⟶ B i , j .
Then we can show from (3.1) that
( g 1 ∘ ( f 1 , 1 ⊗ ⋯ ⊗ f 1 , p 1 ) ) ⊗ ⋯ ⊗ ( g n ∘ ( f n , 1 ⊗ ⋯ ⊗ f n , p n ) ) = ( − 1 ) ∑ 1 ≤ i < k ≤ n , j = 1 , … , p i l i , j m k ( g 1 ⊗ ⋯ ⊗ g n ) ∘ ( f 1 , 1 ⊗ ⋯ ⊗ f 1 , p 1 ⊗ f 2 , 1 ⊗ ⋯ ⊗ f n , 1 ⊗ ⋯ ⊗ f n , p n ) . (61) \begin{aligned}
&\bigl(g_1\circ(f_{1,1}\otimes\cdots\otimes f_{1,p_1})\bigr)\otimes\cdots\otimes\bigl(g_n\circ(f_{n,1}\otimes\cdots\otimes f_{n,p_n})\bigr)\\
&=(-1)^{\sum_{1\leq i<k\leq n,\ j=1,\ldots,p_i}l_{i,j}m_k}
(g_1\otimes\cdots\otimes g_n)\\
&\qquad\circ(f_{1,1}\otimes\cdots\otimes f_{1,p_1}\otimes f_{2,1}\otimes\cdots\otimes f_{n,1}\otimes\cdots\otimes f_{n,p_n}). \tag*{(61)}
\end{aligned} ( g 1 ∘ ( f 1 , 1 ⊗ ⋯ ⊗ f 1 , p 1 ) ) ⊗ ⋯ ⊗ ( g n ∘ ( f n , 1 ⊗ ⋯ ⊗ f n , p n ) ) = ( − 1 ) ∑ 1 ≤ i < k ≤ n , j = 1 , … , p i l i , j m k ( g 1 ⊗ ⋯ ⊗ g n ) ∘ ( f 1 , 1 ⊗ ⋯ ⊗ f 1 , p 1 ⊗ f 2 , 1 ⊗ ⋯ ⊗ f n , 1 ⊗ ⋯ ⊗ f n , p n ) . (61)
This is consistent with the principle that exchanging the order of f i , j f_{i,j} f i , j and g k g_k g k changes signs by ( − 1 ) l i , j m k (-1)^{l_{i,j}m_k} ( − 1 ) l i , j m k .
Of course, we could have defined g 1 ⊗ ⋯ ⊗ g n g_1\otimes\cdots\otimes g_n g 1 ⊗ ⋯ ⊗ g n without the sign ( − 1 ) ∑ i < k d i , j m k (-1)^{\sum_{i<k}d_{i,j}m_k} ( − 1 ) ∑ i < k d i , j m k in (3.1). But then the signs in the rest of the book would have to be be different, and more complicated. Convention 3.1 simplifies the signs later.
A ∞ A_\infty A ∞ -algebras, and units in A ∞ A_\infty A ∞ -algebras
Our sign conventions for A ∞ A_\infty A ∞ -algebras follow [51].
Definition 3.2. Let R R R be a commutative ring. An A ∞ A_\infty A ∞ -algebra ( A , m ) (A,m) ( A , m ) over R R R consists of:
(a) A Z \mathbb{Z} Z -graded R R R -module A = ⨁ d ∈ Z A d A=\bigoplus_{d\in\mathbb{Z}}A^d A = ⨁ d ∈ Z A d ; and
(b) Graded R R R -linear maps m k : A ⊗ k → A m_k:A^{\otimes k}\to A m k : A ⊗ k → A of degree 2 − k 2-k 2 − k for k ≥ 1 k\geq1 k ≥ 1 . Write m = ( m k ) k ≥ 1 m=(m_k)_{k\geq1} m = ( m k ) k ≥ 1 . These must satisfy the equation in maps A ⊗ n → A A^{\otimes n}\to A A ⊗ n → A for all n ≥ 1 n\geq1 n ≥ 1
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (62) \sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}\circ\left(\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\right)=0. \tag*{(62)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (62)
By abuse of notation, we also think of m k m_k m k as a graded R R R -multilinear map m k : A × ⋯ × A → A m_k:A\times\cdots\times A\to A m k : A × ⋯ × A → A of degree 2 − k 2-k 2 − k , with m k ( a 1 , … , a k ) = m k ( a 1 ⊗ ⋯ ⊗ a k ) m_k(a_1,\ldots,a_k)=m_k(a_1\otimes\cdots\otimes a_k) m k ( a 1 , … , a k ) = m k ( a 1 ⊗ ⋯ ⊗ a k ) . Then by Convention 3.1, equation (3.3) is equivalent to the equation for all n ≥ 1 n\geq1 n ≥ 1 and all pure a 1 , … , a n a_1,\ldots,a_n a 1 , … , a n in A A A
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k + j ∑ l = 1 i deg a l m i + k + 1 ( a 1 , … , a i , m j ( a i + 1 , … , a i + j ) , a i + j + 1 , … , a n ) = 0. (63) \sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk+j\sum_{l=1}^{i}\deg a_l}m_{i+k+1}(a_1,\ldots,a_i,m_j(a_{i+1},\ldots,a_{i+j}),a_{i+j+1},\ldots,a_n)=0. \tag*{(63)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k + j ∑ l = 1 i d e g a l m i + k + 1 ( a 1 , … , a i , m j ( a i + 1 , … , a i + j ) , a i + j + 1 , … , a n ) = 0. (63)
An A ∞ A_\infty A ∞ -algebra ( A , m ) (A,m) ( A , m ) is called minimal if m 1 = 0 m_1=0 m 1 = 0 .
Remark 3.3. (a) Our A ∞ A_\infty A ∞ -algebras are also called non-unital A ∞ A_\infty A ∞ -algebras, as in Seidel [84, §1] for instance. See Definitions 3.4 –3.6 for units in A ∞ A_\infty A ∞ -algebras. (b) There is an alternative way to write the definition of A ∞ A_\infty A ∞ -algebras using bar complexes , which is simpler in some ways. See §3.6.
If ( A , m ) (A,\mathfrak m) ( A , m ) is an A ∞ A_\infty A ∞ -algebra then (3.3) for n = 1 n=1 n = 1 becomes m 1 ∘ m 1 = 0 \mathfrak m_1\circ\mathfrak m_1=0 m 1 ∘ m 1 = 0 , with m 1 \mathfrak m_1 m 1 of degree 1. Thus ( A , m 1 ) (A,\mathfrak m_1) ( A , m 1 ) is a complex, and we form its cohomology H ∗ ( A ) H^*(A) H ∗ ( A ) by
H p ( A ) = Ker ( m 1 : A p → A p + 1 ) Im ( m 1 : A p − 1 → A p ) H^p(A)=\frac{\operatorname{Ker}(\mathfrak m_1:A^p\to A^{p+1})}{\operatorname{Im}(\mathfrak m_1:A^{p-1}\to A^p)} H p ( A ) = Im ( m 1 : A p − 1 → A p ) Ker ( m 1 : A p → A p + 1 )
Then m k \mathfrak m_k m k for k > 1 k>1 k > 1 induce various operations on H ∗ ( A ) H^*(A) H ∗ ( A ) . For example, (3.4) when k = 2 k=2 k = 2 yields
− m 2 ( m 1 ( a 1 ) , a 2 ) − ( − 1 ) deg a 1 m 2 ( a 1 , m 1 ( a 2 ) ) + m 1 ( m 2 ( a 1 , a 2 ) ) = 0. -\mathfrak m_2(\mathfrak m_1(a_1),a_2)-(-1)^{\deg a_1}\mathfrak m_2(a_1,\mathfrak m_1(a_2))+\mathfrak m_1(\mathfrak m_2(a_1,a_2))=0. − m 2 ( m 1 ( a 1 ) , a 2 ) − ( − 1 ) d e g a 1 m 2 ( a 1 , m 1 ( a 2 )) + m 1 ( m 2 ( a 1 , a 2 )) = 0.
This implies that the bilinear product ∙ : H p ( A ) × H q ( A ) → H p + q ( A ) \mathbin{\bullet}:H^p(A)\times H^q(A)\to H^{p+q}(A) ∙ : H p ( A ) × H q ( A ) → H p + q ( A ) given by
( a 1 + Im m 1 ) ∙ ( a 2 + Im m 1 ) = m 2 ( a 1 , a 2 ) + Im m 1 (64) (a_1+\operatorname{Im}\mathfrak m_1)\mathbin{\bullet}(a_2+\operatorname{Im}\mathfrak m_1)=\mathfrak m_2(a_1,a_2)+\operatorname{Im}\mathfrak m_1 \tag*{(64)} ( a 1 + Im m 1 ) ∙ ( a 2 + Im m 1 ) = m 2 ( a 1 , a 2 ) + Im m 1 (64)
is well-defined. Then (3.4) when k = 3 k=3 k = 3 implies that ∙ \mathbin{\bullet} ∙ is associative . Thus ( H ∗ ( A ) , ∙ ) (H^*(A),\mathbin{\bullet}) ( H ∗ ( A ) , ∙ ) is an associative R R R -algebra without unit.
Here are three ways to include units in A ∞ A_\infty A ∞ -algebras.
Definition 3.4. Let ( A , m ) (A,\mathfrak m) ( A , m ) be an A ∞ A_\infty A ∞ -algebra over R R R . A unit for ( A , m ) (A,\mathfrak m) ( A , m ) is an element 1 ∈ A 0 \mathbb{1}\in A^0 1 ∈ A 0 satisfying
m 2 ( 1 ⊗ a ) = m 2 ( a ⊗ 1 ) = a for all a ∈ A , \mathfrak m_2(\mathbb{1}\otimes a)=\mathfrak m_2(a\otimes\mathbb{1})=a\quad\text{for all }a\in A, m 2 ( 1 ⊗ a ) = m 2 ( a ⊗ 1 ) = a for all a ∈ A , m k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 for all a j ∈ A , 1 ≤ i ≤ k , k ≠ 2. (65) \mathfrak m_k(a_1\otimes\cdots\otimes a_{i-1}\otimes\mathbb{1}\otimes a_{i+1}\otimes\cdots\otimes a_k)=0\quad\text{for all }a_j\in A,\ 1\leq i\leq k,\ k\neq2. \tag*{(65)} m k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 for all a j ∈ A , 1 ≤ i ≤ k , k = 2. (65)
We call ( A , m ) (A,\mathfrak m) ( A , m ) unital if it has a unit. If 1 , 1 ′ \mathbb{1},\mathbb{1}' 1 , 1 ′ are units for ( A , m ) (A,\mathfrak m) ( A , m ) then 1 = m 2 ( 1 , 1 ′ ) = 1 ′ \mathbb{1}=\mathfrak m_2(\mathbb{1},\mathbb{1}')=\mathbb{1}' 1 = m 2 ( 1 , 1 ′ ) = 1 ′ , so a unit 1 \mathbb{1} 1 is unique if it exists. Thus, to be unital is a property of an A ∞ A_\infty A ∞ -algebra ( A , m ) (A,\mathfrak m) ( A , m ) , not extra data.
Definition 3.5. We call an A ∞ A_\infty A ∞ -algebra ( A , m ) (A,\mathfrak m) ( A , m ) over R R R weakly cohomologically unital if the associative algebra H ∗ ( A ) H^*(A) H ∗ ( A ) has a unit, that is, if there exists 1 A \mathbb{1}_A 1 A in H 0 ( A ) H^0(A) H 0 ( A ) such that 1 A ∙ α = α ∙ 1 A = α \mathbb{1}_A\mathbin{\bullet}\alpha=\alpha\mathbin{\bullet}\mathbb{1}_A=\alpha 1 A ∙ α = α ∙ 1 A = α for all α ∈ H ∗ ( A ) \alpha\in H^*(A) α ∈ H ∗ ( A ) . This 1 A \mathbb{1}_A 1 A is unique if it exists, so to be cohomologically unital is a property of ( A , m ) (A,\mathfrak m) ( A , m ) , not extra data.
Definition 3.6. A cohomological unit for an A ∞ A_\infty A ∞ -algebra ( A , m ) (A,\mathfrak m) ( A , m ) over R R R is e A e_A e A in A 0 A^0 A 0 such that m 1 ( e A ) = 0 \mathfrak m_1(e_A)=0 m 1 ( e A ) = 0 , and the following maps of chain complexes are chain homotopic to the identity:
Φ : ( A , m 1 ) ⟶ ( A , m 1 ) , Φ : a ⟼ m 2 ( e A ⊗ a ) , \Phi:(A,\mathfrak m_1)\longrightarrow(A,\mathfrak m_1),\qquad\Phi:a\longmapsto\mathfrak m_2(e_A\otimes a), Φ : ( A , m 1 ) ⟶ ( A , m 1 ) , Φ : a ⟼ m 2 ( e A ⊗ a ) , Ψ : ( A , m 1 ) ⟶ ( A , m 1 ) , Ψ : a ⟼ m 2 ( a ⊗ e A ) . \Psi:(A,\mathfrak m_1)\longrightarrow(A,\mathfrak m_1),\qquad\Psi:a\longmapsto\mathfrak m_2(a\otimes e_A). Ψ : ( A , m 1 ) ⟶ ( A , m 1 ) , Ψ : a ⟼ m 2 ( a ⊗ e A ) . These imply that the cohomology class 1 A = [ e A ] \mathbb{1}_A=[e_A] 1 A = [ e A ] in H 0 ( A ) H^0(A) H 0 ( A ) is a unit in H ∗ ( A ) H^*(A) H ∗ ( A ) . We call ( A , m ) (A,\mathfrak m) ( A , m ) cohomologically unital if it has a cohomological unit e A e_A e A .
If R R R is a field then weakly cohomologically unital and cohomologically unital coincide, since Φ , Ψ \Phi,\Psi Φ , Ψ are chain homotopic to the identity if and only if they induce the identity maps on cohomology.
A unit 1 A 1_A 1 A in ( A , m ) (A,m) ( A , m ) is a cohomological unit. Thus, ( A , m ) (A,m) ( A , m ) unital implies ( A , m ) (A,m) ( A , m ) cohomologically unital, implies ( A , m ) (A,m) ( A , m ) weakly cohomological unital.
If e A ′ ∈ A 0 e'_A \in A^0 e A ′ ∈ A 0 is any other representative for 1 A 1_A 1 A in H 0 ( A ) H^0(A) H 0 ( A ) , then e A ′ = e A + m 1 ( f ) e'_A=e_A+m_1(f) e A ′ = e A + m 1 ( f ) for f ∈ A − 1 f\in A^{-1} f ∈ A − 1 , and defining Θ , Ξ : A ∗ → A ∗ − 1 \Theta,\Xi:A^*\to A^{*-1} Θ , Ξ : A ∗ → A ∗− 1 by Θ ( a ) = m 2 ( f ⊗ a ) \Theta(a)=m_2(f\otimes a) Θ ( a ) = m 2 ( f ⊗ a ) , Ξ ( a ) = − m 2 ( a ⊗ f ) \Xi(a)=-m_2(a\otimes f) Ξ ( a ) = − m 2 ( a ⊗ f ) , we see from (63) that
Φ ′ ( a ) = Φ ( a ) + m 1 ∘ Θ ( a ) − Θ ∘ m 1 ( a ) , Ψ ′ ( a ) = Ψ ( a ) + m 1 ∘ Ξ ( a ) − Ξ ∘ m 1 ( a ) . \Phi'(a)=\Phi(a)+m_1\circ\Theta(a)-\Theta\circ m_1(a),\qquad\Psi'(a)=\Psi(a)+m_1\circ\Xi(a)-\Xi\circ m_1(a). Φ ′ ( a ) = Φ ( a ) + m 1 ∘ Θ ( a ) − Θ ∘ m 1 ( a ) , Ψ ′ ( a ) = Ψ ( a ) + m 1 ∘ Ξ ( a ) − Ξ ∘ m 1 ( a ) .
Thus, Φ ′ , Ψ ′ \Phi',\Psi' Φ ′ , Ψ ′ are chain homotopic to the identity if and only if Φ , Ψ \Phi,\Psi Φ , Ψ are, and e A ′ e'_A e A ′ is a cohomological unit if and only if e A e_A e A is. Hence, to be cohomologically unital is a property of ( A , m ) (A,m) ( A , m ) , not of the particular choice of element e A e_A e A .
Remark 3.7. Readers are warned that the names used for different kinds of unitality vary widely. Our notion of unital is often called ‘strictly unital’. Our notion of cohomologically unital is sometimes just called ‘unital’. Some references on different forms of units (for both A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories) are Keller [51, §3.5], Seidel [84, §2], Canonaco–Ornaghi–Stellari [12, §1.3], Lyubashenko–Manzyuk [65], and Tanaka [88, §1]. Seidel [84, §2] and Fukaya et al. [33, §3.3] also define homotopy units in A ∞ A_\infty A ∞ -algebras, but we do not discuss them.
Although weakly cohomologically unital A ∞ A_\infty A ∞ -algebras seem a very natural definition, the additional chain homotopy condition in cohomologically unital tends to be needed in proving theorems (see e.g. [65], [88, Th. 1.1]).
Morphisms between A ∞ A_\infty A ∞ -algebras
Here is the notion of morphism of A ∞ A_\infty A ∞ -algebras.
Definition 3.8. Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras over R R R . An A ∞ A_\infty A ∞ -morphism, or just morphism , f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) is f = ( f k ) k ≥ 1 f=(f_k)_{k\geq1} f = ( f k ) k ≥ 1 , where f k : A ⊗ k → B f_k:A^{\otimes k}\to B f k : A ⊗ k → B for k = 1 , 2 , … k=1,2,\ldots k = 1 , 2 , … are graded R R R -linear maps of degree 1 − k 1-k 1 − k , satisfying in maps A ⊗ n → B A^{\otimes n}\to B A ⊗ n → B for all n ≥ 1 n\geq1 n ≥ 1
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k f i + k + 1 ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = ∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (66) \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}f_{i+k+1}\bigl(\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\bigr)
=
\sum_{\substack{n=n_1+\cdots+n_k\\ n_a\geq1}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}n_k\bigl(f_{n_1}\otimes f_{n_2}\otimes\cdots\otimes f_{n_k}\bigr). \tag*{(66)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k f i + k + 1 ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (66)
The identity morphism id A : ( A , m ) → ( A , m ) \operatorname{id}_A:(A,m)\to(A,m) id A : ( A , m ) → ( A , m ) is id A = ( id A , k ) k ≥ 1 \operatorname{id}_A=(\operatorname{id}_{A,k})_{k\geq1} id A = ( id A , k ) k ≥ 1 with id A , 1 = id A \operatorname{id}_{A,1}=\operatorname{id}_A id A , 1 = id A and id A , k = 0 \operatorname{id}_{A,k}=0 id A , k = 0 for k > 1 k>1 k > 1 .
We call a morphism f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) strict if f k = 0 f_k=0 f k = 0 for k ≠ 1 k\ne1 k = 1 , an A ∞ A_\infty A ∞ -isomorphism if f 1 : A → B f_1:A\to B f 1 : A → B is an isomorphism of R R R -modules, and a strict A ∞ A_\infty A ∞ -isomorphism if it is both strict and an A ∞ A_\infty A ∞ -isomorphism. When n = 1 n=1 n = 1 , (66) becomes f 1 ∘ m 1 = n 1 ∘ f 1 : A → B f_1\circ m_1=n_1\circ f_1:A\to B f 1 ∘ m 1 = n 1 ∘ f 1 : A → B . Thus we have a morphism of complexes
f 1 : ( A , m 1 ) ⟶ ( B , n 1 ) , (67) f_1:(A,m_1)\longrightarrow(B,n_1), \tag*{(67)} f 1 : ( A , m 1 ) ⟶ ( B , n 1 ) , (67)
which induces a morphism of cohomology groups
H ∗ ( f 1 ) : H ∗ ( A , m 1 ) ⟶ H ∗ ( B , n 1 ) . (68) H^*(f_1): H^*(A,m_1) \longrightarrow H^*(B,n_1). \tag*{(68)} H ∗ ( f 1 ) : H ∗ ( A , m 1 ) ⟶ H ∗ ( B , n 1 ) . (68)
Following [33, Def. 3.2.10], we call a morphism f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) a weak homotopy equivalence if (67) is a chain homotopy equivalence. This is automatic if f f f is an A ∞ A_\infty A ∞ -isomorphism. As in [33, Rem. 3.2.11], if R R R is a field then f f f is a weak homotopy equivalence if and only if (68) is an isomorphism.
If ( A , m ) , ( B , n ) , ( C , o ) (A,m),(B,n),(C,o) ( A , m ) , ( B , n ) , ( C , o ) are A ∞ A_\infty A ∞ -algebras and f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) , g : ( B , n ) ⟶ ( C , o ) g:(B,n)\longrightarrow(C,o) g : ( B , n ) ⟶ ( C , o ) are morphisms, the composition g ∘ f : ( A , m ) ⟶ ( C , o ) g\circ f:(A,m)\longrightarrow(C,o) g ∘ f : ( A , m ) ⟶ ( C , o ) is given by
( g ∘ f ) n = ∑ n = n 1 + ⋯ + n k n i ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) g k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (69) (g\circ f)_n=\sum_{\substack{n=n_1+\cdots+n_k\\ n_i\geq1}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}g_k(f_{n_1}\otimes f_{n_2}\otimes\cdots\otimes f_{n_k}). \tag*{(69)} ( g ∘ f ) n = n = n 1 + ⋯ + n k n i ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) g k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (69)
Composition is associative. Write A ∞ A l g R A_\infty\mathrm{Alg}_R A ∞ Alg R for the category with objects A ∞ A_\infty A ∞ -algebras ( A , m ) (A,m) ( A , m ) over R R R and morphisms f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) .
If ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) have units 1 A , 1 B 1_A,1_B 1 A , 1 B , as in Definition 3.4, we call a morphism f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) unital if
f 1 ( 1 A ) = 1 B and f k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 A ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (70) f_1(1_A)=1_B\qquad\text{and}\qquad f_k(a_1\otimes\cdots\otimes a_{i-1}\otimes1_A\otimes a_{i+1}\otimes\cdots\otimes a_k)=0 \tag*{(70)} f 1 ( 1 A ) = 1 B and f k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 A ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (70)
for all a j ∈ A a_j\in A a j ∈ A , i = 1 , … , k i=1,\ldots,k i = 1 , … , k , and k ≠ 1 k\ne1 k = 1 .
Unital A ∞ A_\infty A ∞ -algebras over R R R and unital morphisms form a non-full subcategory A ∞ A l g R u n ⊂ A ∞ A l g R A_\infty\mathrm{Alg}^{\mathrm{un}}_R\subset A_\infty\mathrm{Alg}_R A ∞ Alg R un ⊂ A ∞ Alg R .
Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be (weakly) cohomologically unital A ∞ A_\infty A ∞ -algebras, as in Definitions 3.5 and 3.6, and f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) be a morphism. We call f f f (weakly) cohomologically unital if H 0 ( f 1 ) ( 1 A ) = 1 B H^0(f_1)(1_A)=1_B H 0 ( f 1 ) ( 1 A ) = 1 B . Then (weakly) cohomologically unital A ∞ A_\infty A ∞ -algebras over R R R and (weakly) cohomologically unital morphisms form non-full subcategories A ∞ A l g R c u ⊂ A ∞ A l g R w c ⊂ A ∞ A l g R A_\infty\mathrm{Alg}^{\mathrm{cu}}_R\subset A_\infty\mathrm{Alg}^{\mathrm{wc}}_R\subset A_\infty\mathrm{Alg}_R A ∞ Alg R cu ⊂ A ∞ Alg R wc ⊂ A ∞ Alg R . When ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) are unital, unital morphisms f : ( A , m ) ⟶ ( B , n ) f:(A,m)\longrightarrow(B,n) f : ( A , m ) ⟶ ( B , n ) are cohomologically unital, so A ∞ A l g R u n ⊂ A ∞ A l g R c u A_\infty\mathrm{Alg}^{\mathrm{un}}_R\subset A_\infty\mathrm{Alg}^{\mathrm{cu}}_R A ∞ Alg R un ⊂ A ∞ Alg R cu .
Homotopies between A ∞ A_\infty A ∞ -morphisms, and A ∞ A_\infty A ∞ -algebras
Now let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras, and f , g : ( A , m ) ⟶ ( B , n ) f,g:(A,m)\longrightarrow(B,n) f , g : ( A , m ) ⟶ ( B , n ) be morphisms. We will define the notion of homotopy H \mathcal{H} H from f f f to g g g . Our definition is based on Keller [51, §3.7]. Fukaya et al. [33, §4] use a different, more complicated definition, involving ‘models of [ 0 , 1 ] × B [0,1]\times B [ 0 , 1 ] × B ’, but they show in [33, Prop. 4.2.40] that the two definitions yield the same notion of whether f , g f,g f , g are homotopic.
Definition 3.9. Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras over R R R , and f , g : ( A , m ) ⟶ ( B , n ) f,g:(A,m)\longrightarrow(B,n) f , g : ( A , m ) ⟶ ( B , n ) be morphisms. A homotopy H : f ⇒ g \mathcal{H}:f\Rightarrow g H : f ⇒ g from f f f to g g g is H = ( H k ) k ≥ 1 \mathcal{H}=(\mathcal{H}_k)_{k\geq1} H = ( H k ) k ≥ 1 , where
h k : A ⊗ k → B \mathfrak{h}_k : A^{\otimes k} \to B h k : A ⊗ k → B for k = 1 , 2 , … k = 1, 2, \ldots k = 1 , 2 , … are graded R R R -linear maps of degree − k -k − k , satisfying
f n − g n = ∑ n = n 1 + ⋯ + n k , n i ≥ 1 , 1 ≤ l ≤ k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n l − 1 ⊗ h n l ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ) + ∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k h i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) . (71) \begin{aligned}
f_n-g_n =
\sum_{\substack{n=n_1+\cdots+n_k,\\ n_i\geq1,\ 1\leq l\leq k}}
(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}
n_k\left(f_{n_1}\otimes f_{n_2}\otimes\cdots\otimes f_{n_{l-1}}\otimes\mathfrak{h}_{n_l}\otimes g_{n_{l+1}}\otimes\cdots\otimes g_{n_k}\right) \\
&+\sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}
(-1)^{i+jk}\mathfrak{h}_{i+k+1}\circ\left(\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\right). \tag*{(71)}
\end{aligned} f n − g n = n = n 1 + ⋯ + n k , n i ≥ 1 , 1 ≤ l ≤ k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n l − 1 ⊗ h n l ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ) + i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k h i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) . (71)
When n = 1 n=1 n = 1 , equation (71) becomes
f 1 − g 1 = n 1 ∘ h 1 + h 1 ∘ m 1 . f_1-g_1=n_1\circ\mathfrak{h}_1+\mathfrak{h}_1\circ m_1. f 1 − g 1 = n 1 ∘ h 1 + h 1 ∘ m 1 .
This implies that h 1 : A ∗ → B ∗ − 1 \mathfrak{h}_1:A^*\to B^{*-1} h 1 : A ∗ → B ∗− 1 is a chain homotopy between the morphisms of chain complexes f 1 , g 1 : ( A , m 1 ) → ( B , n 1 ) f_1,g_1:(A,m_1)\to(B,n_1) f 1 , g 1 : ( A , m 1 ) → ( B , n 1 ) . Thus for the morphisms (68) on cohomology we have
H ∗ ( f ) = H ∗ ( g ) : H ∗ ( A ) ⟶ H ∗ ( B ) . H^*(f)=H^*(g):H^*(A)\longrightarrow H^*(B). H ∗ ( f ) = H ∗ ( g ) : H ∗ ( A ) ⟶ H ∗ ( B ) .
If f , g : ( A , m ) → ( B , n ) f,g:(A,m)\to(B,n) f , g : ( A , m ) → ( B , n ) are unital morphisms of unital A ∞ A_\infty A ∞ -algebras, as in Definitions 3.4 and 3.8, we call a homotopy h : f ⇒ g \mathfrak{h}:f\Rightarrow g h : f ⇒ g unital if for all 1 ≤ i ≤ k 1\leq i\leq k 1 ≤ i ≤ k and a j ∈ A a_j\in A a j ∈ A we have
h k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 A ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0. (72) \mathfrak{h}_k(a_1\otimes\cdots\otimes a_{i-1}\otimes1_A\otimes a_{i+1}\otimes\cdots\otimes a_k)=0. \tag*{(72)} h k ( a 1 ⊗ ⋯ ⊗ a i − 1 ⊗ 1 A ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0. (72)
For (weakly ) cohomologically unital A ∞ A_\infty A ∞ -algebras and morphisms, as in Definitions 3.5 , 3.6 and 3.8, there seems to be no useful notion of (weakly ) cohomologically unital homotopy between (weakly ) cohomologically unital morphisms, one should allow all homotopies.
Definition 3.10. Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras, and f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) a morphism. A homotopy inverse for f f f is a morphism g : ( B , n ) → ( A , m ) g:(B,n)\to(A,m) g : ( B , n ) → ( A , m ) such that g ∘ f : ( A , m ) → ( A , m ) g\circ f:(A,m)\to(A,m) g ∘ f : ( A , m ) → ( A , m ) is homotopic to id A : ( A , m ) → ( A , m ) \operatorname{id}_A:(A,m)\to(A,m) id A : ( A , m ) → ( A , m ) , and f ∘ g : ( B , n ) → ( B , n ) f\circ g:(B,n)\to(B,n) f ∘ g : ( B , n ) → ( B , n ) is homotopic to id B : ( B , n ) → ( B , n ) \operatorname{id}_B:(B,n)\to(B,n) id B : ( B , n ) → ( B , n ) . If f f f has a homotopy inverse, we call f f f a homotopy equivalence , and we call ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) homotopy equivalent . If f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) is a homotopy equivalence then H ∗ ( f ) : H ∗ ( A ) → H ∗ ( B ) H^*(f):H^*(A)\to H^*(B) H ∗ ( f ) : H ∗ ( A ) → H ∗ ( B ) is an isomorphism.
The next theorem is proved by Fukaya et al. [33, Cor. 4.2.44, Th. 4.2.45], see also Keller [51, §3.7], Lefèvre-Hasegawa [57, Cor. 1.3.1.3] and Seidel [84, §1(h)].
Theorem 3.11. Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras over R R R . Then
(a) Homotopy is an equivalence relation on morphisms f , g : ( A , m ) → ( B , n ) f,g:(A,m)\to(B,n) f , g : ( A , m ) → ( B , n ) .
(b) Homotopy equivalence is an equivalence relation on A ∞ A_\infty A ∞ -algebras.
(c) A morphism f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) is a homotopy equivalence if and only if it is a weak homotopy equivalence in the sense of Definition 3.8. Hence, any A ∞ A_\infty A ∞ -isomorphism is a homotopy equivalence.
When R R R is a field, Kadeishvili [47] proves:
Theorem 3.12. Let ( A , m ) (A,\mathbf{m}) ( A , m ) be an A ∞ A_{\infty} A ∞ -algebra over a field R R R . Then there exists a minimal A ∞ A_{\infty} A ∞ -algebra ( B , n ) (B,\mathbf{n}) ( B , n ) homotopy equivalent to ( A , m ) (A,\mathbf{m}) ( A , m ) , in the sense of Definition 3.2. As graded R R R -modules we have B ≅ H ∗ ( A ) B \cong H^*(A) B ≅ H ∗ ( A ) .
In practice, homotopy equivalence is a more useful notion of when two A ∞ A_{\infty} A ∞ -algebras are ‘the same’ than (strict) A ∞ A_{\infty} A ∞ -isomorphism. We are interested in properties of A ∞ A_{\infty} A ∞ -algebras which are invariant under homotopy. Constructions of A ∞ A_{\infty} A ∞ -algebras generally depend on some arbitrary choices, and different choices yield homotopy equivalent but not (strictly) isomorphic A ∞ A_{\infty} A ∞ -algebras.
Definition 3.13. The homotopy category Ho ( A ∞ A l g R ) \operatorname{Ho}(A_{\infty}\mathrm{Alg}_R) Ho ( A ∞ Alg R ) of A ∞ A l g R A_{\infty}\mathrm{Alg}_R A ∞ Alg R is the category with objects A ∞ A_{\infty} A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m ) over R R R and morphisms homotopy classes [ f ] [f] [ f ] of morphisms f : ( A , m ) → ( B , n ) f:(A,\mathbf{m})\to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) , with composition [ g ] ∘ [ f ] = [ g ∘ f ] [g]\circ[f]=[g\circ f] [ g ] ∘ [ f ] = [ g ∘ f ] .
For cohomologically unital A ∞ A_{\infty} A ∞ -categories, we define the homotopy category Ho ( A ∞ A l g R c u ) \operatorname{Ho}(A_{\infty}\mathrm{Alg}^{\mathrm{cu}}_R) Ho ( A ∞ Alg R cu ) of A ∞ A l g R c u A_{\infty}\mathrm{Alg}^{\mathrm{cu}}_R A ∞ Alg R cu in the same way.
For unital A ∞ A_{\infty} A ∞ -categories, we define the homotopy category Ho ( A ∞ A l g R u n ) \operatorname{Ho}(A_{\infty}\mathrm{Alg}^{\mathrm{un}}_R) Ho ( A ∞ Alg R un ) of A ∞ A l g R u n A_{\infty}\mathrm{Alg}^{\mathrm{un}}_R A ∞ Alg R un by taking equivalence classes of morphisms under unital homotopies.
A ∞ A_{\infty} A ∞ -algebras and differential graded algebras
Definition 3.14. A differential graded algebra (dga ) over R R R is a quadruple ( A , ⋅ , 1 , d ) (A,\cdot,1,d) ( A , ⋅ , 1 , d ) , where A = ⨁ d ∈ Z A d A=\bigoplus_{d\in\mathbb{Z}}A^d A = ⨁ d ∈ Z A d is a Z \mathbb{Z} Z -graded R R R -module, and ⋅ : A × A → A \cdot:A\times A\to A ⋅ : A × A → A is a graded associative R R R -bilinear product of degree 0 0 0 , and 1 1 1 is an identity for ⋅ \cdot ⋅ , and d : A → A d:A\to A d : A → A is graded R R R -bilinear of degree 1 1 1 with d 2 = 0 d^2=0 d 2 = 0 , d ( a ⋅ b ) = ( d a ) ⋅ b + ( − 1 ) deg a a ⋅ ( d b ) d(a\cdot b)=(da)\cdot b+(-1)^{\deg a}a\cdot(db) d ( a ⋅ b ) = ( d a ) ⋅ b + ( − 1 ) d e g a a ⋅ ( d b ) for pure a , b ∈ A a,b\in A a , b ∈ A , and d 1 = 0 d1=0 d 1 = 0 . We usually write dgas as A A A , leaving the data ⋅ , 1 , d \cdot,1,d ⋅ , 1 , d implicit, and sometimes write this as ⋅ A , 1 A , d A \cdot_A,1_A,d_A ⋅ A , 1 A , d A .
Define graded linear m k : A ⊗ k → A m_k:A^{\otimes k}\to A m k : A ⊗ k → A for k ≥ 0 k\geq0 k ≥ 0 by
m 1 ( a ) = d a , m 2 ( a ⊗ b ) = a ⋅ b , m k = 0 , k ≠ 1 , 2. m_1(a)=da,\quad m_2(a\otimes b)=a\cdot b,\quad m_k=0,\quad k\ne1,2. m 1 ( a ) = d a , m 2 ( a ⊗ b ) = a ⋅ b , m k = 0 , k = 1 , 2.
Then ( A , m ) (A,\mathbf{m}) ( A , m ) is an A ∞ A_{\infty} A ∞ -algebra. It has a unit 1 1 1 in the sense of Definition 3.4.
If A , B A,B A , B are dgas over R R R , a dga morphism , or morphism f : A → B f:A\to B f : A → B is an R R R -linear map f : A → B f:A\to B f : A → B which preserves all the structure, so that f f f is graded of degree 0 0 0 with f ( a ⋅ a ′ ) = f ( a ) ⋅ f ( a ′ ) f(a\cdot a')=f(a)\cdot f(a') f ( a ⋅ a ′ ) = f ( a ) ⋅ f ( a ′ ) , d B f ( a ) = f ( d A a ) d_Bf(a)=f(d_Aa) d B f ( a ) = f ( d A a ) , and f ( 1 A ) = 1 B f(1_A)=1_B f ( 1 A ) = 1 B .
Then regarding A , B A,B A , B as A ∞ A_{\infty} A ∞ -algebras, defining f 1 = f f_1=f f 1 = f and f k = 0 f_k=0 f k = 0 for k > 1 k>1 k > 1 , we find that f = ( f k ) k ≥ 1 \mathbf{f}=(f_k)_{k\geq1} f = ( f k ) k ≥ 1 is an A ∞ A_{\infty} A ∞ -algebra morphism f : A → B f:A\to B f : A → B , which is unital in the sense of Definition 3.8. Note however that if f = ( f k ) k ≥ 1 : A → B \mathbf{f}=(f_k)_{k\geq1}:A\to B f = ( f k ) k ≥ 1 : A → B is a unital A ∞ A_{\infty} A ∞ -morphism, then f f f comes from a dga morphism f = f 1 : A → B f=f_1:A\to B f = f 1 : A → B if and only if f k = 0 f_k=0 f k = 0 for k > 1 k>1 k > 1 .
Write D G A R \mathrm{DGA}_R DGA R for the category of dgas over R R R , with morphisms dga morphisms f : A → B f:A\to B f : A → B . Also write D G A R A ∞ \mathrm{DGA}^{A_{\infty}}_R DGA R A ∞ for the category of dgas over R R R , with morphisms unital A ∞ A_{\infty} A ∞ -algebra morphisms f : A → B f:A\to B f : A → B . Then D G A R ⊂ D G A R A ∞ \mathrm{DGA}_R\subset\mathrm{DGA}^{A_{\infty}}_R DGA R ⊂ DGA R A ∞ is a non-full subcategory. Also D G A R A ∞ ⊂ A ∞ A l g R u n \mathrm{DGA}^{A_{\infty}}_R\subset A_{\infty}\mathrm{Alg}^{\mathrm{un}}_R DGA R A ∞ ⊂ A ∞ Alg R un is a full subcategory.
As in Definition 3.13, we write Ho ( D G A R A ∞ ) ⊂ Ho ( A ∞ A l g R u n ) \operatorname{Ho}(\mathrm{DGA}^{A_{\infty}}_R)\subset\operatorname{Ho}(A_{\infty}\mathrm{Alg}^{\mathrm{un}}_R) Ho ( DGA R A ∞ ) ⊂ Ho ( A ∞ Alg R un ) for the homotopy category of D G A R A ∞ \mathrm{DGA}^{A_{\infty}}_R DGA R A ∞ , whose morphisms are unital homotopy equivalence classes of morphisms in D G A R A ∞ \mathrm{DGA}^{A_{\infty}}_R DGA R A ∞ .
A dga morphism f : A → B f:A\to B f : A → B is a quasi-isomorphism if f : ( A , d A ) → ( B , d B ) f:(A,d_A)\to(B,d_B) f : ( A , d A ) → ( B , d B ) is a quasi-isomorphism of complexes, that is, H k ( f ) : H k ( A , d A ) → H k ( B , d B ) H^k(f):H^k(A,d_A)\to H^k(B,d_B) H k ( f ) : H k ( A , d A ) → H k ( B , d B ) is an isomorphism for all k ∈ Z k\in\mathbb{Z} k ∈ Z . Also f f f is a chain homotopy equivalence if f : ( A , d A ) → ( B , d B ) f:(A,d_A)\to(B,d_B) f : ( A , d A ) → ( B , d B ) is a chain homotopy equivalence of complexes. If R R R is a field then every quasi-isomorphism is a chain homotopy equivalence.
Defining A ∞ A_\infty A ∞ -algebras using bar complexes
See Getzler–Jones [36, §1] and Keller [51, §3.6] for the material of this section.
Definition 3.15. Let ( A , m ) (A,m) ( A , m ) be an A ∞ A_\infty A ∞ -algebra over R R R . Write A [ 1 ] A[1] A [ 1 ] for A A A with grading shifted by 1 1 1 , that is, A [ 1 ] k = A k + 1 A[1]^k=A^{k+1} A [ 1 ] k = A k + 1 . Write S [ 1 ] : A → A [ 1 ] S[1]:A\to A[1] S [ 1 ] : A → A [ 1 ] for the identity map id A \operatorname{id}_A id A , thought of as an R R R -linear map graded of degree 1 1 1 . Write S [ − 1 ] : A [ 1 ] → A S[-1]:A[1]\to A S [ − 1 ] : A [ 1 ] → A for its inverse, which is also id A \operatorname{id}_A id A . We call S [ ± 1 ] S[\pm1] S [ ± 1 ] shift maps .
Define R R R -linear maps m ˉ k : A [ 1 ] ⊗ k → A [ 1 ] \bar{m}_k:A[1]^{\otimes k}\to A[1] m ˉ k : A [ 1 ] ⊗ k → A [ 1 ] for k ≥ 1 k\geq1 k ≥ 1 by
m ˉ k = ( − 1 ) k ( k − 1 ) / 2 S [ 1 ] ∘ m k ∘ S [ − 1 ] ⊗ k . (73) \bar{m}_k=(-1)^{k(k-1)/2}S[1]\circ m_k\circ S[-1]^{\otimes k}. \tag*{(73)} m ˉ k = ( − 1 ) k ( k − 1 ) /2 S [ 1 ] ∘ m k ∘ S [ − 1 ] ⊗ k . (73) Since m k m_k m k is graded of degree 2 − k 2-k 2 − k , m ˉ k \bar{m}_k m ˉ k is graded of degree 1 1 1 . As ( S [ − 1 ] ⊗ k ) ∘ ( S [ 1 ] ⊗ k ) = ( − 1 ) k ( k − 1 ) / 2 id (S[-1]^{\otimes k})\circ(S[1]^{\otimes k})=(-1)^{k(k-1)/2}\operatorname{id} ( S [ − 1 ] ⊗ k ) ∘ ( S [ 1 ] ⊗ k ) = ( − 1 ) k ( k − 1 ) /2 id by (61) , equation (73) is equivalent to
m k = S [ − 1 ] ∘ m ˉ k ∘ S [ 1 ] ⊗ k , m_k=S[-1]\circ\bar{m}_k\circ S[1]^{\otimes k}, m k = S [ − 1 ] ∘ m ˉ k ∘ S [ 1 ] ⊗ k , Here is a subtle point: m k m_k m k and m ˉ k \bar{m}_k m ˉ k are essentially the same map, as S [ ± 1 ] S[\pm1] S [ ± 1 ] are id A \operatorname{id}_A id A . However, when we translate equations in the m k m_k m k such as (62) into equations in the m ˉ k \bar{m}_k m ˉ k , there may be sign changes, because of the Koszul sign rule in Convention 3.1. For example, if i , k ≥ 0 i,k\geq0 i , k ≥ 0 and j ≥ 1 j\geq1 j ≥ 1 with i + j + k = n i+j+k=n i + j + k = n then
m ˉ i + k + 1 ∘ [ id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ] = ( − 1 ) ( i + k + 1 ) ( i + k ) / 2 ( − 1 ) j ( j − 1 ) / 2 S [ 1 ] ∘ m i + k + 1 ∘ S [ − 1 ] ⊗ i + k + 1 ∘ [ ( S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ i ⊗ ( S [ 1 ] ∘ m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ k ] = ( − 1 ) n ( n − 1 ) / 2 + ( j + 1 ) ( i + k ) ( − 1 ) k S [ 1 ] ∘ m i + k + 1 ∘ [ ( S [ − 1 ] ∘ S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ i ] ⊗ ( S [ − 1 ] ∘ S [ 1 ] ∘ m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ − 1 ] ∘ S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ k = ( − 1 ) n ( n − 1 ) / 2 + i + i j + j k S [ 1 ] ∘ m i + k + 1 ∘ [ S [ − 1 ] ⊗ i ⊗ ( m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ − 1 ] ) ⊗ k ] = ( − 1 ) n ( n − 1 ) / 2 + i + i j + j k ( − 1 ) i ( 2 − j ) S [ 1 ] ∘ m i + k + 1 ∘ [ id A ⊗ i ⊗ m j ⊗ id A ⊗ k ] ∘ S [ − 1 ] ⊗ i + j + k = ( − 1 ) n ( n − 1 ) / 2 ( − 1 ) i + j k S [ 1 ] ∘ m i + k + 1 ∘ [ id A ⊗ i ⊗ m j ⊗ id A ⊗ k ] ∘ S [ − 1 ] ⊗ i + j + k , (74) \begin{aligned}
\bar{m}_{i+k+1}\circ\left[\operatorname{id}_{A[1]}^{\otimes i}\otimes\bar{m}_j\otimes\operatorname{id}_{A[1]}^{\otimes k}\right]
={}&(-1)^{(i+k+1)(i+k)/2}(-1)^{j(j-1)/2}S[1]\circ m_{i+k+1}\circ S[-1]^{\otimes i+k+1}\\
&\circ\left[(S[1]\circ\operatorname{id}_A\circ S[-1])^{\otimes i}\otimes(S[1]\circ m_j\circ S[-1]^{\otimes j})\otimes(S[1]\circ\operatorname{id}_A\circ S[-1])^{\otimes k}\right]\\
={}&(-1)^{n(n-1)/2+(j+1)(i+k)}(-1)^kS[1]\circ m_{i+k+1}\circ\left[(S[-1]\circ S[1]\circ\operatorname{id}_A\circ S[-1])^{\otimes i}\right]\\
&\otimes(S[-1]\circ S[1]\circ m_j\circ S[-1]^{\otimes j})\otimes(S[-1]\circ S[1]\circ\operatorname{id}_A\circ S[-1])^{\otimes k}\\
={}&(-1)^{n(n-1)/2+i+ij+jk}S[1]\circ m_{i+k+1}\circ\left[S[-1]^{\otimes i}\otimes(m_j\circ S[-1]^{\otimes j})\otimes(S[-1])^{\otimes k}\right]\\
={}&(-1)^{n(n-1)/2+i+ij+jk}(-1)^{i(2-j)}S[1]\circ m_{i+k+1}\circ\left[\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\right]\circ S[-1]^{\otimes i+j+k}\\
={}&(-1)^{n(n-1)/2}(-1)^{i+jk}S[1]\circ m_{i+k+1}\circ\left[\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\right]\circ S[-1]^{\otimes i+j+k}, \tag*{(74)}
\end{aligned} m ˉ i + k + 1 ∘ [ id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ] = = = = = ( − 1 ) ( i + k + 1 ) ( i + k ) /2 ( − 1 ) j ( j − 1 ) /2 S [ 1 ] ∘ m i + k + 1 ∘ S [ − 1 ] ⊗ i + k + 1 ∘ [ ( S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ i ⊗ ( S [ 1 ] ∘ m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ k ] ( − 1 ) n ( n − 1 ) /2 + ( j + 1 ) ( i + k ) ( − 1 ) k S [ 1 ] ∘ m i + k + 1 ∘ [ ( S [ − 1 ] ∘ S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ i ] ⊗ ( S [ − 1 ] ∘ S [ 1 ] ∘ m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ − 1 ] ∘ S [ 1 ] ∘ id A ∘ S [ − 1 ] ) ⊗ k ( − 1 ) n ( n − 1 ) /2 + i + ij + j k S [ 1 ] ∘ m i + k + 1 ∘ [ S [ − 1 ] ⊗ i ⊗ ( m j ∘ S [ − 1 ] ⊗ j ) ⊗ ( S [ − 1 ] ) ⊗ k ] ( − 1 ) n ( n − 1 ) /2 + i + ij + j k ( − 1 ) i ( 2 − j ) S [ 1 ] ∘ m i + k + 1 ∘ [ id A ⊗ i ⊗ m j ⊗ id A ⊗ k ] ∘ S [ − 1 ] ⊗ i + j + k ( − 1 ) n ( n − 1 ) /2 ( − 1 ) i + j k S [ 1 ] ∘ m i + k + 1 ∘ [ id A ⊗ i ⊗ m j ⊗ id A ⊗ k ] ∘ S [ − 1 ] ⊗ i + j + k , (74) using (61) in the second and fourth steps.
Combining (62) and (74) shows that the m ˉ k \bar{m}_k m ˉ k satisfy the identities for n ≥ 1 n\geq1 n ≥ 1
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n m ˉ i + k + 1 ∘ ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) = 0 , (75) \sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}\bar{m}_{i+k+1}\circ\left(\operatorname{id}_{A[1]}^{\otimes i}\otimes\bar{m}_j\otimes\operatorname{id}_{A[1]}^{\otimes k}\right)=0, \tag*{(75)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ m ˉ i + k + 1 ∘ ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) = 0 , (75) as in Keller [51, §3.6]. Note the sign change compared to (62) .
Now let f : ( A , m ) → ( B , n ) f : (A,m) \to(B,n) f : ( A , m ) → ( B , n ) be a morphism of A ∞ A_\infty A ∞ -algebras. Define R R R -linear maps f ˉ k : A [ 1 ] ⊗ k → B [ 1 ] \bar{f}_k : A[1]^{\otimes k} \to B[1] f ˉ k : A [ 1 ] ⊗ k → B [ 1 ] for k ≥ 1 k \ge1 k ≥ 1 by
f ˉ k = ( − 1 ) k ( k − 1 ) / 2 S B [ 1 ] ∘ f k ∘ S A [ − 1 ] ⊗ k . \bar{f}_k = (-1)^{k(k-1)/2}S_{B[1]} \circ f_k \circ S_{A[-1]}^{\otimes k}. f ˉ k = ( − 1 ) k ( k − 1 ) /2 S B [ 1 ] ∘ f k ∘ S A [ − 1 ] ⊗ k .
Note that although f k f_k f k is graded of degree 1 − k 1-k 1 − k , f ˉ k \bar{f}_k f ˉ k is graded of degree 0 0 0 . Then as for (62) and (75) , equation (66) is equivalent to the equation for n ≥ 1 n \ge1 n ≥ 1
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n f ˉ i + k + 1 ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) = ∑ n = k 1 + ⋯ + k l , k a ≥ 1 n ˉ l ( f ˉ k 1 ⊗ f ˉ k 2 ⊗ ⋯ ⊗ f ˉ k l ) . (76) \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}
\bar{f}_{i+k+1}\left(\operatorname{id}_{A[1]}^{\otimes i}\otimes\bar{m}_j\otimes\operatorname{id}_{A[1]}^{\otimes k}\right)
=
\sum_{\substack{n=k_1+\cdots+k_l,\ k_a\geq1}}
\bar{n}_l\left(\bar{f}_{k_1}\otimes\bar{f}_{k_2}\otimes\cdots\otimes\bar{f}_{k_l}\right). \tag*{(76)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ f ˉ i + k + 1 ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) = n = k 1 + ⋯ + k l , k a ≥ 1 ∑ n ˉ l ( f ˉ k 1 ⊗ f ˉ k 2 ⊗ ⋯ ⊗ f ˉ k l ) . (76)
Next let f , g : ( A , m ) → ( B , n ) f,g : (A,m) \to(B,n) f , g : ( A , m ) → ( B , n ) be morphisms of A ∞ A_\infty A ∞ -algebras, and H : f ⇒ g H:f\Rightarrow g H : f ⇒ g be a homotopy. Define R R R -linear maps H ˉ k : A [ 1 ] ⊗ k → B [ 1 ] \bar{H}_k : A[1]^{\otimes k}\to B[1] H ˉ k : A [ 1 ] ⊗ k → B [ 1 ] for k ≥ 1 k\geq1 k ≥ 1 by
H ˉ k = ( − 1 ) k ( k − 1 ) / 2 S B [ 1 ] ∘ H k ∘ S A [ − 1 ] ⊗ k . \bar{H}_k=(-1)^{k(k-1)/2}S_{B[1]}\circ H_k\circ S_{A[-1]}^{\otimes k}. H ˉ k = ( − 1 ) k ( k − 1 ) /2 S B [ 1 ] ∘ H k ∘ S A [ − 1 ] ⊗ k .
Then H ˉ k \bar{H}_k H ˉ k is graded of degree − 1 -1 − 1 , and equation (71) is equivalent to the equation for n ≥ 1 n\geq1 n ≥ 1
f ˉ n − g ˉ n = ∑ l , m ≥ 0 , i 1 , … , i l , j , k 1 , … , k m ≥ 1 n = i 1 + ⋯ + i l + j + k 1 + ⋯ + k m n ˉ l + m + 1 ∘ ( f ˉ i 1 ⊗ ⋯ ⊗ f ˉ i l ⊗ H ˉ j ⊗ g ˉ k 1 ⊗ ⋯ ⊗ g ˉ k m ) + ∑ i , k ≥ 0 , j ≥ 1 i + j + k = n H ˉ i + k + 1 ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) . \begin{aligned}
\bar{f}_n-\bar{g}_n
={}&\sum_{\substack{l,m\geq0,\ i_1,\ldots,i_l,j,k_1,\ldots,k_m\geq1\\
n=i_1+\cdots+i_l+j+k_1+\cdots+k_m}}
\bar{n}_{l+m+1}\circ
\left(\bar{f}_{i_1}\otimes\cdots\otimes\bar{f}_{i_l}\otimes\bar{H}_j\otimes\bar{g}_{k_1}\otimes\cdots\otimes\bar{g}_{k_m}\right)\\
&+\sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}
\bar{H}_{i+k+1}\left(\operatorname{id}_{A[1]}^{\otimes i}\otimes\bar{m}_j\otimes\operatorname{id}_{A[1]}^{\otimes k}\right).
\end{aligned} f ˉ n − g ˉ n = l , m ≥ 0 , i 1 , … , i l , j , k 1 , … , k m ≥ 1 n = i 1 + ⋯ + i l + j + k 1 + ⋯ + k m ∑ n ˉ l + m + 1 ∘ ( f ˉ i 1 ⊗ ⋯ ⊗ f ˉ i l ⊗ H ˉ j ⊗ g ˉ k 1 ⊗ ⋯ ⊗ g ˉ k m ) + i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ H ˉ i + k + 1 ( id A [ 1 ] ⊗ i ⊗ m ˉ j ⊗ id A [ 1 ] ⊗ k ) .
We can also use labelled trees Γ \Gamma Γ as in §6.1 to represent compositions of the shifted operators m ˉ j \bar{m}_j m ˉ j , f ˉ j \bar{f}_j f ˉ j , H ˉ j \bar{H}_j H ˉ j , … above. Then we will write the compositions as C ˉ ( Γ ) \bar{C}(\Gamma) C ˉ ( Γ ) rather than C ( Γ ) C(\Gamma) C ( Γ ) .
As in Keller [51, §3.6], §3.2–§ 3.4 can be rewritten in terms of bar complexes .
Definition 3.16. Let ( A , m ) (A,m) ( A , m ) be an A ∞ A_\infty A ∞ -algebra over R R R . Write A [ 1 ] A[1] A [ 1 ] for A A A with grading shifted by 1 1 1 , that is, A [ 1 ] k = A k + 1 A[1]^k=A^{k+1} A [ 1 ] k = A k + 1 . The tensor coalgebra T ( A [ 1 ] ) T(A[1]) T ( A [ 1 ]) of A [ 1 ] A[1] A [ 1 ] is
T ( A [ 1 ] ) = ⨁ n ≥ 0 A [ 1 ] ⊗ n , T(A[1])=\bigoplus_{n\geq0}A[1]^{\otimes n}, T ( A [ 1 ]) = n ≥ 0 ⨁ A [ 1 ] ⊗ n , where we write A ⊗ 0 = R A^{\otimes0}=R A ⊗ 0 = R . It is graded in the obvious way, so that
T ( A [ 1 ] ) d = ⨁ d 1 + ⋯ + d n = d A [ 1 ] d 1 ⊗ ⋯ ⊗ A [ 1 ] d n = ⨁ d 1 + ⋯ + d n = d A d 1 + 1 ⊗ ⋯ ⊗ A d n + 1 . T(A[1])^d=\bigoplus_{d_1+\cdots+d_n=d}A[1]^{d_1}\otimes\cdots\otimes A[1]^{d_n}
=\bigoplus_{d_1+\cdots+d_n=d}A^{d_1+1}\otimes\cdots\otimes A^{d_n+1}. T ( A [ 1 ] ) d = d 1 + ⋯ + d n = d ⨁ A [ 1 ] d 1 ⊗ ⋯ ⊗ A [ 1 ] d n = d 1 + ⋯ + d n = d ⨁ A d 1 + 1 ⊗ ⋯ ⊗ A d n + 1 . It has a coproduct Δ : T ( A [ 1 ] ) → T ( A [ 1 ] ) ⊗ T ( A [ 1 ] ) \Delta:T(A[1])\to T(A[1])\otimes T(A[1]) Δ : T ( A [ 1 ]) → T ( A [ 1 ]) ⊗ T ( A [ 1 ]) given by
Δ ( a 1 ⊗ ⋯ ⊗ a n ) = ∑ k = 0 n ( a 1 ⊗ ⋯ ⊗ a k ) ⊗ ( a k + 1 ⊗ ⋯ ⊗ a n ) , \Delta(a_1\otimes\cdots\otimes a_n)=\sum_{k=0}^{n}(a_1\otimes\cdots\otimes a_k)\otimes(a_{k+1}\otimes\cdots\otimes a_n), Δ ( a 1 ⊗ ⋯ ⊗ a n ) = k = 0 ∑ n ( a 1 ⊗ ⋯ ⊗ a k ) ⊗ ( a k + 1 ⊗ ⋯ ⊗ a n ) , taking the k = 0 k=0 k = 0 and k = n k=n k = n terms to be 1 ⊗ ( a 1 ⊗ ⋯ ⊗ a n ) 1\otimes(a_1\otimes\cdots\otimes a_n) 1 ⊗ ( a 1 ⊗ ⋯ ⊗ a n ) and ( a 1 ⊗ ⋯ ⊗ a n ) ⊗ 1 (a_1\otimes\cdots\otimes a_n)\otimes1 ( a 1 ⊗ ⋯ ⊗ a n ) ⊗ 1 .
It has an increasing filtration T ( A [ 1 ] ) ≤ 0 ⊂ T ( A [ 1 ] ) ≤ 1 ⊂ T ( A [ 1 ] ) ≤ 2 ⊂ ⋯ T(A[1])_{\leq0} \subset T(A[1])_{\leq1} \subset T(A[1])_{\leq2} \subset\cdots T ( A [ 1 ] ) ≤ 0 ⊂ T ( A [ 1 ] ) ≤ 1 ⊂ T ( A [ 1 ] ) ≤ 2 ⊂ ⋯ , with T ( A [ 1 ] ) ≤ k = ⨁ n = 0 k A ⊗ n T(A[1])_{\leq k} = \bigoplus_{n=0}^{k} A^{\otimes n} T ( A [ 1 ] ) ≤ k = ⨁ n = 0 k A ⊗ n . The coproduct maps
Δ : T ( A [ 1 ] ) ≤ k ⟶ ∑ i + j = k T ( A [ 1 ] ) ≤ i ⊗ T ( A [ 1 ] ) ≤ j . \Delta:T(A[1])_{\leq k}\longrightarrow\sum_{i+j=k}T(A[1])_{\leq i}\otimes T(A[1])_{\leq j}. Δ : T ( A [ 1 ] ) ≤ k ⟶ i + j = k ∑ T ( A [ 1 ] ) ≤ i ⊗ T ( A [ 1 ] ) ≤ j .
Define R R R -linear maps m ‾ k : A [ 1 ] ⊗ k → A [ 1 ] \overline{m}_k:A[1]^{\otimes k}\to A[1] m k : A [ 1 ] ⊗ k → A [ 1 ] from m k m_k m k as in Definition 3.15. Define an R R R -linear map d ‾ : T ( A [ 1 ] ) → T ( A [ 1 ] ) \overline{d}:T(A[1])\to T(A[1]) d : T ( A [ 1 ]) → T ( A [ 1 ]) by, for all n ≥ 0 n\geq0 n ≥ 0
d ‾ ∣ A [ 1 ] ⊗ n = ∑ i , k ≥ 0 , j ≥ 1 i + j + k = n id A [ 1 ] ⊗ i ⊗ m ‾ j ⊗ id A [ 1 ] ⊗ k . (77) \left.\overline{d}\right|_{A[1]^{\otimes n}}=\sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}\operatorname{id}_{A[1]}^{\otimes i}\otimes\overline{m}_j\otimes\operatorname{id}_{A[1]}^{\otimes k}. \tag*{(77)} d A [ 1 ] ⊗ n = i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ id A [ 1 ] ⊗ i ⊗ m j ⊗ id A [ 1 ] ⊗ k . (77)
Then d ‾ \overline{d} d is a graded R R R -linear map of degree 1, and equation (3.16) is equivalent to d ‾ ∘ d ‾ = 0 \overline{d}\circ\overline{d}=0 d ∘ d = 0 , so that ( T ( A [ 1 ] ) , d ‾ ) (T(A[1]),\overline{d}) ( T ( A [ 1 ]) , d ) is a complex, the bar complex of ( A , m ) (A,m) ( A , m ) . Also d ‾ \overline{d} d is a derivation for the coproduct Δ \Delta Δ , that is,
Δ ∘ d ‾ = ( d ‾ ⊗ id T ( A [ 1 ] ) + id T ( A [ 1 ] ) ⊗ d ‾ ) ∘ Δ (78) \Delta\circ\overline{d}=(\overline{d}\otimes\operatorname{id}_{T(A[1])}+\operatorname{id}_{T(A[1])}\otimes\overline{d})\circ\Delta \tag*{(78)} Δ ∘ d = ( d ⊗ id T ( A [ 1 ]) + id T ( A [ 1 ]) ⊗ d ) ∘ Δ (78)
so that ( T ( A [ 1 ] ) , Δ , d ‾ ) (T(A[1]),\Delta,\overline{d}) ( T ( A [ 1 ]) , Δ , d ) is a differential graded coalgebra . And d ‾ \overline{d} d preserves the filtration, that is,
d ‾ ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( A [ 1 ] ) ≤ k . (79) \overline{d}(T(A[1])_{\leq k})\subseteq T(A[1])_{\leq k}. \tag*{(79)} d ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( A [ 1 ] ) ≤ k . (79)
Definition 3.17. Let f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) be a morphism of A ∞ A_\infty A ∞ -algebras over R R R , as in Definition 3.8. Use the notation of Definition 3.15. We can rewrite (3.7) in terms of the bar complexes of ( A , m ) (A,m) ( A , m ) and ( B , n ) (B,n) ( B , n ) : define an R R R -linear map f ‾ : T ( A [ 1 ] ) → T ( B [ 1 ] ) \overline{f}:T(A[1])\to T(B[1]) f : T ( A [ 1 ]) → T ( B [ 1 ]) by, for n > 0 n>0 n > 0
f ‾ ∣ A [ 1 ] ⊗ n = ∑ n = k 1 + ⋯ + k l , k a ≥ 1 f ‾ k 1 ⊗ f ‾ k 2 ⊗ ⋯ ⊗ f ‾ k l . (80) \left.\overline{f}\right|_{A[1]^{\otimes n}}=\sum_{\substack{n=k_1+\cdots+k_l,\ k_a\geq1}}\overline{f}_{k_1}\otimes\overline{f}_{k_2}\otimes\cdots\otimes\overline{f}_{k_l}. \tag*{(80)} f A [ 1 ] ⊗ n = n = k 1 + ⋯ + k l , k a ≥ 1 ∑ f k 1 ⊗ f k 2 ⊗ ⋯ ⊗ f k l . (80)
When n = 0 n=0 n = 0 , we set f ‾ ( λ ) = λ \overline{f}(\lambda)=\lambda f ( λ ) = λ for λ ∈ R = A ⊗ 0 = B ⊗ 0 \lambda\in R=A^{\otimes0}=B^{\otimes0} λ ∈ R = A ⊗ 0 = B ⊗ 0 . Then f ‾ \overline{f} f is graded of degree 0, and (3.17) is equivalent to
d ‾ B ∘ f ‾ = f ‾ ∘ d ‾ A : T ( A [ 1 ] ) ⟶ T ( B [ 1 ] ) , (81) \overline{d}_B\circ\overline{f}=\overline{f}\circ\overline{d}_A:T(A[1])\longrightarrow T(B[1]), \tag*{(81)} d B ∘ f = f ∘ d A : T ( A [ 1 ]) ⟶ T ( B [ 1 ]) , (81)
that is, f ‾ \overline{f} f is a morphism of bar complexes ( T ( A [ 1 ] ) , d ‾ A ) → ( T ( B [ 1 ] ) , d ‾ B ) (T(A[1]),\overline{d}_A)\to(T(B[1]),\overline{d}_B) ( T ( A [ 1 ]) , d A ) → ( T ( B [ 1 ]) , d B ) . It also intertwines the coproducts Δ A , Δ B \Delta_A,\Delta_B Δ A , Δ B on T ( A [ 1 ] ) , T ( B [ 1 ] ) T(A[1]),T(B[1]) T ( A [ 1 ]) , T ( B [ 1 ]) , that is
Δ B ∘ f ‾ = ( f ‾ ⊗ f ‾ ) ∘ Δ A , \Delta_B\circ\overline{f}=(\overline{f}\otimes\overline{f})\circ\Delta_A, Δ B ∘ f = ( f ⊗ f ) ∘ Δ A ,
and it preserves filtrations T ( A [ 1 ] ) ≤ k , T ( B [ 1 ] ) ≤ k T(A[1])_{\leq k},T(B[1])_{\leq k} T ( A [ 1 ] ) ≤ k , T ( B [ 1 ] ) ≤ k , namely
f ‾ ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( B [ 1 ] ) ≤ k . (82) \overline{f}(T(A[1])_{\leq k})\subseteq T(B[1])_{\leq k}. \tag*{(82)} f ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( B [ 1 ] ) ≤ k . (82)
For identity morphisms id A : ( A , m ) → ( A , m ) \operatorname{id}_A:(A,m)\to(A,m) id A : ( A , m ) → ( A , m ) we have id ‾ A = id T ( A [ 1 ] ) \overline{\operatorname{id}}_A=\operatorname{id}_{T(A[1])} id A = id T ( A [ 1 ]) on bar complexes. For morphisms f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) , g : ( B , n ) → ( C , o ) g:(B,n)\to(C,o) g : ( B , n ) → ( C , o ) , equation (3.10) implies that ( g ∘ f ) ‾ = g ‾ ∘ f ‾ \overline{(g\circ f)}=\overline{g}\circ\overline{f} ( g ∘ f ) = g ∘ f on bar complexes.
Definition 3.18. Let f , g : ( A , m ) → ( B , n ) f,g:(A,\mathfrak{m})\to(B,\mathfrak{n}) f , g : ( A , m ) → ( B , n ) be morphisms of A ∞ A_\infty A ∞ -algebras over R R R and H : f ⇒ g \mathcal{H}:f\Rightarrow g H : f ⇒ g be a homotopy, as in Definition 3.9. Use the notation of Definition 3.15. Define an R R R -linear map H ‾ : T ( A [ 1 ] ) → T ( B [ 1 ] ) \overline{\mathcal{H}}:T(A[1])\to T(B[1]) H : T ( A [ 1 ]) → T ( B [ 1 ]) by, for n ≥ 0 n\geq0 n ≥ 0
H ‾ ∣ A [ 1 ] ⊗ n = ∑ l , m ≥ 0 , i 1 , … , i l , j , k 1 , … , k m ≥ 1 : n = i 1 + ⋯ + i l + j + k 1 + ⋯ + k m f ‾ i 1 ⊗ ⋯ ⊗ f ‾ i l ⊗ H ‾ j ⊗ g ‾ k 1 ⊗ ⋯ ⊗ g ‾ k m . (83) \left.\overline{\mathcal{H}}\right|_{A[1]^{\otimes n}}=
\sum_{\substack{l,m\geq0,\ i_{1},\ldots,i_{l},j,k_{1},\ldots,k_{m}\geq1:\\ n=i_{1}+\cdots+i_{l}+j+k_{1}+\cdots+k_{m}}}
\overline{f}_{i_{1}}\otimes\cdots\otimes\overline{f}_{i_{l}}\otimes\overline{\mathcal{H}}_{j}\otimes\overline{g}_{k_{1}}\otimes\cdots\otimes\overline{g}_{k_{m}}. \tag*{(83)} H A [ 1 ] ⊗ n = l , m ≥ 0 , i 1 , … , i l , j , k 1 , … , k m ≥ 1 : n = i 1 + ⋯ + i l + j + k 1 + ⋯ + k m ∑ f i 1 ⊗ ⋯ ⊗ f i l ⊗ H j ⊗ g k 1 ⊗ ⋯ ⊗ g k m . (83) Then H ‾ \overline{\mathcal{H}} H is a graded R R R -linear map of degree − 1 -1 − 1 , and (3.12) is equivalent to
f ‾ − g ‾ = d ‾ B ∘ H ‾ + H ‾ ∘ d ‾ A , (84) \overline{f}-\overline{g}=\overline{d}_{B}\circ\overline{\mathcal{H}}+\overline{\mathcal{H}}\circ\overline{d}_{A}, \tag*{(84)} f − g = d B ∘ H + H ∘ d A , (84) so that f ‾ \overline{f} f and g ‾ \overline{g} g are homotopic as morphisms of chain complexes in the usual sense. Also H ‾ \overline{\mathcal{H}} H satisfies
I B ∘ H ‾ = ( f ‾ ⊗ H ‾ + H ‾ ⊗ g ‾ ) ∘ I A , (85) \mathbb{I}_{B}\circ\overline{\mathcal{H}}=(\overline{f}\otimes\overline{\mathcal{H}}+\overline{\mathcal{H}}\otimes\overline{g})\circ\mathbb{I}_{A}, \tag*{(85)} I B ∘ H = ( f ⊗ H + H ⊗ g ) ∘ I A , (85) and H ‾ \overline{\mathcal{H}} H preserves filtrations, that is
H ‾ ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( B [ 1 ] ) ≤ k . (86) \overline{\mathcal{H}}\left(T(A[1])_{\leq k}\right)\subseteq T(B[1])_{\leq k}. \tag*{(86)} H ( T ( A [ 1 ] ) ≤ k ) ⊆ T ( B [ 1 ] ) ≤ k . (86)
These are equivalent to the previous definitions, in the following sense:
Proposition 3.19. (a) Let R R R be a commutative ring and A A A be a Z \mathbb{Z} Z -graded R R R -module. Then A ∞ A_\infty A ∞ -algebra structures ( A , m ) (A,\mathfrak{m}) ( A , m ) on A A A are equivalent under (3.18) to R R R -linear differentials d ‾ : T ( A [ 1 ] ) → T ( A [ 1 ] ) \overline{d}:T(A[1])\to T(A[1]) d : T ( A [ 1 ]) → T ( A [ 1 ]) which are graded of degree + 1 +1 + 1 and satisfy d ‾ ∘ d ‾ = 0 \overline{d}\circ\overline{d}=0 d ∘ d = 0 and equations (3.19)–(3.20).
(b) Let ( A , m ) , ( B , n ) (A,\mathfrak{m}),(B,\mathfrak{n}) ( A , m ) , ( B , n ) be A ∞ A_\infty A ∞ -algebras over R R R . Then morphisms f : ( A , m ) → ( B , n ) f:(A,\mathfrak{m})\to(B,\mathfrak{n}) f : ( A , m ) → ( B , n ) are equivalent under (3.21) to R R R -linear maps f ‾ : T ( A [ 1 ] ) → T ( B [ 1 ] ) \overline{f}:T(A[1])\to T(B[1]) f : T ( A [ 1 ]) → T ( B [ 1 ]) which are graded of degree 0 0 0 and satisfy (3.22)–(3.23).
(c) Let f , g : ( A , m ) → ( B , n ) f,g:(A,\mathfrak{m})\to(B,\mathfrak{n}) f , g : ( A , m ) → ( B , n ) be morphisms of A ∞ A_\infty A ∞ -algebras over R R R . Then homotopies H : f ⇒ g \mathcal{H}:f\Rightarrow g H : f ⇒ g are equivalent under (3.24) to R R R -linear maps H ‾ : T ( A [ 1 ] ) → T ( B [ 1 ] ) \overline{\mathcal{H}}:T(A[1])\to T(B[1]) H : T ( A [ 1 ]) → T ( B [ 1 ]) which are graded of degree − 1 -1 − 1 and satisfy (3.25)–(3.27).
Mazuir and Cazassus' work on ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras
As part of a project to study the higher structure of Morse theory of a compact manifold [68, 69], Mazuir [69, §2.4] discusses several attempts to construct an ∞ \infty ∞ -category of A ∞ A_\infty A ∞ -algebras, but he does not succeed in this goal.
If ( A , m ) (A,\mathfrak{m}) ( A , m ) and ( B , n ) (B,\mathfrak{n}) ( B , n ) are A ∞ A_\infty A ∞ -algebras over R R R , Mazuir [69, Th. 1] defines a Kan complex Hom ( ( A , m ) , ( B , n ) ) \operatorname{Hom}((A,\mathfrak{m}),(B,\mathfrak{n})) Hom (( A , m ) , ( B , n )) , which may be related to mapping spaces Hom L ( ( A , m ) , ( B , n ) ) \operatorname{Hom}^{L}((A,\mathfrak{m}),(B,\mathfrak{n})) Hom L (( A , m ) , ( B , n )) in our ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras in Chapter 6.
Cazassus [15, §5.2] defines an ∞ \infty ∞ -category f - A l g R \mathrm{f\text{-}Alg}_{R} f - Alg R of A ∞ A_{\infty} A ∞ -algebras over R R R . Part of his aim is to understand the sense in which Morse (co)homology of smooth manifolds X , Y X,Y X , Y is functorial under smooth maps ϕ : X → Y \phi:X\to Y ϕ : X → Y . Singular (co)homology, for example, is functorial in the obvious sense on (co)chains. But to define the Morse cochain complex C M ∙ ( X , R ) CM^\bullet(X,R) C M ∙ ( X , R ) of X X X (which can be given the structure of an A ∞ A_{\infty} A ∞ -algebra) one must choose a Morse function f : X → R f:X\to\mathbb{R} f : X → R and a Palais–Smale pseudo-gradient v v v for f f f (and more choices for the A ∞ A_{\infty} A ∞ -structure), and these choices cannot be made functorial at the 1-categorical level.
Cazassus defines an ∞ \infty ∞ -functor C M ∙ : N ( M a n o p ) → f - A l g R CM^\bullet:\mathcal{N}(\mathrm{Man}^{\mathrm{op}})\to\mathrm{f\text{-}Alg}_{R} C M ∙ : N ( Man op ) → f - Alg R which restores the functoriality at the ∞ \infty ∞ -categorical level. Here C M ∙ CM^\bullet C M ∙ makes an arbitrary choice of f , v f,v f , v as above on each 0-simplex X X X of N ( M a n o p ) \mathcal{N}(\mathrm{Man}^{\mathrm{op}}) N ( Man op ) , and on N N N -simplices for N ≥ 1 N\geq1 N ≥ 1 makes more complicated choices of Morse-type data. It is necessary to define an ∞ \infty ∞ -category f - A l g R \mathrm{f\text{-}Alg}_{R} f - Alg R of A ∞ A_{\infty} A ∞ -algebras as the target of this functor.
Cazassus also defines ∞ \infty ∞ -categories f - C o a l g R \mathrm{f\text{-}Coalg}_{R} f - Coalg R of ‘A ∞ A_{\infty} A ∞ -coalgebras’, which are the target of an ∞ \infty ∞ -functor for Morse homology, and f - B i a l g R \mathrm{f\text{-}Bialg}_{R} f - Bialg R of ‘A ∞ A_{\infty} A ∞ -bialgebras’, which are the target of an ∞ \infty ∞ -functor for Morse cohomology of Lie groups. All these are intended for eventual applications in (equivariant) Lagrangian Floer cohomology and (equivariant) Fukaya categories, since Floer theories are (morally) infinite-dimensional Morse theories.
To define f - A l g R \mathrm{f\text{-}Alg}_{R} f - Alg R , Cazassus defines an R R R -linear category ( R , u ) (R,u) ( R , u ) -mod of ascending foresty R R R -modules, or ( R , u ) (R,u) ( R , u ) -modules, with objects graded R R R -modules A , B A,B A , B and morphisms certain collections of graded R R R -linear maps A ⊗ k → B [ l ] A^{\otimes k}\to B[l] A ⊗ k → B [ l ] . He forms the dg-category C h ( R , u ) - m o d \mathrm{Ch}_{(R,u)\text{-}\mathrm{mod}} Ch ( R , u ) - mod of complexes of ( R , u ) (R,u) ( R , u ) -modules. As in Lurie [61, §1.3.1] this has a dg-nerve N d g ( C h ( R , u ) - m o d ) \mathcal{N}_{\mathrm{dg}}(\mathrm{Ch}_{(R,u)\text{-}\mathrm{mod}}) N dg ( Ch ( R , u ) - mod ) , which is an ∞ \infty ∞ -category.
Cazassus defines a faithful (but not full) inclusion functor I : A ∞ A l g R ↪ C h ( R , u ) - m o d I:A_{\infty}\mathrm{Alg}_{R}\hookrightarrow\mathrm{Ch}_{(R,u)\text{-}\mathrm{mod}} I : A ∞ Alg R ↪ Ch ( R , u ) - mod . Then he sets f - A l g R \mathrm{f\text{-}Alg}_{R} f - Alg R to be the ∞ \infty ∞ -subcategory of N d g ( C h ( R , u ) - m o d ) \mathcal{N}_{\mathrm{dg}}(\mathrm{Ch}_{(R,u)\text{-}\mathrm{mod}}) N dg ( Ch ( R , u ) - mod ) whose 0- and 1-simplices lie in the image of I I I , but whose N N N -simplices for N ≥ 2 N\geq2 N ≥ 2 are all N N N -simplices in N d g ( C h ( R , u ) - m o d ) \mathcal{N}_{\mathrm{dg}}(\mathrm{Ch}_{(R,u)\text{-}\mathrm{mod}}) N dg ( Ch ( R , u ) - mod ) with these 0- and 1-simplices.
The classical theory of A ∞ A_{\infty} A ∞ -categories
The definition of A ∞ A_{\infty} A ∞ -categories, and units
Some references on A ∞ A_{\infty} A ∞ -categories are Fukaya [29, §1] and Seidel [84].
Definition 4.1. Let R R R be a commutative ring. An A ∞ A_{\infty} A ∞ -category A \mathcal{A} A over R R R consists of:
(a) A set, or proper class, Obj A \operatorname{Obj}\mathcal{A} Obj A of objects of A \mathcal{A} A .
(b) For all objects x , y ∈ Obj A x,y\in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , a Z \mathbb{Z} Z -graded R R R -module
Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) . \operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y). Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) . We also write Hom A ( x , y ) \operatorname{Hom}_{\mathcal{A}}(x,y) Hom A ( x , y ) when we want to identify the A ∞ A_{\infty} A ∞ -category A \mathcal{A} A .
(c) For all k ≥ 1 k\geq1 k ≥ 1 and objects x 0 , x 1 , … , x k x_{0},x_{1},\ldots,x_{k} x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
m k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) m_{k}:\operatorname{Hom}(x_{0},x_{1})\otimes_{R}\operatorname{Hom}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\operatorname{Hom}(x_{k-1},x_{k})\longrightarrow\operatorname{Hom}(x_{0},x_{k}) m k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k )
of degree 2 − k 2-k 2 − k . We also write m k A m_k^{\mathcal{A}} m k A , or m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k , or m k A ; x 0 , … , x k m_k^{\mathcal{A};x_0,\ldots,x_k} m k A ; x 0 , … , x k for m k m_k m k when we want to identify x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k or A \mathcal{A} A . We write m = ( m k ) k ≥ 1 \mathbf{m}=(m_k)_{k\geq1} m = ( m k ) k ≥ 1 .
Generalizing (3.3), we require that for all n ≥ 1 n\geq1 n ≥ 1 and objects x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (87) \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}\circ\left(\operatorname{id}_{\operatorname{Hom}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{i-1},x_i)}\otimes m_j^{x_i,\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{n-1},x_n)}\right)=0. \tag*{(87)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (87)
By abuse of notation, we also regard m k m_k m k in (c) as a graded R R R -multilinear map
m k : Hom ( x 0 , x 1 ) × Hom ( x 1 , x 2 ) × ⋯ × Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) . m_k:\operatorname{Hom}(x_0,x_1)\times\operatorname{Hom}(x_1,x_2)\times\cdots\times\operatorname{Hom}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}(x_0,x_k). m k : Hom ( x 0 , x 1 ) × Hom ( x 1 , x 2 ) × ⋯ × Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) .
Then generalizing (3.4), by Convention 3.1, equation (4.1) is equivalent to the equation for all n ≥ 1 n\geq1 n ≥ 1 and all pure a i ∈ Hom ( x i − 1 , x i ) a_i\in\operatorname{Hom}(x_{i-1},x_i) a i ∈ Hom ( x i − 1 , x i ) for $i=1,\ldots,n
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k + j ∑ l = 1 i deg a l m i + k + 1 ( a 1 , … , a i , m j ( a i + 1 , … , a i + j ) , a i + j + 1 , … , a n ) = 0. (88) \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk+j\sum_{l=1}^{i}\deg a_l}m_{i+k+1}(a_1,\ldots,a_i,m_j(a_{i+1},\ldots,a_{i+j}),a_{i+j+1},\ldots,a_n)=0. \tag*{(88)} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k + j ∑ l = 1 i d e g a l m i + k + 1 ( a 1 , … , a i , m j ( a i + 1 , … , a i + j ) , a i + j + 1 , … , a n ) = 0. (88)
An A ∞ A_\infty A ∞ -category A \mathcal{A} A for which Obj A \operatorname{Obj}\mathcal{A} Obj A is a set (rather than a proper class) is called a small A ∞ A_\infty A ∞ -category.
Remark 4.2. (On Set Theory foundations.) As in Remark 1.23, to study A ∞ A_\infty A ∞ -categories A \mathcal{A} A using ∞ \infty ∞ -categories, we will need Obj A \operatorname{Obj}\mathcal{A} Obj A to be a set, so A \mathcal{A} A is a small A ∞ A_\infty A ∞ -category. In this book we fix a Grothendieck universe U \mathcal{U} U , and assume implicitly from now on that Obj A \operatorname{Obj}\mathcal{A} Obj A is a subset of U \mathcal{U} U .
Remark 4.3. (Relations between A ∞ A_\infty A ∞ -categories and A ∞ A_\infty A ∞ -algebras.) If Obj A = { ∗ } \operatorname{Obj}\mathcal{A}=\{\ast\} Obj A = { ∗ } is a single point then Definition 4.1 reduces to Definition 3.2. That is, A ∞ A_\infty A ∞ -algebras are essentially the same as A ∞ A_\infty A ∞ -categories with one object.
Remark 4.4. (Other remarks.) (a) Note that A ∞ A_\infty A ∞ -categories are not actually categories, as by (4.1) the composition maps
m 2 x , y , z : Hom ( x , y ) × Hom ( y , z ) ⟶ Hom ( x , z ) m_2^{x,y,z}:\operatorname{Hom}(x,y)\times\operatorname{Hom}(y,z)\longrightarrow\operatorname{Hom}(x,z) m 2 x , y , z : Hom ( x , y ) × Hom ( y , z ) ⟶ Hom ( x , z )
need not be associative, and also we do not require the existence of identities (units). In Definition 4.7 we will show that a (cohomologically) unital A ∞ A_\infty A ∞ -category A \mathcal{A} A has a cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) which is an honest category.
(b) A ∞ A_\infty A ∞ -categories are important in Symplectic Geometry: Fukaya categories of symplectic manifolds are initially defined as A ∞ A_\infty A ∞ -categories [29, 84]. They also have applications elsewhere. For example, dg-categories are special cases of A ∞ A_\infty A ∞ -categories, and to study the homotopy theory of dg-categories it is helpful to regard them as A ∞ A_\infty A ∞ -categories, with A ∞ A_\infty A ∞ -functors between them.
If A \mathcal{A} A is an A ∞ A_\infty A ∞ -category then (4.1) for n = 1 n=1 n = 1 becomes m 1 x 0 , x 1 ∘ m 1 x 0 , x 1 = 0 m_1^{x_0,x_1}\circ m_1^{x_0,x_1}=0 m 1 x 0 , x 1 ∘ m 1 x 0 , x 1 = 0 , with m 1 x 0 , x 1 m_1^{x_0,x_1} m 1 x 0 , x 1 of degree 1. Thus ( Hom ( x 0 , x 1 ) , m 1 x 0 , x 1 ) (\operatorname{Hom}(x_0,x_1),m_1^{x_0,x_1}) ( Hom ( x 0 , x 1 ) , m 1 x 0 , x 1 ) is a complex, and we form its cohomology H ∗ ( Hom ( x 0 , x 1 ) ) H^*(\operatorname{Hom}(x_0,x_1)) H ∗ ( Hom ( x 0 , x 1 )) by
H p ( Hom ( x 0 , x 1 ) ) = Ker ( m 1 x 0 , x 1 : Hom p ( x 0 , x 1 ) ⟶ Hom p + 1 ( x 0 , x 1 ) ) Im ( m 1 x 0 , x 1 : Hom p − 1 ( x 0 , x 1 ) ⟶ Hom p ( x 0 , x 1 ) ) . (89) H^p(\operatorname{Hom}(x_0,x_1))=\frac{\operatorname{Ker}\left(m_1^{x_0,x_1}:\operatorname{Hom}^p(x_0,x_1)\longrightarrow\operatorname{Hom}^{p+1}(x_0,x_1)\right)}{\operatorname{Im}\left(m_1^{x_0,x_1}:\operatorname{Hom}^{p-1}(x_0,x_1)\longrightarrow\operatorname{Hom}^p(x_0,x_1)\right)}. \tag*{(89)} H p ( Hom ( x 0 , x 1 )) = Im ( m 1 x 0 , x 1 : Hom p − 1 ( x 0 , x 1 ) ⟶ Hom p ( x 0 , x 1 ) ) Ker ( m 1 x 0 , x 1 : Hom p ( x 0 , x 1 ) ⟶ Hom p + 1 ( x 0 , x 1 ) ) . (89)
As in (64) , for all objects x 0 , x 1 , x 2 x_0,x_1,x_2 x 0 , x 1 , x 2 in A \mathcal{A} A we have a well-defined, associative R R R -bilinear product, which we call composition ∘ \circ ∘
∘ : H p ( Hom ( x 0 , x 1 ) ) × H q ( Hom ( x 1 , x 2 ) ) ⟶ H p + q ( Hom ( x 0 , x 2 ) ) (90) \circ: H^p(\operatorname{Hom}(x_0,x_1)) \times H^q(\operatorname{Hom}(x_1,x_2)) \longrightarrow H^{p+q}(\operatorname{Hom}(x_0,x_2)) \tag*{(90)} ∘ : H p ( Hom ( x 0 , x 1 )) × H q ( Hom ( x 1 , x 2 )) ⟶ H p + q ( Hom ( x 0 , x 2 )) (90)
given by
( a 1 + Im m 1 x 0 , x 1 ) ∘ ( a 2 + Im m 1 x 1 , x 2 ) = m 2 x 0 , x 1 , x 2 ( a 1 , a 2 ) + Im m 1 x 0 , x 2 . (a_1+\operatorname{Im}m_1^{x_0,x_1})\circ(a_2+\operatorname{Im}m_1^{x_1,x_2})=m_2^{x_0,x_1,x_2}(a_1,a_2)+\operatorname{Im}m_1^{x_0,x_2}. ( a 1 + Im m 1 x 0 , x 1 ) ∘ ( a 2 + Im m 1 x 1 , x 2 ) = m 2 x 0 , x 1 , x 2 ( a 1 , a 2 ) + Im m 1 x 0 , x 2 .
Here are the analogues of Definitions 3.4–3.6.
Definition 4.5. Let A \mathcal{A} A be an A ∞ A_\infty A ∞ -category. We say that A \mathcal{A} A is unital if for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A there exists 1 x ∈ Hom 0 ( x , x ) 1_x\in\operatorname{Hom}^0(x,x) 1 x ∈ Hom 0 ( x , x ) , called the identity for x x x , satisfying
m 2 x 0 , x 0 , x 1 ( 1 x 0 ⊗ a ) = m 2 x 0 , x 1 , x 1 ( a ⊗ 1 x 1 ) = a (91) m_2^{x_0,x_0,x_1}(1_{x_0}\otimes a)=m_2^{x_0,x_1,x_1}(a\otimes1_{x_1})=a \tag*{(91)} m 2 x 0 , x 0 , x 1 ( 1 x 0 ⊗ a ) = m 2 x 0 , x 1 , x 1 ( a ⊗ 1 x 1 ) = a (91) for all objects x 0 , x 1 x_0,x_1 x 0 , x 1 in A \mathcal{A} A and a ∈ Hom ( x 0 , x 1 ) a\in\operatorname{Hom}(x_0,x_1) a ∈ Hom ( x 0 , x 1 ) , and
m k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (92) m_{k+1}^{x_0,\ldots,x_i,x_i,\ldots,x_k}(a_1\otimes\cdots\otimes a_i\otimes1_{x_i}\otimes a_{i+1}\otimes\cdots\otimes a_k)=0 \tag*{(92)} m k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (92) for all k ≠ 1 k\ne1 k = 1 , i = 0 , … , k i=0,\ldots,k i = 0 , … , k , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and a j ∈ Hom ( x j − 1 , x j ) a_j\in\operatorname{Hom}(x_{j-1},x_j) a j ∈ Hom ( x j − 1 , x j ) . If 1 x , 1 x ′ 1_x,1'_x 1 x , 1 x ′ are alternative choices for 1 x 1_x 1 x then (4.5) gives 1 x = m 2 x , x , x ( 1 x , 1 x ′ ) = 1 x ′ 1_x=m_2^{x,x,x}(1_x,1'_x)=1'_x 1 x = m 2 x , x , x ( 1 x , 1 x ′ ) = 1 x ′ , so identities are unique if they exist, and to be unital is a property of A \mathcal{A} A , not extra data.
Definition 4.6. Let A \mathcal{A} A be an A ∞ A_\infty A ∞ -category over a commutative ring R R R . We call A \mathcal{A} A weakly cohomologically unital if for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A there exists 1 x 1_x 1 x in H 0 ( Hom ( x , x ) ) H^0(\operatorname{Hom}(x,x)) H 0 ( Hom ( x , x )) , called the cohomological identity for x x x , satisfying
1 x 0 ∘ α = α ∘ 1 x 1 = α (93) 1_{x_0}\circ\alpha=\alpha\circ1_{x_1}=\alpha \tag*{(93)} 1 x 0 ∘ α = α ∘ 1 x 1 = α (93) for all objects x 0 , x 1 x_0,x_1 x 0 , x 1 in A \mathcal{A} A and α ∈ H ∗ ( Hom ( x 0 , x 1 ) ) \alpha\in H^*(\operatorname{Hom}(x_0,x_1)) α ∈ H ∗ ( Hom ( x 0 , x 1 )) , where ∘ \circ ∘ is as in (4.4). Again, if 1 x , 1 x ′ 1_x,1'_x 1 x , 1 x ′ are alternative choices for 1 x 1_x 1 x then (4.7) gives 1 x = 1 x ∘ 1 x ′ = 1 x ′ 1_x=1_x\circ1'_x=1'_x 1 x = 1 x ∘ 1 x ′ = 1 x ′ , so cohomological identities are unique if they exist, and to be cohomologically unital is a property of A \mathcal{A} A , not extra data.
Let A \mathcal{A} A be weakly cohomologically unital. The cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) has objects x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A , morphisms Hom H 0 ( A ) ( x , y ) = H 0 ( Hom A ( x , y ) ) \operatorname{Hom}_{H^0(\mathcal{A})}(x,y)=H^0(\operatorname{Hom}_{\mathcal{A}}(x,y)) Hom H 0 ( A ) ( x , y ) = H 0 ( Hom A ( x , y )) for x , y ∈ Obj A x,y\in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , composition of morphisms ∘ : H 0 ( Hom ( x , y ) ) × H 0 ( Hom ( y , z ) ) ⟶ H 0 ( Hom ( x , z ) ) \circ:H^0(\operatorname{Hom}(x,y))\times H^0(\operatorname{Hom}(y,z))\longrightarrow H^0(\operatorname{Hom}(x,z)) ∘ : H 0 ( Hom ( x , y )) × H 0 ( Hom ( y , z )) ⟶ H 0 ( Hom ( x , z )) as in (4.4), and identities 1 x ∈ H 0 ( Hom ( x , x ) ) 1_x\in H^0(\operatorname{Hom}(x,x)) 1 x ∈ H 0 ( Hom ( x , x )) . Then H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) is an ordinary R R R -linear category.
Definition 4.7. Let A \mathcal{A} A be an A ∞ A_\infty A ∞ -category over a commutative ring R R R , and x x x an object in A \mathcal{A} A . A cohomological unit for x x x is an element e x e_x e x in Hom A 0 ( x , x ) \operatorname{Hom}^0_{\mathcal{A}}(x,x) Hom A 0 ( x , x ) such that m 1 x , x ( e x ) = 0 m_1^{x,x}(e_x)=0 m 1 x , x ( e x ) = 0 , and for all objects y y y in A \mathcal{A} A the following maps of chain complexes are chain homotopic to the identity:
Φ : ( Hom A ( x , y ) , m 1 x , y ) ⟶ ( Hom A ( x , y ) , m 1 x , y ) , Φ : a ⟼ m 2 x , y ( e x ⊗ a ) , \Phi:(\operatorname{Hom}_{\mathcal{A}}(x,y),m_1^{x,y})\longrightarrow(\operatorname{Hom}_{\mathcal{A}}(x,y),m_1^{x,y}),\qquad\Phi:a\longmapsto m_2^{x,y}(e_x\otimes a), Φ : ( Hom A ( x , y ) , m 1 x , y ) ⟶ ( Hom A ( x , y ) , m 1 x , y ) , Φ : a ⟼ m 2 x , y ( e x ⊗ a ) , Ψ : ( Hom A ( y , x ) , m 1 y , x ) ⟶ ( Hom A ( y , x ) , m 1 y , x ) , Ψ : a ⟼ m 2 y , x ( a ⊗ e x ) . \Psi:(\operatorname{Hom}_{\mathcal{A}}(y,x),m_1^{y,x})\longrightarrow(\operatorname{Hom}_{\mathcal{A}}(y,x),m_1^{y,x}),\qquad\Psi:a\longmapsto m_2^{y,x}(a\otimes e_x). Ψ : ( Hom A ( y , x ) , m 1 y , x ) ⟶ ( Hom A ( y , x ) , m 1 y , x ) , Ψ : a ⟼ m 2 y , x ( a ⊗ e x ) .
Then the cohomology class 1 x = [ e x ] 1_x=[e_x] 1 x = [ e x ] in H 0 ( Hom ( x , x ) ) H^0(\operatorname{Hom}(x,x)) H 0 ( Hom ( x , x )) satisfies (93) .
If e x ′ ∈ Hom A 0 ( x , x ) e'_x\in\operatorname{Hom}^0_{\mathcal{A}}(x,x) e x ′ ∈ Hom A 0 ( x , x ) is any other representative for 1 x 1_x 1 x in H 0 ( Hom ( x , x ) ) H^0(\operatorname{Hom}(x,x)) H 0 ( Hom ( x , x )) , then e x ′ = e x + m 1 x , x ( f ) e'_x=e_x+m^{x,x}_1(f) e x ′ = e x + m 1 x , x ( f ) for f ∈ Hom A − 1 ( x , x ) f\in\operatorname{Hom}^{-1}_{\mathcal{A}}(x,x) f ∈ Hom A − 1 ( x , x ) , and defining Θ , Ξ \Theta,\Xi Θ , Ξ by Θ ( a ) = m 2 x , y ( f ⊗ a ) \Theta(a)=m^{x,y}_2(f\otimes a) Θ ( a ) = m 2 x , y ( f ⊗ a ) , Ξ ( a ) = − m 2 y , x ( a ⊗ f ) \Xi(a)=-m^{y,x}_2(a\otimes f) Ξ ( a ) = − m 2 y , x ( a ⊗ f ) , we see from (3.4) that
Φ ′ ( a ) = Φ ( a ) + m 1 x , y ∘ Θ ( a ) − Θ ∘ m 1 x , y ( a ) , Ψ ′ ( a ) = Ψ ( a ) + m 1 y , x ∘ Ξ ( a ) − Ξ ∘ m 1 y , x ( a ) . \Phi'(a)=\Phi(a)+m^{x,y}_1\circ\Theta(a)-\Theta\circ m^{x,y}_1(a),\qquad\Psi'(a)=\Psi(a)+m^{y,x}_1\circ\Xi(a)-\Xi\circ m^{y,x}_1(a). Φ ′ ( a ) = Φ ( a ) + m 1 x , y ∘ Θ ( a ) − Θ ∘ m 1 x , y ( a ) , Ψ ′ ( a ) = Ψ ( a ) + m 1 y , x ∘ Ξ ( a ) − Ξ ∘ m 1 y , x ( a ) .
Thus, Φ ′ , Ψ ′ \Phi',\Psi' Φ ′ , Ψ ′ are chain homotopic to the identity if and only if Φ , Ψ \Phi,\Psi Φ , Ψ are, and e x ′ e'_x e x ′ is a cohomological unit if and only if e x e_x e x is. Thus, for x x x to have a cohomologically unit is a property of A \mathcal{A} A , not of the particular choice of element e x e_x e x .
We call A \mathcal{A} A cohomologically unital if every x ∈ A x\in\mathcal{A} x ∈ A has a cohomological unit.
Clearly, A \mathcal{A} A unital implies A \mathcal{A} A cohomologically unital, implies A \mathcal{A} A weakly cohomologically unital. If R R R is a field then weakly cohomologically unital and cohomologically unital coincide, since Φ , Ψ \Phi,\Psi Φ , Ψ are chain homotopic to the identity if and only if they induce the identity maps on cohomology.
Readers are warned that the names used for different kinds of unitality in A ∞ A_\infty A ∞ -categories vary widely. Our notion of unital is often called ‘strictly unital’. Our notion of cohomologically unital is sometimes just called ‘unital’.
The next definition follows Fukaya [29, Def. 6.22].
Definition 4.8. Suppose A \mathcal{A} A is a unital or cohomologically unital A ∞ A_\infty A ∞ -category over R R R , and x , y x,y x , y are objects in A \mathcal{A} A , and f ∈ Hom 0 ( x , y ) f\in\operatorname{Hom}^0(x,y) f ∈ Hom 0 ( x , y ) with m A 1 ( f ) = 0 m^1_{\mathcal{A}}(f)=0 m A 1 ( f ) = 0 , so that f f f defines a morphism [ f ] : x → y [f]:x\to y [ f ] : x → y in the cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) . We call f f f a homotopy equivalence if [ f ] [f] [ f ] is an isomorphism in H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) . We say that x , y x,y x , y are homotopy equivalent if there exists a homotopy equivalence between them. Homotopy equivalence is an equivalence relation on objects in A \mathcal{A} A .
Functors between A ∞ A_\infty A ∞ -categories
Here is the analogue of Definition 3.8.
Definition 4.9. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_\infty A ∞ -categories over a commutative ring R R R . An A ∞ A_\infty A ∞ -functor, or just functor , F ~ : A → B \widetilde{F}:\mathcal{A}\to\mathcal{B} F : A → B consists of:
(a) A map F ~ O b j : Obj A → Obj B \widetilde{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{A}\to\operatorname{Obj}\mathcal{B} F Obj : Obj A → Obj B . When this will not cause confusion, for x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A we write F ~ ( x ) ∈ Obj B \widetilde{F}(x)\in\operatorname{Obj}\mathcal{B} F ( x ) ∈ Obj B rather than F ~ O b j ( x ) \widetilde{F}_{\mathrm{Obj}}(x) F Obj ( x ) .
(b) For all k ≥ 1 k\ge1 k ≥ 1 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
F ~ k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F ~ O b j ( x 0 ) , F ~ O b j ( x k ) ) , \widetilde{F}_k:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\widetilde{F}_{\mathrm{Obj}}(x_0),\widetilde{F}_{\mathrm{Obj}}(x_k)), F k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , F Obj ( x k )) ,
of degree 1 − k 1-k 1 − k . We also write F ~ k x 0 , … , x k \widetilde{F}^{x_0,\ldots,x_k}_k F k x 0 , … , x k for F ~ k \widetilde{F}_k F k when we want to identify x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k . Generalizing (3.7), these must satisfy for all n ≥ 1 n\ge1 n ≥ 1 and objects x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n in Obj A \operatorname{Obj}\mathcal{A} Obj A
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) = ∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ( F k Obj ( x 0 ) , F Obj ( x n 1 ) , … , F Obj ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) ) . (94) \begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk}\mathcal{F}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}_{i+k+1}\circ\left(\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_i)}\right.\\
&\left.\qquad\otimes m^{\mathcal{A};x_i,\ldots,x_{i+j}}_j\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{n-1},x_n)}\right)\\
&=\sum_{\substack{n=n_1+\cdots+n_k\\n_a\geq1}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}m^{\mathcal{B}}_k\left(\mathcal{F}^{\operatorname{Obj}(x_0),\mathcal{F}\operatorname{Obj}(x_{n_1}),\ldots,\mathcal{F}\operatorname{Obj}(x_n)}_k\left(\mathcal{F}^{x_0,\ldots,x_{n_1}}_{n_1}\otimes\mathcal{F}^{x_{n_1},\ldots,x_{n_1+n_2}}_{n_2}\otimes\cdots\otimes\mathcal{F}^{x_{n-n_k},\ldots,x_n}_{n_k}\right)\right). \tag*{(94)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) = n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ( F k Obj ( x 0 ) , F Obj ( x n 1 ) , … , F Obj ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) ) . (94)
The identity functor id A : A → A \operatorname{id}_{\mathcal{A}}:\mathcal{A}\to\mathcal{A} id A : A → A has id Obj A = id Obj A \operatorname{id}_{\operatorname{Obj}}^{\mathcal{A}}=\operatorname{id}_{\operatorname{Obj}_{\mathcal{A}}} id Obj A = id Obj A , and id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) \operatorname{id}^{\mathcal{A},x}_{1}=\operatorname{id}:\operatorname{Hom}_{\mathcal{A}}(x,x)\to\operatorname{Hom}_{\mathcal{A}}(x,x) id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) , and id k A = 0 \operatorname{id}^{\mathcal{A}}_k=0 id k A = 0 for k > 1 k>1 k > 1 .
When n = 1 n=1 n = 1 , equation (94) becomes
F 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ F 1 x 0 , x 1 : \mathcal{F}^{x_0,x_1}_1\circ m^{\mathcal{A};x_0,x_1}_1=m^{\mathcal{B};\mathcal{F}(x_0),\mathcal{F}(x_1)}_1\circ\mathcal{F}^{x_0,x_1}_1: F 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ F 1 x 0 , x 1 :
Hom A ( x 0 , x 1 ) ⟶ Hom B ( F ( x 0 ) , F ( x 1 ) ) . \operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_1)). Hom A ( x 0 , x 1 ) ⟶ Hom B ( F ( x 0 ) , F ( x 1 )) .
Thus we have a morphism of complexes
F 1 x 0 , x 1 : ( Hom A ( x 0 , x 1 ) , m 1 A ; x 0 , x 1 ) ⟶ ( Hom B ( F ( x 0 ) , F ( x 1 ) ) , m 1 B ; F ( x 0 ) , F ( x 1 ) ) , \mathcal{F}^{x_0,x_1}_1:\left(\operatorname{Hom}_{\mathcal{A}}(x_0,x_1),m^{\mathcal{A};x_0,x_1}_1\right)\longrightarrow\left(\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_1)),m^{\mathcal{B};\mathcal{F}(x_0),\mathcal{F}(x_1)}_1\right), F 1 x 0 , x 1 : ( Hom A ( x 0 , x 1 ) , m 1 A ; x 0 , x 1 ) ⟶ ( Hom B ( F ( x 0 ) , F ( x 1 )) , m 1 B ; F ( x 0 ) , F ( x 1 ) ) ,
which induces a morphism of cohomology groups
H ∗ ( F 1 x 0 , x 1 ) : H ∗ ( Hom A ( x 0 , x 1 ) ) ⟶ H ∗ ( Hom B ( F ( x 0 ) , F ( x 1 ) ) ) . (95) H^*(\mathcal{F}^{x_0,x_1}_1):H^*\left(\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\right)\longrightarrow H^*\left(\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_1))\right). \tag*{(95)} H ∗ ( F 1 x 0 , x 1 ) : H ∗ ( Hom A ( x 0 , x 1 ) ) ⟶ H ∗ ( Hom B ( F ( x 0 ) , F ( x 1 )) ) . (95)
If F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , G : B → C \mathcal{G}:\mathcal{B}\to\mathcal{C} G : B → C are functors of A ∞ A_\infty A ∞ -categories over R R R , the composition G ∘ F : A → C \mathcal{G}\circ\mathcal{F}:\mathcal{A}\to\mathcal{C} G ∘ F : A → C is given by ( G ∘ F ) Obj = G Obj ∘ F Obj (\mathcal{G}\circ\mathcal{F})_{\operatorname{Obj}}=\mathcal{G}_{\operatorname{Obj}}\circ\mathcal{F}_{\operatorname{Obj}} ( G ∘ F ) Obj = G Obj ∘ F Obj and
( G ∘ F ) n x 0 , … , x n = (96) (\mathcal{G}\circ\mathcal{F})^{x_0,\ldots,x_n}_n= \tag*{(96)} ( G ∘ F ) n x 0 , … , x n = (96)
∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) , \sum_{\substack{n=n_1+\cdots+n_k\\n_a\geq1}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\mathcal{G}^{\mathcal{F}(x_0),\mathcal{F}(x_{n_1}),\mathcal{F}(x_{n_1+n_2}),\ldots,\mathcal{F}(x_n)}_k\left(\mathcal{F}^{x_0,\ldots,x_{n_1}}_{n_1}\otimes\mathcal{F}^{x_{n_1},\ldots,x_{n_1+n_2}}_{n_2}\otimes\cdots\otimes\mathcal{F}^{x_{n-n_k},\ldots,x_n}_{n_k}\right), n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) ,
for all n ≥ 1 n\geq1 n ≥ 1 and x 0 , … , x n ∈ Obj A x_0,\ldots,x_n\in\operatorname{Obj}\mathcal{A} x 0 , … , x n ∈ Obj A . Composition is associative.
Write A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R for the category with objects A ∞ A_\infty A ∞ -categories A \mathcal{A} A over R R R and morphisms functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B .
If A , B \mathcal{A},\mathcal{B} A , B are unital A ∞ A_\infty A ∞ -categories over R R R , as in Definition 4.5, we call a functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B unital if
F 1 x , x ( 1 x ) = 1 F Obj ( x ) (97) \mathcal{F}^{x,x}_1(1_x)=1_{\mathcal{F}_{\operatorname{Obj}}(x)} \tag*{(97)} F 1 x , x ( 1 x ) = 1 F Obj ( x ) (97)
for all objects x x x in A \mathcal{A} A , and
F k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (98) \mathcal{F}^{x_0,\ldots,x_i,x_i,\ldots,x_k}_{k+1}(a_1\otimes\cdots\otimes a_i\otimes1_{x_i}\otimes a_{i+1}\otimes\cdots\otimes a_k)=0 \tag*{(98)} F k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (98)
for all k ≥ 1 k\geq1 k ≥ 1 , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A , i = 0 , … , k i=0,\ldots,k i = 0 , … , k and a j ∈ Hom A ( x j − 1 , x j ) a_j\in\operatorname{Hom}_{\mathcal{A}}(x_{j-1},x_j) a j ∈ Hom A ( x j − 1 , x j ) for j = 1 , … , k j=1,\ldots,k j = 1 , … , k . Unital A ∞ A_\infty A ∞ -categories over R R R and unital functors form a non-full subcategory A ∞ Cat R u n ⊂ A ∞ Cat R A_\infty\operatorname{Cat}^{\mathrm{un}}_R\subset A_\infty\operatorname{Cat}_R A ∞ Cat R un ⊂ A ∞ Cat R .
Let A , B \mathcal{A},\mathcal{B} A , B be (weakly) cohomologically unital A ∞ A_\infty A ∞ -categories, as in Definitions 4.6 and 4.7, and F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a functor. We call F \mathcal{F} F (weakly) cohomologically unital if H 0 ( F 1 x ) ( 1 x ) = 1 F Obj ( x ) H^0(\mathcal{F}^x_1)(1_x)=1_{\mathcal{F}_{\operatorname{Obj}}(x)} H 0 ( F 1 x ) ( 1 x ) = 1 F Obj ( x ) for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A . Then (weakly) cohomologically unital A ∞ A_\infty A ∞ -categories over R R R and (weakly) cohomologically unital functors form non-full subcategories A ∞ C a t R c u ⊂ A ∞ C a t R w c ⊂ A ∞ C a t R A_\infty\mathrm{Cat}^{\mathrm{cu}}_R \subset A_\infty\mathrm{Cat}^{\mathrm{wc}}_R \subset A_\infty\mathrm{Cat}_R A ∞ Cat R cu ⊂ A ∞ Cat R wc ⊂ A ∞ Cat R . When A , B \mathcal{A},\mathcal{B} A , B are unital, unital functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B are cohomologically unital, so A ∞ C a t R u n ⊂ A ∞ C a t R c u A_\infty\mathrm{Cat}^{\mathrm{un}}_R \subset A_\infty\mathrm{Cat}^{\mathrm{cu}}_R A ∞ Cat R un ⊂ A ∞ Cat R cu .
If F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a (cohomologically) unital functor of (cohomologically) unital A ∞ A_\infty A ∞ -categories over R R R , the morphisms (95) induce an R R R -linear functor
H 0 ( F 1 ) : H 0 ( A ) ⟶ H 0 ( B ) H^0(\mathcal{F}_1):H^0(\mathcal{A})\longrightarrow H^0(\mathcal{B}) H 0 ( F 1 ) : H 0 ( A ) ⟶ H 0 ( B )
of ordinary R R R -linear categories.
The A ∞ A_\infty A ∞ -category of A ∞ A_\infty A ∞ -functors
Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_\infty A ∞ -categories over R R R . Then one can define an A ∞ A_\infty A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) whose objects are A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B . This idea seems to be due to Kontsevich circa 1998, and accounts can be found in Fukaya [29, Th.-Def. 7.55], Lefèvre-Hasegawa [57, §8], Lyubashenko [64, §5], and Seidel [84, §1d, §2e]. See also Keller [52, §5], and Bespalov–Lyubashenko–Manzyuk [4, Prop. 8.15] who give a version for A ∞ A_\infty A ∞ -multifunctors F : A 1 × ⋯ × A n → B \mathcal{F}:\mathcal{A}_1\times\cdots\times\mathcal{A}_n\to\mathcal{B} F : A 1 × ⋯ × A n → B . Our presentation is based on [29,57], but these are written in terms of bar complexes as in §3.6, so we have used the method of §3.6 to compute the signs in (101) and (103) from the (much simpler) signs in [29,57].
Definition 4.10. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_\infty A ∞ -categories over a commutative ring R R R . We will define an A ∞ A_\infty A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) . Objects of Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) are A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , as in §4.2. For objects F , G \mathcal{F},\mathcal{G} F , G , we define a graded R R R -module
Hom Fun A ∞ ( A , B ) ( F , G ) = Hom A ∞ ( F , G ) = ⨁ l ∈ Z Hom A ∞ l ( F , G ) \operatorname{Hom}_{\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B})}(\mathcal{F},\mathcal{G})=\operatorname{Hom}_{A_\infty}(\mathcal{F},\mathcal{G})=\bigoplus_{l\in\mathbb{Z}}\operatorname{Hom}_{A_\infty}^{l}(\mathcal{F},\mathcal{G}) Hom Fun A ∞ ( A , B ) ( F , G ) = Hom A ∞ ( F , G ) = l ∈ Z ⨁ Hom A ∞ l ( F , G )
as follows. An element T \mathcal{T} T of Hom A ∞ l ( F , G ) \operatorname{Hom}_{A_\infty}^{l}(\mathcal{F},\mathcal{G}) Hom A ∞ l ( F , G ) consists of, for all k ≥ 0 k\geq0 k ≥ 0 and objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in A \mathcal{A} A , a graded R R R -linear map of degree l − k l-k l − k
T k = T k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F O b j ( x 0 ) , G O b j ( x k ) ) . (99) \mathcal{T}_k=\mathcal{T}^{x_0,\ldots,x_k}_k:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{G}_{\mathrm{Obj}}(x_k)). \tag*{(99)} T k = T k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , G Obj ( x k )) . (99)
Note that here we allow k = 0 k=0 k = 0 , in contrast to Definitions 4.1 and 4.9 when we consider m k , F k m_k,F_k m k , F k for k ≥ 1 k\geq1 k ≥ 1 , and T 0 x 0 \mathcal{T}_0^{x_0} T 0 x 0 maps R → Hom B l ( F ( x 0 ) , G ( x 0 ) ) R\to\operatorname{Hom}_{\mathcal{B}}^{l}(\mathcal{F}(x_0),\mathcal{G}(x_0)) R → Hom B l ( F ( x 0 ) , G ( x 0 )) . Thus T 0 x 0 \mathcal{T}_0^{x_0} T 0 x 0 is equivalent to an element T 0 x 0 ( 1 ) ∈ Hom B l ( F ( x 0 ) , G ( x 0 ) ) \mathcal{T}_0^{x_0}(1)\in\operatorname{Hom}_{\mathcal{B}}^{l}(\mathcal{F}(x_0),\mathcal{G}(x_0)) T 0 x 0 ( 1 ) ∈ Hom B l ( F ( x 0 ) , G ( x 0 )) .
Such T \mathcal{T} T are called pre natural transformations from F \mathcal{F} F to G \mathcal{G} G [29, Def. 7.49]. We write the A ∞ A_\infty A ∞ -operations on the A ∞ A_\infty A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) as n p n_p n p for p ≥ 1 p\geq1 p ≥ 1 . Firstly, when p = 1 p=1 p = 1 we define
n 1 = n 1 F , G : Hom A ∞ ( F , G ) ⟶ Hom A ∞ ( F , G ) (100) n_1=n_1^{\mathcal{F},\mathcal{G}}:\operatorname{Hom}_{A_\infty}(\mathcal{F},\mathcal{G})\longrightarrow\operatorname{Hom}_{A_\infty}(\mathcal{F},\mathcal{G}) \tag*{(100)} n 1 = n 1 F , G : Hom A ∞ ( F , G ) ⟶ Hom A ∞ ( F , G ) (100)
by n 1 ( T ) = L \mathfrak n_1(\mathcal T)=\mathcal L n 1 ( T ) = L , where if n ≥ 0 n\geq0 n ≥ 0 and x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n are objects in A \mathcal A A then we define
L n x 0 , … , x n = ∑ 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ j ≤ k : l i > l i − 1 , i ≠ 0 , j ( − 1 ) ( k + 1 ) ( deg T + 1 ) ⋅ ( − 1 ) ∑ i = 1 j − 1 ( l i − l i − 1 + 1 ) ( deg T + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) (101) \mathcal L_n^{x_0,\ldots,x_n} =
\sum_{\substack{0=l_0\leq l_1\leq\cdots\leq l_k=n,\\1\leq j\leq k:\\l_i>l_{i-1},\ i\neq0,j}}
(-1)^{(k+1)(\deg\mathcal T+1)}\cdot(-1)^{\sum_{i=1}^{j-1}(l_i-l_{i-1}+1)(\deg\mathcal T+1)}
\cdot
(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)} \tag*{(101)} L n x 0 , … , x n = 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ j ≤ k : l i > l i − 1 , i = 0 , j ∑ ( − 1 ) ( k + 1 ) ( d e g T + 1 ) ⋅ ( − 1 ) ∑ i = 1 j − 1 ( l i − l i − 1 + 1 ) ( d e g T + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) (101)
m k B ; F ( x l 0 ) , F ( x l 1 ) , F ( x l 2 ) , … , F ( x l j − 1 ) , G ( x l j ) , G ( x l j + 1 ) , … , G ( x l k ) ( F l 1 − l 0 x l 0 , … , x l 1 ⊗ F l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ F l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ F l j − l j − 1 x l j − 1 , … , x l j ⊗ G l j + 1 − l j x l j , … , x l j + 1 ⊗ ⋯ ⊗ G l k − l k − 1 x l k − 1 , … , x l k ) + ∑ i , k ≥ 0 , j ≥ 1 : i + j + k = n ( − 1 ) deg T ⋅ ( − 1 ) i + j k ⋅ T i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) . \begin{aligned}
\mathfrak m_k^{\mathcal B;\mathcal F(x_{l_0}),\mathcal F(x_{l_1}),\mathcal F(x_{l_2}),\ldots,\mathcal F(x_{l_{j-1}}),\mathcal G(x_{l_j}),\mathcal G(x_{l_{j+1}}),\ldots,\mathcal G(x_{l_k})}
\left(
\mathcal F_{l_1-l_0}^{x_{l_0},\ldots,x_{l_1}}\otimes
\mathcal F_{l_2-l_1}^{x_{l_1},\ldots,x_{l_2}}\otimes\cdots\otimes
\mathcal F_{l_{j-1}-l_{j-2}}^{x_{l_{j-2}},\ldots,x_{l_{j-1}}}
\otimes
\mathcal F_{l_j-l_{j-1}}^{x_{l_{j-1}},\ldots,x_{l_j}}
\otimes
\mathcal G_{l_{j+1}-l_j}^{x_{l_j},\ldots,x_{l_{j+1}}}
\otimes\cdots\otimes
\mathcal G_{l_k-l_{k-1}}^{x_{l_{k-1}},\ldots,x_{l_k}}
\right) \\
&+\sum_{\substack{i,k\geq0,\ j\geq1:\\i+j+k=n}}
(-1)^{\deg\mathcal T}\cdot(-1)^{i+jk}\cdot
\mathcal T_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}
\circ
\left(
\operatorname{id}_{\operatorname{Hom}_{\mathcal A}(x_0,x_1)}\otimes\cdots\otimes
\operatorname{id}_{\operatorname{Hom}_{\mathcal A}(x_{i-1},x_i)}
\otimes
\mathfrak m_j^{\mathcal A;x_i,\ldots,x_{i+j}}
\otimes
\operatorname{id}_{\operatorname{Hom}_{\mathcal A}(x_{i+j},x_{i+j+1})}
\otimes\cdots\otimes
\operatorname{id}_{\operatorname{Hom}_{\mathcal A}(x_{n-1},x_n)}
\right).
\end{aligned} m k B ; F ( x l 0 ) , F ( x l 1 ) , F ( x l 2 ) , … , F ( x l j − 1 ) , G ( x l j ) , G ( x l j + 1 ) , … , G ( x l k ) ( F l 1 − l 0 x l 0 , … , x l 1 ⊗ F l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ F l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ F l j − l j − 1 x l j − 1 , … , x l j ⊗ G l j + 1 − l j x l j , … , x l j + 1 ⊗ ⋯ ⊗ G l k − l k − 1 x l k − 1 , … , x l k ) + i , k ≥ 0 , j ≥ 1 : i + j + k = n ∑ ( − 1 ) d e g T ⋅ ( − 1 ) i + j k ⋅ T i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) .
Secondly, let p ≥ 2 p\geq2 p ≥ 2 and F 0 , F 1 , … , F p : A → B \mathcal F^0,\mathcal F^1,\ldots,\mathcal F^p:\mathcal A\to\mathcal B F 0 , F 1 , … , F p : A → B be A ∞ A_\infty A ∞ -functors. We define the A ∞ A_\infty A ∞ -operation n p \mathfrak n_p n p in Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal A,\mathcal B) Fun A ∞ ( A , B ) ,
n p = n p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ R ⋯ ⊗ R Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 , F p ) (102) \mathfrak n_p=\mathfrak n_p^{\mathcal F^0,\ldots,\mathcal F^p}:
\operatorname{Hom}_{A_\infty}(\mathcal F^0,\mathcal F^1)\otimes_R\cdots\otimes_R
\operatorname{Hom}_{A_\infty}(\mathcal F^{p-1},\mathcal F^p)
\longrightarrow
\operatorname{Hom}_{A_\infty}(\mathcal F^0,\mathcal F^p) \tag*{(102)} n p = n p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ R ⋯ ⊗ R Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 , F p ) (102)
by, if T i ∈ Hom A ∞ ( F i − 1 , F i ) \mathcal T^i\in\operatorname{Hom}_{A_\infty}(\mathcal F^{i-1},\mathcal F^i) T i ∈ Hom A ∞ ( F i − 1 , F i ) for i = 1 , … , p i=1,\ldots,p i = 1 , … , p , then n p F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) = L \mathfrak n_p^{\mathcal F^0,\ldots,\mathcal F^p}(\mathcal T^1\otimes\cdots\otimes\mathcal T^p)=\mathcal L n p F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) = L in Hom A ∞ ( F 0 , F p ) \operatorname{Hom}_{A_\infty}(\mathcal F^0,\mathcal F^p) Hom A ∞ ( F 0 , F p ) , where if n ≥ 0 n\geq0 n ≥ 0 and x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n are objects in A \mathcal A A then we define
L n x 0 , … , x n = ∑ 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ a 1 < a 2 < ⋯ < a p ≤ k : l j > l j − 1 , j ≠ 0 , a i ( − 1 ) ∑ i = 1 p ( k + 1 ) ( deg T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 p ∑ j = 1 a i − 1 ( l j − l j − 1 + 1 ) ( deg T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) (103) \mathcal L_n^{x_0,\ldots,x_n} =
\sum_{\substack{0=l_0\leq l_1\leq\cdots\leq l_k=n,\\1\leq a_1<a_2<\cdots<a_p\leq k:\\l_j>l_{j-1},\ j\neq0,a_i}}
(-1)^{\sum_{i=1}^{p}(k+1)(\deg\mathcal T^i+1)}
\cdot
(-1)^{\sum_{i=1}^{p}\sum_{j=1}^{a_i-1}(l_j-l_{j-1}+1)(\deg\mathcal T^i+1)}
\cdot
(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)} \tag*{(103)} L n x 0 , … , x n = 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ a 1 < a 2 < ⋯ < a p ≤ k : l j > l j − 1 , j = 0 , a i ∑ ( − 1 ) ∑ i = 1 p ( k + 1 ) ( d e g T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 p ∑ j = 1 a i − 1 ( l j − l j − 1 + 1 ) ( d e g T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) (103)
\mathfrak m_k^{\mathcal B;\mathcal F^0(x_{l_0}),\mathcal F^0(x_{l_1}),\ldots,\mathcal F^0(x_{l_{a_1-1}}),\mathcal F^1(x_{l_{a_1}}),\mathcal F^1(x_{l_{a_1+1}}),\ldots,\mathcal F^1(x_{l_{a_2-1}}),\ldots,\mathcal F^p(x_{l_{a_p}}),\mathcal F^p(x_{l_{a_p+1}}),\ldots,\mathcal F^p(x_{l_k})}
\left(
\mathcal F^0_{l_1-l_0}^{x_{l_0},\ldots,x_{l_1}}\otimes
\mathcal F^0_{l_2-l_1}^{x_{l_1},\ldots,x_{l_2}}\otimes\cdots\otimes
\mathcal F^0_{l_{a_1-1}-l_{a_1-2}}^{x_{l_{a_1-2}},\ldots,x_{l_{a_1-1}}}
\otimes
\mathcal T^1_{l_{a_1}-l_{a_1-1}}^{x_{l_{a_1-1}},\ldots,x_{l_{a_1}}}
\otimes
\mathcal F^1_{l_{a_1+1}-l_{a_1}}^{x_{l_{a_1}},\ldots,x_{l_{a_1+1}}}
\otimes\cdots\otimes
\mathcal F^1_{l_{a_2-1}-l_{a_2-2}}^{x_{l_{a_2-2}},\ldots,x_{l_{a_2-1}}}
\otimes
\mathcal T^2_{l_{a_2}-l_{a_2-1}}^{x_{l_{a_2-1}},\ldots,x_{l_{a_2}}}
\otimes\cdots\otimes
\mathcal T^p_{l_{a_p}-l_{a_p-1}}^{x_{l_{a_p-1}},\ldots,x_{l_{a_p}}}
\otimes
\mathcal F^p_{l_{a_p+1}-l_{a_p}}^{x_{l_{a_p}},\ldots,x_{l_{a_p+1}}}
\otimes\cdots\otimes
\mathcal F^p_{l_k-l_{k-1}}^{x_{l_{k-1}},\ldots,x_{l_k}}
\right).
When p = 1 p=1 p = 1 , equation (103) for ( F 0 , F 1 ) = ( F , G ) (\mathcal F^0,\mathcal F^1)=(\mathcal F,\mathcal G) ( F 0 , F 1 ) = ( F , G ) reduces to the first sum in (101) , but omits the second sum, which is why we treat p = 1 p=1 p = 1 separately. As in [28,57,64,84], the above make Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal A,\mathcal B) Fun A ∞ ( A , B ) into an A ∞ A_\infty A ∞ -category over R R R .
Next suppose A , B \mathcal A,\mathcal B A , B are cohomologically unital A ∞ A_\infty A ∞ -categories. Following [84, §2e], define Fun A ∞ c u ( A , B ) \operatorname{Fun}^{\mathrm{cu}}_{A_\infty}(\mathcal A,\mathcal B) Fun A ∞ cu ( A , B ) to be the full A ∞ A_\infty A ∞ -subcategory of Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal A,\mathcal B) Fun A ∞ ( A , B ) with objects cohomologically unital A ∞ A_\infty A ∞ -functors F : A → B \mathcal F:\mathcal A\to\mathcal B F : A → B . Then Fun A ∞ c u ( A , B ) \operatorname{Fun}^{\mathrm{cu}}_{A_\infty}(\mathcal A,\mathcal B) Fun A ∞ cu ( A , B ) is a cohomologically unital A ∞ A_\infty A ∞ -category by Lyubashenko [64, Prop. 7.7].
Definition 4.11. Suppose A , B \mathcal{A},\mathcal{B} A , B are unital A ∞ A_\infty A ∞ -categories over R R R . Following [57, §8.1.3], define a non-full A ∞ A_\infty A ∞ -subcategory Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) of Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) to have objects unital A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , and if F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B are unital functors, we define
Hom Fun A ∞ u n ( A , B ) ( F , G ) = Hom A ∞ u n ( F , G ) ⊂ Hom Fun A ∞ ( A , B ) ( F , G ) \operatorname{Hom}_{\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B})}(\mathcal{F},\mathcal{G})=\operatorname{Hom}^{\mathrm{un}}_{A_\infty}(\mathcal{F},\mathcal{G})\subset\operatorname{Hom}_{\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B})}(\mathcal{F},\mathcal{G}) Hom Fun A ∞ un ( A , B ) ( F , G ) = Hom A ∞ un ( F , G ) ⊂ Hom Fun A ∞ ( A , B ) ( F , G )
to be the subcomplex of T \mathcal{T} T in Hom A ∞ ( F , G ) \operatorname{Hom}_{A_\infty}(\mathcal{F},\mathcal{G}) Hom A ∞ ( F , G ) such that
T k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (104) \mathcal{T}^{x_0,\ldots,x_i,x_i,\ldots,x_k}_{k+1}(a_1\otimes\cdots\otimes a_i\otimes1_{x_i}\otimes a_{i+1}\otimes\cdots\otimes a_k)=0 \tag*{(104)} T k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0 (104)
for all 0 ≤ i ≤ k 0\leq i\leq k 0 ≤ i ≤ k , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and a j ∈ Hom ( x j − 1 , x j ) a_j\in\operatorname{Hom}(x_{j-1},x_j) a j ∈ Hom ( x j − 1 , x j ) for j = 1 , … , k j=1,\ldots,k j = 1 , … , k . As in [57, §8.1.3], these are closed under the operations n p F 0 , … , F p n^{\mathcal{F}^0,\ldots,\mathcal{F}^p}_p n p F 0 , … , F p , so Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) is an A ∞ A_\infty A ∞ -category.
In fact Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) is a unital A ∞ A_\infty A ∞ -category. For each unital A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , the identity is 1 F = S 1_{\mathcal{F}}=\mathcal{S} 1 F = S , where
S 0 x 0 : R ⟶ Hom B ( F ( x 0 ) , F ( x 0 ) ) , S 0 x 0 : λ ⟼ λ 1 F ( x 0 ) , \mathcal{S}^{x_0}_0:R\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_0)),\qquad\mathcal{S}^{x_0}_0:\lambda\longmapsto\lambda1_{\mathcal{F}(x_0)}, S 0 x 0 : R ⟶ Hom B ( F ( x 0 ) , F ( x 0 )) , S 0 x 0 : λ ⟼ λ 1 F ( x 0 ) ,
S k x 0 , … , x k = 0 for all k ≥ 1 and x 0 , … , x k ∈ Obj A . (105) \mathcal{S}^{x_0,\ldots,x_k}_k=0\quad\text{for all }k\geq1\text{ and }x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A}. \tag*{(105)} S k x 0 , … , x k = 0 for all k ≥ 1 and x 0 , … , x k ∈ Obj A . (105)
We can check this satisfies (4.5)–(4.6). For (4.5), if x 0 , x 1 ∈ Obj A x_0,x_1\in\operatorname{Obj}\mathcal{A} x 0 , x 1 ∈ Obj A and T ∈ Hom A ∞ ( F 0 , F 1 ) \mathcal{T}\in\operatorname{Hom}_{A_\infty}(\mathcal{F}^0,\mathcal{F}^1) T ∈ Hom A ∞ ( F 0 , F 1 ) then (4.17) and (4.5) for B \mathcal{B} B yield
n 2 F 0 , F 0 , F 1 ( 1 F 0 ⊗ T ) x 0 , … , x n = m 2 F 0 , F 1 ( 1 F 0 ( x 0 ) ⊗ T x 0 , … , x n ) = T x 0 , … , x n , n^{\mathcal{F}^0,\mathcal{F}^0,\mathcal{F}^1}_2(1_{\mathcal{F}^0}\otimes\mathcal{T})^{x_0,\ldots,x_n}
=m^{\mathcal{F}^0,\mathcal{F}^1}_{2}(1_{\mathcal{F}^0(x_0)}\otimes\mathcal{T}^{x_0,\ldots,x_n})
=\mathcal{T}^{x_0,\ldots,x_n}, n 2 F 0 , F 0 , F 1 ( 1 F 0 ⊗ T ) x 0 , … , x n = m 2 F 0 , F 1 ( 1 F 0 ( x 0 ) ⊗ T x 0 , … , x n ) = T x 0 , … , x n ,
n 2 F 0 , F 1 , F 1 ( T ⊗ 1 F 1 ) x 0 , … , x n = m 2 F 0 , F 1 ( T x 0 , … , x n ⊗ 1 F 1 ( x n ) ) = T x 0 , … , x n . n^{\mathcal{F}^0,\mathcal{F}^1,\mathcal{F}^1}_2(\mathcal{T}\otimes1_{\mathcal{F}^1})^{x_0,\ldots,x_n}
=m^{\mathcal{F}^0,\mathcal{F}^1}_{2}(\mathcal{T}^{x_0,\ldots,x_n}\otimes1_{\mathcal{F}^1(x_n)})
=\mathcal{T}^{x_0,\ldots,x_n}. n 2 F 0 , F 1 , F 1 ( T ⊗ 1 F 1 ) x 0 , … , x n = m 2 F 0 , F 1 ( T x 0 , … , x n ⊗ 1 F 1 ( x n ) ) = T x 0 , … , x n .
Remark 4.12. Definition 4.10 shows that in the category A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R , the mappings spaces Hom A ∞ Cat R ( A , B ) \operatorname{Hom}_{A_\infty\operatorname{Cat}_R}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R ( A , B ) can be given the structure of A ∞ A_\infty A ∞ -categories Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) . That is, A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R is morally ‘a category enriched over itself’.
This is relevant to a central question for this book: what kind of higher category should A ∞ A_\infty A ∞ -categories form? By Lyubashenko [64, §7], A ∞ A_\infty A ∞ -categories should be at least a 2-category. It is clear that A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R should really be an ( ∞ , ∞ ) (\infty,\infty) ( ∞ , ∞ ) -category, since if A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R has (possibly non-invertible) n n n -morphisms, then Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) should have (possibly non-invertible) n n n -morphisms, and thus A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R must have (possibly non-invertible) ( n + 1 ) (n+1) ( n + 1 ) -morphisms.
We do not really understand ( ∞ , ∞ ) (\infty,\infty) ( ∞ , ∞ ) -categories, but we do have a good theory of ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -categories, namely ∞ \infty ∞ -categories (quasicategories). To make A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R into an ∞ \infty ∞ -category, we allow only invertible n n n -morphisms for n ≥ 2 n\geq2 n ≥ 2 , which roughly corresponds to restricting (in Definition 7.12 below) to the subcategory of Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) with morphisms homotopy equivalences.
In Theorem 7.17 we will also define an ∞ \infty ∞ -bicategory of unital A ∞ A_\infty A ∞ -categories over R R R , which is a kind of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category, as in §2.8.
Next we discuss functoriality of Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) under A ∞ A_\infty A ∞ -functors E : A 1 → A 2 \mathcal{E}:\mathcal{A}_1\to\mathcal{A}_2 E : A 1 → A 2 or G : B 1 → B 2 \mathcal{G}:\mathcal{B}_1\to\mathcal{B}_2 G : B 1 → B 2 , following Lefèvre-Hasegawa [57, §8.1.2], Lyubashenko [64, Prop. 6.2], and Fukaya [29, Prop.-Def. 8.41].
Definition 4.13. (a) Let A 1 , A 2 , B \mathcal{A}_1,\mathcal{A}_2,\mathcal{B} A 1 , A 2 , B be A ∞ A_\infty A ∞ -categories over a commutative ring R R R , and E : A 1 → A 2 \mathcal{E}:\mathcal{A}_1\to\mathcal{A}_2 E : A 1 → A 2 be an A ∞ A_\infty A ∞ -functor. We define an A ∞ A_\infty A ∞ -functor E ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) \mathcal{E}^*:\operatorname{Fun}_{A_\infty}(\mathcal{A}_2,\mathcal{B})\to\operatorname{Fun}_{A_\infty}(\mathcal{A}_1,\mathcal{B}) E ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) to act on objects by E O b j ∗ : F ↦ F ∘ E \mathcal{E}^*_{\mathrm{Obj}}:\mathcal{F}\mapsto\mathcal{F}\circ\mathcal{E} E Obj ∗ : F ↦ F ∘ E for F : A 2 → B \mathcal{F}:\mathcal{A}_2\to\mathcal{B} F : A 2 → B an A ∞ A_\infty A ∞ -functor, and if p ≥ 1 p\ge1 p ≥ 1 and F 0 , … , F p \mathcal{F}_0,\ldots,\mathcal{F}_p F 0 , … , F p are objects in Fun A ∞ ( A 2 , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A}_2,\mathcal{B}) Fun A ∞ ( A 2 , B ) , we define
( E ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 ∘ E , F p ∘ E ) (\mathcal{E}^*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p:
\operatorname{Hom}_{A_\infty}(\mathcal{F}_0,\mathcal{F}_1)\otimes\cdots\otimes
\operatorname{Hom}_{A_\infty}(\mathcal{F}_{p-1},\mathcal{F}_p)
\longrightarrow
\operatorname{Hom}_{A_\infty}(\mathcal{F}_0\circ\mathcal{E},\mathcal{F}_p\circ\mathcal{E}) ( E ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 ∘ E , F p ∘ E )
by ( E ∗ ) p F 0 , … , F p = 0 (\mathcal{E}^*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p=0 ( E ∗ ) p F 0 , … , F p = 0 if p ≠ 1 p\ne1 p = 1 , and
( ( E ∗ ) 1 F 0 , F 1 ( T ) ) x 0 , … , x n = ∑ 0 = l 0 < l 1 < ⋯ < l k = n ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) T k E ( x l 0 ) , E ( x l 1 ) , E ( x l 2 ) , … , E ( x l k ) ( E l 1 − l 0 x l 0 , … , x l 1 ⊗ E l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ E l k − l k − 1 x l k − 1 , … , x l k ) . (106) \left((\mathcal{E}^*)^{\mathcal{F}_0,\mathcal{F}_1}_1(\mathcal{T})\right)^{x_0,\ldots,x_n}
=
\sum_{0=l_0<l_1<\cdots<l_k=n}
(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)}
\mathcal{T}^{\mathcal{E}(x_{l_0}),\mathcal{E}(x_{l_1}),\mathcal{E}(x_{l_2}),\ldots,\mathcal{E}(x_{l_k})}_k
\left(
\mathcal{E}^{x_{l_0},\ldots,x_{l_1}}_{l_1-l_0}\otimes
\mathcal{E}^{x_{l_1},\ldots,x_{l_2}}_{l_2-l_1}\otimes\cdots\otimes
\mathcal{E}^{x_{l_{k-1}},\ldots,x_{l_k}}_{l_k-l_{k-1}}
\right). \tag*{(106)} ( ( E ∗ ) 1 F 0 , F 1 ( T ) ) x 0 , … , x n = 0 = l 0 < l 1 < ⋯ < l k = n ∑ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) T k E ( x l 0 ) , E ( x l 1 ) , E ( x l 2 ) , … , E ( x l k ) ( E l 1 − l 0 x l 0 , … , x l 1 ⊗ E l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ E l k − l k − 1 x l k − 1 , … , x l k ) . (106)
This defines an A ∞ A_\infty A ∞ -functor as in [29,57,64], contravariantly functorial in E \mathcal{E} E .
(b) Let A , B 1 , B 2 \mathcal{A},\mathcal{B}_1,\mathcal{B}_2 A , B 1 , B 2 be A ∞ A_\infty A ∞ -categories over R R R , and G : B 1 → B 2 \mathcal{G}:\mathcal{B}_1\to\mathcal{B}_2 G : B 1 → B 2 be an A ∞ A_\infty A ∞ -functor. We define an A ∞ A_\infty A ∞ -functor G ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) \mathcal{G}_*:\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}_1)\to\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}_2) G ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) to act on objects by ( G ∗ ) O b j : F ↦ G ∘ F (\mathcal{G}_*)_{\mathrm{Obj}}:\mathcal{F}\mapsto\mathcal{G}\circ\mathcal{F} ( G ∗ ) Obj : F ↦ G ∘ F for F : A → B 1 \mathcal{F}:\mathcal{A}\to\mathcal{B}_1 F : A → B 1 an A ∞ A_\infty A ∞ -functor, and if p ≥ 1 p\ge1 p ≥ 1 and F 0 , … , F p \mathcal{F}_0,\ldots,\mathcal{F}_p F 0 , … , F p are objects in Fun A ∞ ( A , B 1 ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}_1) Fun A ∞ ( A , B 1 ) , we define
( G ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( G ∘ F 0 , G ∘ F p ) (107) (\mathcal{G}_*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p:
\operatorname{Hom}_{A_\infty}(\mathcal{F}_0,\mathcal{F}_1)\otimes\cdots\otimes
\operatorname{Hom}_{A_\infty}(\mathcal{F}_{p-1},\mathcal{F}_p)
\longrightarrow
\operatorname{Hom}_{A_\infty}(\mathcal{G}\circ\mathcal{F}_0,\mathcal{G}\circ\mathcal{F}_p) \tag*{(107)} ( G ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( G ∘ F 0 , G ∘ F p ) (107) by
\begin{aligned}
\left((\mathcal{G}_*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p
(\mathcal{T}_1\otimes\cdots\otimes\mathcal{T}_p)\right)^{x_0,\ldots,x_n}
={}&
\sum_{\substack{0=l_0\le l_1\le\cdots\le l_k=n,\\
1\le a_1<a_2<\cdots<a_p\le k:\\
l_j>l_{j-1},\ j\ne0,a_i}}
(-1)^{\sum_{i=1}^{p}(k+1)(\deg\mathcal{T}_{i+1})}
(-1)^{\sum_{i=1}^{p}\sum_{j=1}^{a_i-1}(l_j-l_{j-1}+1)(\deg\mathcal{T}_{i+1})}\\
&\cdot(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)}
\mathcal{G}^{\mathcal{F}_0(x_{l_0}),\mathcal{F}_0(x_{l_1}),\ldots,\mathcal{F}_0(x_{l_{a_1-1}}),
\mathcal{F}_1(x_{l_{a_1}}),\mathcal{F}_1(x_{l_{a_1+1}}),\ldots,
\mathcal{F}_1(x_{l_{a_2-1}}),\ldots,\mathcal{F}_p(x_{l_{a_p}}),
\mathcal{F}_p(x_{l_{a_p+1}}),\ldots,\mathcal{F}_p(x_{l_k})}_k\\
&\cdot\left(
\mathcal{F}_0^{x_{l_0},\ldots,x_{l_1}}_{l_1-l_0}\otimes
\mathcal{F}_0^{x_{l_1},\ldots,x_{l_2}}_{l_2-l_1}\otimes\cdots\otimes
\mathcal{F}_0^{x_{l_{a_1-2}},\ldots,x_{l_{a_1-1}}}_{l_{a_1-1}-l_{a_1-2}}\right.\\
&\qquad\otimes\mathcal{T}_1^{x_{l_{a_1-1}},\ldots,x_{l_{a_1}}}_{l_{a_1}-l_{a_1-1}}
\otimes\mathcal{F}_1^{x_{l_{a_1}},\ldots,x_{l_{a_1+1}}}_{l_{a_1+1}-l_{a_1}}
\otimes\cdots\otimes
\mathcal{F}_1^{x_{l_{a_2-2}},\ldots,x_{l_{a_2-1}}}_{l_{a_2-1}-l_{a_2-2}}\\
&\qquad\otimes\mathcal{T}_2^{x_{l_{a_2-1}},\ldots,x_{l_{a_2}}}_{l_{a_2}-l_{a_2-1}}
\otimes\cdots\otimes
\mathcal{T}_p^{x_{l_{a_p-1}},\ldots,x_{l_{a_p}}}_{l_{a_p}-l_{a_p-1}}\\
&\qquad\left.\otimes\mathcal{F}_p^{x_{l_{a_p}},\ldots,x_{l_{a_p+1}}}_{l_{a_p+1}-l_{a_p}}
\otimes\cdots\otimes
\mathcal{F}_p^{x_{l_{k-1}},\ldots,x_{l_k}}_{l_k-l_{k-1}}
\right).
\end{aligned}
This defines an A ∞ A_\infty A ∞ -functor as in [29,57,64], covariantly functorial in G \mathcal{G} G .
(c) Combining Definition 4.10 and (a),(b) gives a functor
Fun A ∞ : ( A ∞ Cat R ) o p × A ∞ Cat R ⟶ A ∞ Cat R . (108) \operatorname{Fun}_{A_\infty}:(A_\infty\operatorname{Cat}_R)^{\mathrm{op}}\times A_\infty\operatorname{Cat}_R\longrightarrow A_\infty\operatorname{Cat}_R. \tag*{(108)} Fun A ∞ : ( A ∞ Cat R ) op × A ∞ Cat R ⟶ A ∞ Cat R . (108) acting on objects ( A , B ) ↦ Fun A ∞ ( A , B ) (\mathcal{A},\mathcal{B})\mapsto\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) ( A , B ) ↦ Fun A ∞ ( A , B ) and on morphisms by ( E o p , G ) ↦ E ∗ ∘ G ∗ (\mathcal{E}^{\mathrm{op}},\mathcal{G})\mapsto\mathcal{E}^*\circ\mathcal{G}_* ( E op , G ) ↦ E ∗ ∘ G ∗ . Here E ∗ \mathcal{E}^* E ∗ and G ∗ \mathcal{G}_* G ∗ commute, so their order does not matter.
(d) The analogue of (a)–(c) works for cohomologically unital A ∞ A_\infty A ∞ -categories, cohomologically unital A ∞ A_\infty A ∞ -functors, and Fun A ∞ c u ( − , − ) \operatorname{Fun}^{\mathrm{cu}}_{A_\infty}(-,-) Fun A ∞ cu ( − , − ) in Definition 4.10, giving a functor
Fun A ∞ c u : ( A ∞ Cat R c u ) o p × A ∞ Cat R c u ⟶ A ∞ Cat R c u . (109) \operatorname{Fun}^{\mathrm{cu}}_{A_\infty} : (A_\infty\operatorname{Cat}^{\mathrm{cu}}_R)^{\mathrm{op}} \times A_\infty\operatorname{Cat}^{\mathrm{cu}}_R \longrightarrow A_\infty\operatorname{Cat}^{\mathrm{cu}}_R . \tag*{(109)} Fun A ∞ cu : ( A ∞ Cat R cu ) op × A ∞ Cat R cu ⟶ A ∞ Cat R cu . (109)
(e) The analogue of (a)–(c) works for unital A ∞ A_\infty A ∞ -categories, unital A ∞ A_\infty A ∞ -functors, and Fun A ∞ u n ( − , − ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty}(-,-) Fun A ∞ un ( − , − ) in Definition 4.11, giving a functor
Fun A ∞ u n : ( A ∞ Cat R u n ) o p × A ∞ Cat R u n ⟶ A ∞ Cat R u n . (110) \operatorname{Fun}^{\mathrm{un}}_{A_\infty} : (A_\infty\operatorname{Cat}^{\mathrm{un}}_R)^{\mathrm{op}} \times A_\infty\operatorname{Cat}^{\mathrm{un}}_R \longrightarrow A_\infty\operatorname{Cat}^{\mathrm{un}}_R . \tag*{(110)} Fun A ∞ un : ( A ∞ Cat R un ) op × A ∞ Cat R un ⟶ A ∞ Cat R un . (110)
We explain two simple examples of functor A ∞ A_\infty A ∞ -categories, which will be important in §5.7–§5.8 and §9.3 below.
Definition 4.15. Let R R R be a commutative ring. Define an A ∞ A_\infty A ∞ -category P \mathcal{P} P over R R R to have Obj P = { ∗ } \operatorname{Obj}\mathcal{P}=\{\ast\} Obj P = { ∗ } , that is, P \mathcal{P} P has one object ∗ \ast ∗ , and set Hom P ( ∗ , ∗ ) = 0 \operatorname{Hom}_{\mathcal{P}}(\ast,\ast)=0 Hom P ( ∗ , ∗ ) = 0 with maps m k ∗ , … , ∗ = 0 m_k^{\ast,\ldots,\ast}=0 m k ∗ , … , ∗ = 0 for k ≥ 1 k\geq1 k ≥ 1 . We call P \mathcal{P} P the (non-unital) point A ∞ A_\infty A ∞ -category.
Proposition 4.16. Let A \mathcal{A} A be any A ∞ A_\infty A ∞ -category over R R R . Then there is a canonical isomorphism of A ∞ A_\infty A ∞ -categories A ≅ Fun A ∞ ( P , A ) \mathcal{A}\cong\operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A}) A ≅ Fun A ∞ ( P , A ) , where Fun A ∞ ( P , A ) \operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A}) Fun A ∞ ( P , A ) is as in Definition 4.10.
Proof. Objects of Fun A ∞ ( P , A ) \operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A}) Fun A ∞ ( P , A ) are A ∞ A_\infty A ∞ -functors F : P → A \mathcal{F}:\mathcal{P}\to\mathcal{A} F : P → A . Such F \mathcal{F} F has a map F O b j : Obj P = { ∗ } → Obj A \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{P}=\{\ast\}\to\operatorname{Obj}\mathcal{A} F Obj : Obj P = { ∗ } → Obj A , and so determines an object x = F ( ∗ ) x=\mathcal{F}(\ast) x = F ( ∗ ) in A \mathcal{A} A . Since Hom P ( ∗ , ∗ ) = 0 \operatorname{Hom}_{\mathcal{P}}(\ast,\ast)=0 Hom P ( ∗ , ∗ ) = 0 we have F k ∗ , … , ∗ = 0 \mathcal{F}^{\ast,\ldots,\ast}_k=0 F k ∗ , … , ∗ = 0 for all k ≥ 1 k\geq1 k ≥ 1 . Hence objects F \mathcal{F} F in Fun A ∞ ( P , A ) \operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A}) Fun A ∞ ( P , A ) are in 1-1 correspondence with objects x x x in A \mathcal{A} A , and we write F x \mathcal{F}^x F x for the functor with F x ( ∗ ) = x \mathcal{F}^x(\ast)=x F x ( ∗ ) = x .
For objects x , y x,y x , y in A \mathcal{A} A , consider the graded R R R -module Hom A ( F x , F y ) = Hom Fun A ∞ ( P , A ) ( F x , F y ) \operatorname{Hom}_{\mathcal{A}}(\mathcal{F}^x,\mathcal{F}^y)=\operatorname{Hom}_{\operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A})}(\mathcal{F}^x,\mathcal{F}^y) Hom A ( F x , F y ) = Hom Fun A ∞ ( P , A ) ( F x , F y ) . Elements of this are ( T k ∗ , … , ∗ ) k ≥ 0 (\mathcal{T}^{\ast,\ldots,\ast}_k)_{k\geq0} ( T k ∗ , … , ∗ ) k ≥ 0 . As Hom P ( ∗ , ∗ ) = 0 \operatorname{Hom}_{\mathcal{P}}(\ast,\ast)=0 Hom P ( ∗ , ∗ ) = 0 we have T k ∗ , … , ∗ = 0 \mathcal{T}^{\ast,\ldots,\ast}_k=0 T k ∗ , … , ∗ = 0 for k ≥ 1 k\geq1 k ≥ 1 , so by (4.13) the only data is
T 0 ∗ , ∗ ∈ Hom R ( R , Hom A ( x , y ) ) ≅ Hom A ( x , y ) . \mathcal{T}^{\ast,\ast}_0\in\operatorname{Hom}_R(R,\operatorname{Hom}_{\mathcal{A}}(x,y))\cong\operatorname{Hom}_{\mathcal{A}}(x,y). T 0 ∗ , ∗ ∈ Hom R ( R , Hom A ( x , y )) ≅ Hom A ( x , y ) .
This gives a canonical isomorphism of graded R R R -modules
Hom Fun A ∞ ( P , A ) ( F x , F y ) ≅ Hom A ( x , y ) . \operatorname{Hom}_{\operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A})}(\mathcal{F}^x,\mathcal{F}^y)\cong\operatorname{Hom}_{\mathcal{A}}(x,y). Hom Fun A ∞ ( P , A ) ( F x , F y ) ≅ Hom A ( x , y ) .
It is now easy to check that these isomorphisms also identify the maps m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k in A \mathcal{A} A and Fun A ∞ ( P , A ) \operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A}) Fun A ∞ ( P , A ) .
Here are the unital analogues of Definition 4.15 and Proposition 4.16.
Definition 4.17. Let R R R be a commutative ring. Define a unital A ∞ A_\infty A ∞ -category P u n \mathcal{P}^{\mathrm{un}} P un over R R R to have Obj P u n = { ∗ } \operatorname{Obj}\mathcal{P}^{\mathrm{un}}=\{\ast\} Obj P un = { ∗ } , that is, P u n \mathcal{P}^{\mathrm{un}} P un has one object ∗ \ast ∗ , and set Hom P u n ( ∗ , ∗ ) = R ⋅ 1 ∗ P u n \operatorname{Hom}_{\mathcal{P}^{\mathrm{un}}}(\ast,\ast)=R\cdot1_{\ast}^{\mathcal{P}^{\mathrm{un}}} Hom P un ( ∗ , ∗ ) = R ⋅ 1 ∗ P un with 1 ∗ P u n 1_{\ast}^{\mathcal{P}^{\mathrm{un}}} 1 ∗ P un in degree 0, and define the maps m k ∗ , … , ∗ m_k^{\ast,\ldots,\ast} m k ∗ , … , ∗ by
m 2 ∗ , ∗ , ∗ ( 1 ∗ P u n ⊗ 1 ∗ P u n ) = 1 ∗ P u n , m k ∗ , … , ∗ = 0 , k ≠ 2. m_2^{\ast,\ast,\ast}(1_{\ast}^{\mathcal{P}^{\mathrm{un}}}\otimes1_{\ast}^{\mathcal{P}^{\mathrm{un}}})=1_{\ast}^{\mathcal{P}^{\mathrm{un}}},\qquad m_k^{\ast,\ldots,\ast}=0,\quad k\ne2. m 2 ∗ , ∗ , ∗ ( 1 ∗ P un ⊗ 1 ∗ P un ) = 1 ∗ P un , m k ∗ , … , ∗ = 0 , k = 2. We call P u n \mathcal{P}^{\mathrm{un}} P un the unital point A ∞ A_\infty A ∞ -category. It is a dg-category in §4.5 below.
Proposition 4.18. Let A \mathcal{A} A be any unital A ∞ A_{\infty} A ∞ -category over R R R . Then there is a canonical isomorphism of unital A ∞ A_{\infty} A ∞ -categories A ≅ Fun A ∞ u n ( P u n , A ) \mathcal{A} \cong\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) A ≅ Fun A ∞ un ( P un , A ) , where Fun A ∞ u n ( P u n , A ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) Fun A ∞ un ( P un , A ) is as in Definitions 4.10–4.11.
Proof. Objects of Fun A ∞ u n ( P u n , A ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) Fun A ∞ un ( P un , A ) are unital A ∞ A_{\infty} A ∞ -functors F : P u n → A \mathcal{F}:\mathcal{P}^{\mathrm{un}}\to\mathcal{A} F : P un → A . Such F \mathcal{F} F has a map F O b j : Obj P u n = { ∗ } → Obj A \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{P}^{\mathrm{un}}=\{\ast\}\to\operatorname{Obj}\mathcal{A} F Obj : Obj P un = { ∗ } → Obj A , and so determines an object x = F ( ∗ ) x=\mathcal{F}(\ast) x = F ( ∗ ) in A \mathcal{A} A . Since Hom P u n ( ∗ , ∗ ) = R ⋅ 1 ∗ P u n \operatorname{Hom}_{\mathcal{P}^{\mathrm{un}}}(\ast,\ast)=R\cdot1_{\ast}^{\mathcal{P}^{\mathrm{un}}} Hom P un ( ∗ , ∗ ) = R ⋅ 1 ∗ P un , by (4.11)–(4.12), the maps F k ∗ , … , ∗ \mathcal{F}^{\ast,\ldots,\ast}_{k} F k ∗ , … , ∗ are determined uniquely by F 1 ∗ ( 1 ∗ P u n ) = 1 x A \mathcal{F}^{\ast}_{1}(1_{\ast}^{\mathcal{P}^{\mathrm{un}}})=1_{x}^{\mathcal{A}} F 1 ∗ ( 1 ∗ P un ) = 1 x A and F k ∗ , … , ∗ = 0 \mathcal{F}^{\ast,\ldots,\ast}_{k}=0 F k ∗ , … , ∗ = 0 for k ≠ 1 k\ne1 k = 1 . Hence objects F \mathcal{F} F in Fun A ∞ u n ( P u n , A ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) Fun A ∞ un ( P un , A ) are in 1-1 correspondence with objects x x x in A \mathcal{A} A , and we write F x \mathcal{F}^{x} F x for the functor with F x ( ∗ ) = x \mathcal{F}^{x}(\ast)=x F x ( ∗ ) = x .
For objects x , y x,y x , y in A \mathcal{A} A , consider the graded R R R -module Hom Fun A ∞ u n ( P u n , A ) ( F x , F y ) \operatorname{Hom}_{\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A})}(\mathcal{F}^{x},\mathcal{F}^{y}) Hom Fun A ∞ un ( P un , A ) ( F x , F y ) . Elements of this are ( T k ∗ , … , ∗ ) k ≥ 0 (T^{\ast,\ldots,\ast}_{k})_{k\geq0} ( T k ∗ , … , ∗ ) k ≥ 0 . As Hom P u n ( ∗ , ∗ ) = R ⋅ 1 ∗ P u n \operatorname{Hom}_{\mathcal{P}^{\mathrm{un}}}(\ast,\ast)=R\cdot1_{\ast}^{\mathcal{P}^{\mathrm{un}}} Hom P un ( ∗ , ∗ ) = R ⋅ 1 ∗ P un , we have T k ∗ , … , ∗ = 0 T^{\ast,\ldots,\ast}_{k}=0 T k ∗ , … , ∗ = 0 for k ≥ 1 k\geq1 k ≥ 1 by (4.18), so by (4.13) the only data is
T 0 ∗ , ∗ ∈ Hom R ( R , Hom A ( x , y ) ) ≅ Hom A ( x , y ) . T^{\ast,\ast}_{0}\in\operatorname{Hom}_{R}(R,\operatorname{Hom}_{\mathcal{A}}(x,y))\cong\operatorname{Hom}_{\mathcal{A}}(x,y). T 0 ∗ , ∗ ∈ Hom R ( R , Hom A ( x , y )) ≅ Hom A ( x , y ) . This gives a canonical isomorphism of graded R R R -modules
Hom Fun A ∞ u n ( P u n , A ) ( F x , F y ) ≅ Hom A ( x , y ) . \operatorname{Hom}_{\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A})}(\mathcal{F}^{x},\mathcal{F}^{y})\cong\operatorname{Hom}_{\mathcal{A}}(x,y). Hom Fun A ∞ un ( P un , A ) ( F x , F y ) ≅ Hom A ( x , y ) . When x = y x=y x = y , this isomorphism identifies 1 F x Fun A ∞ u n ( P u n , A ) 1_{\mathcal{F}^{x}}^{\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A})} 1 F x Fun A ∞ un ( P un , A ) in (4.19) with 1 x A 1_{x}^{\mathcal{A}} 1 x A in Hom A 0 ( x , x ) \operatorname{Hom}_{\mathcal{A}}^{0}(x,x) Hom A 0 ( x , x ) . It is now easy to check that these isomorphisms also identify the maps m k x 0 , … , x k m_{k}^{x_{0},\ldots,x_{k}} m k x 0 , … , x k in A \mathcal{A} A and Fun A ∞ u n ( P u n , A ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) Fun A ∞ un ( P un , A ) . □ \square □
We will write T A [ 1 ] T\mathcal{A}[1] T A [ 1 ] for the counital dg-cocategory corresponding to A \mathcal{A} A . A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B are identified with suitable functors T A [ 1 ] → T B [ 1 ] T\mathcal{A}[1]\to T\mathcal{B}[1] T A [ 1 ] → T B [ 1 ] , and similarly for A ∞ A_\infty A ∞ -natural transformations.
There is a good notion of tensor product ⊗ \otimes ⊗ of counital dg-cocategories. Note that this does not give a definition of tensor product of A ∞ A_\infty A ∞ -categories, since ( T A [ 1 ] ) ⊗ ( T B [ 1 ] ) (T\mathcal{A}[1])\otimes(T\mathcal{B}[1]) ( T A [ 1 ]) ⊗ ( T B [ 1 ]) need not be of the form T C [ 1 ] T\mathcal{C}[1] T C [ 1 ] .
Let A 1 , … , A k , B \mathcal{A}_1,\ldots,\mathcal{A}_k,\mathcal{B} A 1 , … , A k , B be A ∞ A_\infty A ∞ -categories over R R R . Following Lyubashenko [64, 5.3], we can define an A ∞ A_\infty A ∞ -multifunctor A 1 × ⋯ × A k → B \mathcal{A}_1\times\cdots\times\mathcal{A}_k\to\mathcal{B} A 1 × ⋯ × A k → B to be a suitable kind of functor of counital dg-cocategories ( T A 1 [ 1 ] ) ⊗ ⋯ ⊗ ( T A k [ 1 ] ) → T B [ 1 ] (T\mathcal{A}_1[1])\otimes\cdots\otimes(T\mathcal{A}_k[1])\to T\mathcal{B}[1] ( T A 1 [ 1 ]) ⊗ ⋯ ⊗ ( T A k [ 1 ]) → T B [ 1 ] . When k = 1 k=1 k = 1 this recovers ordinary A ∞ A_\infty A ∞ -functors.
Given A ∞ A_\infty A ∞ -categories A , B , C \mathcal{A},\mathcal{B},\mathcal{C} A , B , C over R R R , Lyubashenko [64, §4, p. 14 & §6, p. 21] defines a functor of counital dg-cocategories
Λ A B C : T Fun A ∞ ( A , B ) [ 1 ] ⊗ T Fun A ∞ ( B , C ) [ 1 ] ⟶ T Fun A ∞ ( A , C ) [ 1 ] , (112) \Lambda_{\mathcal{A}\mathcal{B}\mathcal{C}}:T\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B})[1]\otimes T\operatorname{Fun}_{A_\infty}(\mathcal{B},\mathcal{C})[1]\longrightarrow T\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{C})[1], \tag*{(112)} Λ A BC : T Fun A ∞ ( A , B ) [ 1 ] ⊗ T Fun A ∞ ( B , C ) [ 1 ] ⟶ T Fun A ∞ ( A , C ) [ 1 ] , (112)
which is by definition an A ∞ A_\infty A ∞ -bifunctor (111) . These λ A B C , Λ A B C \lambda_{\mathcal{A}\mathcal{B}\mathcal{C}},\Lambda_{\mathcal{A}\mathcal{B}\mathcal{C}} λ A BC , Λ A BC have strict associativity and unit properties.
If A , B , C \mathcal{A},\mathcal{B},\mathcal{C} A , B , C are (cohomologically) unital, then λ A B C , Λ A B C \lambda_{\mathcal{A}\mathcal{B}\mathcal{C}},\Lambda_{\mathcal{A}\mathcal{B}\mathcal{C}} λ A BC , Λ A BC above restrict to bifunctors on Fun A ∞ u n ( ⋯ ) , Fun A ∞ c u ( ⋯ ) \operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\cdots),\operatorname{Fun}_{A_\infty}^{\mathrm{cu}}(\cdots) Fun A ∞ un ( ⋯ ) , Fun A ∞ cu ( ⋯ ) .
We now consider different notions of when two A ∞ A_\infty A ∞ -functors F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B are ‘the same’. The next definition follows Seidel [84, §1h], and is the analogue of homotopies of morphisms of A ∞ A_\infty A ∞ -algebras in Definition 3.9.
Definition 4.21. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_\infty A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be A ∞ A_\infty A ∞ -functors which act the same way on objects, F O b j = G O b j \mathcal{F}_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}} F Obj = G Obj . An A ∞ A_\infty A ∞ -homotopy , or just homotopy , H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G consists of for all k ≥ 1 k\geq1 k ≥ 1 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in A \mathcal{A} A , a graded R R R -linear map
H k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F O b j ( x 0 ) , F O b j ( x k ) ) , \mathcal{H}^{x_0,\ldots,x_k}_k:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{F}_{\mathrm{Obj}}(x_k)), H k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , F Obj ( x k )) , of degree − k -k − k , satisfying the analogue of (3.12): for all n ≥ 1 n\geq1 n ≥ 1 and objects x 0 , x 1 , … , x n x_0,x_1,\ldots,x_n x 0 , x 1 , … , x n in A \mathcal{A} A , we require that
F n x 0 , … , x n − G n x 0 , … , x n = ∑ 0 = l 0 < l 1 ≤ ⋯ ≤ l k = n 1 ≤ j ≤ k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ j − 1 i = 1 ( l i − l i − 1 + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) m k B ; F ( x l 0 ) , F ( x l 1 ) , … , F ( x l k ) ∘ [ F l 1 − l 0 x 0 , … , x l 1 ⊗ ⋯ ⊗ F l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ H l j − l j − 1 x l j − 1 , … , x l j ⊗ ⋯ ⊗ G l k − l k − 1 x l k − 1 , … , x l k ] (113) \mathcal{F}^{x_0,\ldots,x_n}_n-\mathcal{G}^{x_0,\ldots,x_n}_n
=
\sum_{\substack{0=l_0<l_1\leq\cdots\leq l_k=n\\1\leq j\leq k}}
(-1)^{k+1}\cdot(-1)^{\sum_{j-1}^{i=1}(l_i-l_{i-1}+1)}\cdot(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)}
m_k^{\mathcal{B};\mathcal{F}(x_{l_0}),\mathcal{F}(x_{l_1}),\ldots,\mathcal{F}(x_{l_k})}
\circ\left[\mathcal{F}^{x_0,\ldots,x_{l_1}}_{l_1-l_0}\otimes\cdots\otimes\mathcal{F}^{x_{l_{j-2}},\ldots,x_{l_{j-1}}}_{l_{j-1}-l_{j-2}}\otimes\mathcal{H}^{x_{l_{j-1}},\ldots,x_{l_j}}_{l_j-l_{j-1}}\otimes\cdots\otimes\mathcal{G}^{x_{l_{k-1}},\ldots,x_{l_k}}_{l_k-l_{k-1}}\right] \tag*{(113)} F n x 0 , … , x n − G n x 0 , … , x n = 0 = l 0 < l 1 ≤ ⋯ ≤ l k = n 1 ≤ j ≤ k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ j − 1 i = 1 ( l i − l i − 1 + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) m k B ; F ( x l 0 ) , F ( x l 1 ) , … , F ( x l k ) ∘ [ F l 1 − l 0 x 0 , … , x l 1 ⊗ ⋯ ⊗ F l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ H l j − l j − 1 x l j − 1 , … , x l j ⊗ ⋯ ⊗ G l k − l k − 1 x l k − 1 , … , x l k ] (113) + ∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k H i + k + 1 x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , x i + 1 , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) . +\sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}
(-1)^{i+jk}\mathcal{H}^{x_0,x_1,\ldots,x_i,x_{i+j},x_{i+j+1},\ldots,x_n}_{i+k+1}\circ\left(\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_i)}\otimes m_j^{\mathcal{A};x_i,x_{i+1},\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{n-1},x_n)}\right). + i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k H i + k + 1 x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , x i + 1 , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) .
When n = 1 n=1 n = 1 , equation (113) becomes
F 1 x 0 , x 1 − G 1 x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ H 1 x 0 , x 1 + H 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 . \mathfrak{F}_{1}^{x_{0},x_{1}}-\mathfrak{G}_{1}^{x_{0},x_{1}}=m_{1}^{\mathcal{B};\mathfrak{F}(x_{0}),\mathfrak{F}(x_{1})}\circ\mathfrak{H}_{1}^{x_{0},x_{1}}+\mathfrak{H}_{1}^{x_{0},x_{1}}\circ m_{1}^{\mathcal{A};x_{0},x_{1}}. F 1 x 0 , x 1 − G 1 x 0 , x 1 = m 1 B ; F ( x 0 ) , F ( x 1 ) ∘ H 1 x 0 , x 1 + H 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 .
This implies that H 1 x 0 , x 1 \mathfrak{H}_{1}^{x_{0},x_{1}} H 1 x 0 , x 1 is a chain homotopy between the morphisms of chain complexes F 1 x 0 , x 1 , G 1 x 0 , x 1 : ( Hom A ( x 0 , x 1 ) , m 1 A ; x 0 , x 1 ) ⟶ ( Hom B ( F ( x 0 ) , F ( x 1 ) ) , m 1 B ; F ( x 0 ) , F ( x 1 ) ) \mathfrak{F}_{1}^{x_{0},x_{1}},\mathfrak{G}_{1}^{x_{0},x_{1}}:(\operatorname{Hom}_{\mathcal{A}}(x_{0},x_{1}),m_{1}^{\mathcal{A};x_{0},x_{1}})\longrightarrow(\operatorname{Hom}_{\mathcal{B}}(\mathfrak{F}(x_{0}),\mathfrak{F}(x_{1})),m_{1}^{\mathcal{B};\mathfrak{F}(x_{0}),\mathfrak{F}(x_{1})}) F 1 x 0 , x 1 , G 1 x 0 , x 1 : ( Hom A ( x 0 , x 1 ) , m 1 A ; x 0 , x 1 ) ⟶ ( Hom B ( F ( x 0 ) , F ( x 1 )) , m 1 B ; F ( x 0 ) , F ( x 1 ) ) . Thus for the morphisms (4.9) on cohomology we have
H ∗ ( F 1 x 0 , x 1 ) = H ∗ ( G 1 x 0 , x 1 ) : H ∗ ( Hom A ( x 0 , x 1 ) ) ⟶ H ∗ ( Hom B ( F ( x 0 ) , F ( x 1 ) ) ) . H^{*}(\mathfrak{F}_{1}^{x_{0},x_{1}})=H^{*}(\mathfrak{G}_{1}^{x_{0},x_{1}}):H^{*}(\operatorname{Hom}_{\mathcal{A}}(x_{0},x_{1}))\longrightarrow H^{*}(\operatorname{Hom}_{\mathcal{B}}(\mathfrak{F}(x_{0}),\mathfrak{F}(x_{1}))). H ∗ ( F 1 x 0 , x 1 ) = H ∗ ( G 1 x 0 , x 1 ) : H ∗ ( Hom A ( x 0 , x 1 )) ⟶ H ∗ ( Hom B ( F ( x 0 ) , F ( x 1 ))) .
Hence in functors of cohomology categories we have
H 0 ( F 1 ) = H 0 ( G 1 ) : H 0 ( A ) ⟶ H 0 ( B ) . H^{0}(\mathfrak{F}_{1})=H^{0}(\mathfrak{G}_{1}):H^{0}(\mathcal{A})\longrightarrow H^{0}(\mathcal{B}). H 0 ( F 1 ) = H 0 ( G 1 ) : H 0 ( A ) ⟶ H 0 ( B ) .
If F , G \mathfrak{F},\mathfrak{G} F , G are unital A ∞ A_{\infty} A ∞ -functors of unital A ∞ A_{\infty} A ∞ -categories A , B \mathcal{A},\mathcal{B} A , B , we call a homotopy H : F ⇒ G \mathfrak{H}:\mathfrak{F}\Rightarrow\mathfrak{G} H : F ⇒ G unital if for all k ≥ 0 k\geq0 k ≥ 0 , i = 0 , … , k i=0,\ldots,k i = 0 , … , k , x 0 , … , x k ∈ Obj A x_{0},\ldots,x_{k}\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and a j ∈ Hom A ( x j − 1 , x j ) a_{j}\in\operatorname{Hom}_{\mathcal{A}}(x_{j-1},x_{j}) a j ∈ Hom A ( x j − 1 , x j ) we have
H k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0. (114) \mathfrak{H}_{k+1}^{x_{0},\ldots,x_{i},x_{i},\ldots,x_{k}}(a_{1}\otimes\cdots\otimes a_{i}\otimes1_{x_{i}}\otimes a_{i+1}\otimes\cdots\otimes a_{k})=0. \tag*{(114)} H k + 1 x 0 , … , x i , x i , … , x k ( a 1 ⊗ ⋯ ⊗ a i ⊗ 1 x i ⊗ a i + 1 ⊗ ⋯ ⊗ a k ) = 0. (114)
Definition 4.22. Let A , B \mathcal{A},\mathcal{B} A , B be A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathfrak{F},\mathfrak{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be A ∞ A_{\infty} A ∞ -functors. Using the notation of Definition 4.10, an A ∞ A_{\infty} A ∞ -natural transformation T : F ⇒ G \mathfrak{T}:\mathfrak{F}\Rightarrow\mathfrak{G} T : F ⇒ G is an element T ∈ Hom 0 ( F , G ) \mathfrak{T}\in\operatorname{Hom}^{0}(\mathfrak{F},\mathfrak{G}) T ∈ Hom 0 ( F , G ) with n 1 F , G ( T ) = 0 \mathfrak{n}_{1}^{\mathfrak{F},\mathfrak{G}}(\mathfrak{T})=0 n 1 F , G ( T ) = 0 . This is the most natural notion of morphism F → G \mathfrak{F}\to\mathfrak{G} F → G in the A ∞ A_{\infty} A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) .
Note that T \mathfrak{T} T is data ( T k ) k ≥ 0 (\mathfrak{T}_{k})_{k\geq0} ( T k ) k ≥ 0 , where T 0 \mathfrak{T}_{0} T 0 gives T 0 x ( 1 ) ∈ Hom B 0 ( F ( x ) , G ( x ) ) \mathfrak{T}_{0}^{x}(1)\in\operatorname{Hom}_{\mathcal{B}}^{0}(\mathfrak{F}(x),\mathfrak{G}(x)) T 0 x ( 1 ) ∈ Hom B 0 ( F ( x ) , G ( x )) for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A , and the n = 0 n=0 n = 0 equation in n 1 F , G ( T ) = 0 \mathfrak{n}_{1}^{\mathfrak{F},\mathfrak{G}}(\mathfrak{T})=0 n 1 F , G ( T ) = 0 from (4.15) gives m 1 B ; F ( x ) , G ( x ) ( T 0 x ( 1 ) ) = 0 m_{1}^{\mathcal{B};\mathfrak{F}(x),\mathfrak{G}(x)}(\mathfrak{T}_{0}^{x}(1))=0 m 1 B ; F ( x ) , G ( x ) ( T 0 x ( 1 )) = 0 . In cohomology this gives a morphism [ T 0 x ( 1 ) ] : H 0 ( F 1 ) ( x ) → H 0 ( G 1 ) ( x ) [\mathfrak{T}_{0}^{x}(1)]:H^{0}(\mathfrak{F}_{1})(x)\to H^{0}(\mathfrak{G}_{1})(x) [ T 0 x ( 1 )] : H 0 ( F 1 ) ( x ) → H 0 ( G 1 ) ( x ) in the (nonunital) category H 0 ( B ) H^{0}(\mathcal{B}) H 0 ( B ) .
The n = 0 n=0 n = 0 equation in n 1 F , G ( T ) = 0 \mathfrak{n}_{1}^{\mathfrak{F},\mathfrak{G}}(\mathfrak{T})=0 n 1 F , G ( T ) = 0 from (4.15) becomes
T 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 + m 1 B ; F ( x 0 ) , G ( x 1 ) ∘ T 1 x 0 , x 1 − m 2 B ; F ( x 0 ) , F ( x 1 ) , G ( x 1 ) ( F 1 x 0 , x 1 ⊗ T 0 x 1 ( 1 ) ) + m 2 B ; F ( x 0 ) , G ( x 0 ) , G ( x 1 ) ( T 0 x 0 ( 1 ) ⊗ G 1 x 0 , x 1 ) = 0. \mathfrak{T}_{1}^{x_{0},x_{1}}\circ m_{1}^{\mathcal{A};x_{0},x_{1}}+m_{1}^{\mathcal{B};\mathfrak{F}(x_{0}),\mathfrak{G}(x_{1})}\circ\mathfrak{T}_{1}^{x_{0},x_{1}}
-m_{2}^{\mathcal{B};\mathfrak{F}(x_{0}),\mathfrak{F}(x_{1}),\mathfrak{G}(x_{1})}(\mathfrak{F}_{1}^{x_{0},x_{1}}\otimes\mathfrak{T}_{0}^{x_{1}}(1))
+m_{2}^{\mathcal{B};\mathfrak{F}(x_{0}),\mathfrak{G}(x_{0}),\mathfrak{G}(x_{1})}(\mathfrak{T}_{0}^{x_{0}}(1)\otimes\mathfrak{G}_{1}^{x_{0},x_{1}})=0. T 1 x 0 , x 1 ∘ m 1 A ; x 0 , x 1 + m 1 B ; F ( x 0 ) , G ( x 1 ) ∘ T 1 x 0 , x 1 − m 2 B ; F ( x 0 ) , F ( x 1 ) , G ( x 1 ) ( F 1 x 0 , x 1 ⊗ T 0 x 1 ( 1 )) + m 2 B ; F ( x 0 ) , G ( x 0 ) , G ( x 1 ) ( T 0 x 0 ( 1 ) ⊗ G 1 x 0 , x 1 ) = 0.
By applying this to a cochain representative of a morphism α : x 0 → x 1 \alpha:x_{0}\to x_{1} α : x 0 → x 1 in H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) , we find the following commutes in the (nonunital) category H 0 ( B ) H^{0}(\mathcal{B}) H 0 ( B ) :
H 0 ( F 1 ) ( x 0 ) → [ T 0 x 0 ( 1 ) ] H 0 ( G 1 ) ( x 0 ) H 0 ( F 1 ) ( α ) ↓ ↓ H 0 ( G 1 ) ( α ) H 0 ( F 1 ) ( x 1 ) → [ T 0 x 1 ( 1 ) ] H 0 ( G 1 ) ( x 1 ) . \begin{CD}
H^{0}(\mathfrak{F}_{1})(x_{0}) @>{[\mathfrak{T}_{0}^{x_{0}}(1)]}>> H^{0}(\mathfrak{G}_{1})(x_{0}) \\
@V{H^{0}(\mathfrak{F}_{1})(\alpha)}VV @VV{H^{0}(\mathfrak{G}_{1})(\alpha)}V \\
H^{0}(\mathfrak{F}_{1})(x_{1}) @>{[\mathfrak{T}_{0}^{x_{1}}(1)]}>> H^{0}(\mathfrak{G}_{1})(x_{1}).
\end{CD} H 0 ( F 1 ) ( x 0 ) H 0 ( F 1 ) ( α ) ↓ ⏐ H 0 ( F 1 ) ( x 1 ) [ T 0 x 0 ( 1 )] [ T 0 x 1 ( 1 )] H 0 ( G 1 ) ( x 0 ) ↓ ⏐ H 0 ( G 1 ) ( α ) H 0 ( G 1 ) ( x 1 ) .
Thus, we can define a natural transformation H 0 ( T 0 ) : H 0 ( F 1 ) ⇒ H 0 ( G 1 ) H^{0}(\mathfrak{T}_{0}):H^{0}(\mathfrak{F}_{1})\Rightarrow H^{0}(\mathfrak{G}_{1}) H 0 ( T 0 ) : H 0 ( F 1 ) ⇒ H 0 ( G 1 ) of (nonunital) functors H 0 ( A ) → H 0 ( B ) H^{0}(\mathcal{A})\to H^{0}(\mathcal{B}) H 0 ( A ) → H 0 ( B ) by H 0 ( T 0 ) ( x ) = [ T 0 x ( 1 ) ] H^{0}(\mathfrak{T}_{0})(x)=[\mathfrak{T}_{0}^{x}(1)] H 0 ( T 0 ) ( x ) = [ T 0 x ( 1 )] .
Now suppose A , B , F , G \mathcal{A},\mathcal{B},\mathfrak{F},\mathfrak{G} A , B , F , G are unital, so that F , G \mathfrak{F},\mathfrak{G} F , G are objects in the A ∞ A_{\infty} A ∞ -subcategory Fun A ∞ u n ( A , B ) ⊂ Fun A ∞ ( A , B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\subset\operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) ⊂ Fun A ∞ ( A , B ) . We call T : F ⇒ G \mathfrak{T}:\mathfrak{F}\Rightarrow\mathfrak{G} T : F ⇒ G a unital A ∞ A_{\infty} A ∞ -natural transformation if T ∈ Hom A ∞ u n , 0 ( F , G ) ⊂ Hom A ∞ 0 ( F , G ) \mathfrak{T}\in\operatorname{Hom}_{A_{\infty}}^{\mathrm{un},0}(\mathfrak{F},\mathfrak{G})\subset\operatorname{Hom}_{A_{\infty}}^{0}(\mathfrak{F},\mathfrak{G}) T ∈ Hom A ∞ un , 0 ( F , G ) ⊂ Hom A ∞ 0 ( F , G ) . Then $H^{0}(\mathfrak{T}_{0}) is a natural transformation of honest categories and functors, not just nonunital categories and nonunital functors.
A unital A ∞ A_{\infty} A ∞ -natural transformation T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is called a (unital) A ∞ A_{\infty} A ∞ -natural isomorphism if it is a homotopy equivalence in the unital A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) , in the sense of Definition 4.8, that is, [ T ] [\mathcal{T}] [ T ] is an isomorphism in H 0 ( Fun A ∞ u n ( A , B ) ) H^{0}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ un ( A , B )) . Unital A ∞ A_{\infty} A ∞ -natural isomorphism is an equivalence relation on unital functors A → B \mathcal{A}\to\mathcal{B} A → B .
Similarly, suppose A , B , F , G \mathcal{A},\mathcal{B},\mathcal{F},\mathcal{G} A , B , F , G are cohomologically unital, so that F , G \mathcal{F},\mathcal{G} F , G are objects in the cohomologically unital A ∞ A_{\infty} A ∞ -category Fun A ∞ c u ( A , B ) \operatorname{Fun}^{\mathrm{cu}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ cu ( A , B ) . An A ∞ A_{\infty} A ∞ -natural transformation T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is called a (cohomologically unital) A ∞ A_{\infty} A ∞ -natural isomorphism if it is a homotopy equivalence in Fun A ∞ c u ( A , B ) \operatorname{Fun}^{\mathrm{cu}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ cu ( A , B ) .
Remark 4.23. Fukaya [29, Def. 8.1] calls an A ∞ A_{\infty} A ∞ -natural isomorphism T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G a homotopy equivalence between F \mathcal{F} F and G \mathcal{G} G . Lefèvre-Hasegawa [57, Def. 8.2.2.1] calls A ∞ A_{\infty} A ∞ -natural isomorphisms weak equivalences .
Lefèvre-Hasegawa [57, Prop. 8.2.2.3] proves:
Proposition 4.24. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be unital A ∞ A_{\infty} A ∞ -functors, and T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G be a unital A ∞ A_{\infty} A ∞ -natural transformation. Then T \mathcal{T} T is an A ∞ A_{\infty} A ∞ -natural isomorphism if and only if H 0 ( T 0 ) : H 0 ( F 1 ) ⇒ H 0 ( G 1 ) H^{0}(\mathcal{T}_{0}):H^{0}(\mathcal{F}_{1})\Rightarrow H^{0}(\mathcal{G}_{1}) H 0 ( T 0 ) : H 0 ( F 1 ) ⇒ H 0 ( G 1 ) is a natural isomorphism of ordinary functors.
The next proposition, outlined in Seidel [84, §2c], explains that A ∞ A_{\infty} A ∞ -homotopies are special A ∞ A_{\infty} A ∞ -natural isomorphisms. We give a detailed proof.
Proposition 4.25. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be unital A ∞ A_{\infty} A ∞ -functors which act the same way on objects, F O b j = G O b j \mathcal{F}_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}} F Obj = G Obj , and H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G be a unital A ∞ A_{\infty} A ∞ -homotopy. Define T ∈ Hom A ∞ u n , 0 ( F , G ) \mathcal{T}\in\operatorname{Hom}^{\mathrm{un},0}_{A_{\infty}}(\mathcal{F},\mathcal{G}) T ∈ Hom A ∞ un , 0 ( F , G ) by
T 0 x 0 : R ⟶ Hom B 0 ( F ( x 0 ) , G ( x 0 ) ) is given by T 0 x 0 : λ ⟼ λ 1 F ( x 0 ) , T k x 0 , … , x k = H k x 0 , … , x k , all k ≥ 1 and x 0 , … , x k ∈ Obj A . (115) \mathcal{T}_{0}^{x_{0}}:R\longrightarrow\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x_{0}),\mathcal{G}(x_{0}))\ \text{is given by}\ \mathcal{T}_{0}^{x_{0}}:\lambda\longmapsto\lambda1_{\mathcal{F}(x_{0})},
\qquad
\mathcal{T}_{k}^{x_{0},\ldots,x_{k}}=\mathcal{H}_{k}^{x_{0},\ldots,x_{k}},\quad\text{all }k\geq1\text{ and }x_{0},\ldots,x_{k}\in\operatorname{Obj}\mathcal{A}. \tag*{(115)} T 0 x 0 : R ⟶ Hom B 0 ( F ( x 0 ) , G ( x 0 )) is given by T 0 x 0 : λ ⟼ λ 1 F ( x 0 ) , T k x 0 , … , x k = H k x 0 , … , x k , all k ≥ 1 and x 0 , … , x k ∈ Obj A . (115) Then T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is an A ∞ A_{\infty} A ∞ -natural isomorphism. This gives a 1-1 correspondence between unital A ∞ A_{\infty} A ∞ -homotopies and A ∞ A_{\infty} A ∞ -natural isomorphisms T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G with T 0 x 0 : λ ↦ λ 1 F ( x 0 ) \mathcal{T}_{0}^{x_{0}}:\lambda\mapsto\lambda1_{\mathcal{F}(x_{0})} T 0 x 0 : λ ↦ λ 1 F ( x 0 ) for all x 0 ∈ Obj A x_{0}\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A .
Proof. We claim that for T \mathcal{T} T as in (4.29), the equation n 1 F , G ( T ) = 0 n_{1}^{\mathcal{F},\mathcal{G}}(\mathcal{T})=0 n 1 F , G ( T ) = 0 from (101) is equivalent to (4.27). This holds as n 1 F , G ( T ) n_{1}^{\mathcal{F},\mathcal{G}}(\mathcal{T}) n 1 F , G ( T ) in (4.15) is the sum of terms from T k x 0 , … , x k = H k x 0 , … , x k \mathcal{T}_{k}^{x_{0},\ldots,x_{k}}=\mathcal{H}_{k}^{x_{0},\ldots,x_{k}} T k x 0 , … , x k = H k x 0 , … , x k for k ≥ 1 k\geq1 k ≥ 1 , which is just the right hand side of (4.27), and terms in T 0 x 0 \mathcal{T}_{0}^{x_{0}} T 0 x 0 . The terms in T 0 x 0 \mathcal{T}_{0}^{x_{0}} T 0 x 0 come only from the first sum in (4.15), and as B \mathcal{B} B is unital, when T 0 x 0 = 1 F ( x 0 ) \mathcal{T}_{0}^{x_{0}}=1_{\mathcal{F}(x_{0})} T 0 x 0 = 1 F ( x 0 ) is substituted into m k B ; F ( x 0 ) , … , F ( x n ) m_{k}^{\mathcal{B};\mathcal{F}(x_{0}),\ldots,\mathcal{F}(x_{n})} m k B ; F ( x 0 ) , … , F ( x n ) we get zero unless k = 2 k=2 k = 2 . Thus there are only two possibly nonzero terms in T 0 x 0 \mathcal{T}_{0}^{x_{0}} T 0 x 0 , which come from k = 2 k=2 k = 2 , j = 1 j=1 j = 1 , l 0 = l 1 = 0 l_{0}=l_{1}=0 l 0 = l 1 = 0 , l 2 = n l_{2}=n l 2 = n and from k = 2 k=2 k = 2 , j = 2 j=2 j = 2 , l 0 = 0 l_{0}=0 l 0 = 0 , l 1 = l 2 = n l_{1}=l_{2}=n l 1 = l 2 = n in the first sum in (4.15), which are
m 2 B ; F ( x 0 ) , F ( x 1 ) ∘ ( T 0 x 0 ( 1 ) ⊗ G n x 0 , … , x n ) − m 2 B ; F ( x 0 ) , F ( x 1 ) ∘ ( F n x 0 , … , x n ⊗ T 0 x n ( 1 ) ) = G n x 0 , … , x n − F n x 0 , … , x n , m_{2}^{\mathcal{B};\mathcal{F}(x_{0}),\mathcal{F}(x_{1})}\circ\left(\mathcal{T}_{0}^{x_{0}}(1)\otimes\mathcal{G}_{n}^{x_{0},\ldots,x_{n}}\right)-m_{2}^{\mathcal{B};\mathcal{F}(x_{0}),\mathcal{F}(x_{1})}\circ\left(\mathcal{F}_{n}^{x_{0},\ldots,x_{n}}\otimes\mathcal{T}_{0}^{x_{n}}(1)\right)
=\mathcal{G}_{n}^{x_{0},\ldots,x_{n}}-\mathcal{F}_{n}^{x_{0},\ldots,x_{n}}, m 2 B ; F ( x 0 ) , F ( x 1 ) ∘ ( T 0 x 0 ( 1 ) ⊗ G n x 0 , … , x n ) − m 2 B ; F ( x 0 ) , F ( x 1 ) ∘ ( F n x 0 , … , x n ⊗ T 0 x n ( 1 ) ) = G n x 0 , … , x n − F n x 0 , … , x n ,
which is minus the right hand side of (4.27). Hence (4.27) for H \mathcal{H} H is equivalent to n 1 F ~ , G ( T ) = 0 n_1^{\tilde{\mathcal{F}},\mathcal{G}}(\mathcal{T})=0 n 1 F ~ , G ( T ) = 0 , and T \mathcal{T} T is an A ∞ A_\infty A ∞ -natural transformation.
We will prove directly that T \mathcal{T} T is an A ∞ A_\infty A ∞ -natural isomorphism. First we claim that there is a unique U ∈ Hom A ∞ u n , 0 ( G , F ~ ) \mathcal{U}\in\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\mathcal{G},\tilde{\mathcal{F}}) U ∈ Hom A ∞ un , 0 ( G , F ~ ) such that U 0 x 0 : λ ↦ λ 1 F ~ ( x 0 ) \mathcal{U}^{x_0}_0:\lambda\mapsto\lambda1_{\tilde{\mathcal{F}}(x_0)} U 0 x 0 : λ ↦ λ 1 F ~ ( x 0 ) for all x 0 ∈ Obj A x_0\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A and
n 2 F ~ , G , F ~ ( T ⊗ U ) = 1 F ~ in Hom A ∞ u n , 0 ( F ~ , F ~ ) . (116) n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes\mathcal{U})=1_{\tilde{\mathcal{F}}}\quad\text{in }\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\tilde{\mathcal{F}},\tilde{\mathcal{F}}). \tag*{(116)} n 2 F ~ , G , F ~ ( T ⊗ U ) = 1 F ~ in Hom A ∞ un , 0 ( F ~ , F ~ ) . (116)
This holds as by (4.15), the ( ⋯ ) n x 0 , … , x n (\cdots)^{x_0,\ldots,x_n}_n ( ⋯ ) n x 0 , … , x n component in (4.30) is of the form
m 2 F ~ ( x 0 ) , F ~ ( x 0 ) , F ~ ( x n ) ( T 0 x 0 ( 1 ) ⊗ U n x 0 , … , x n ) + ( terms in U k x 0 ′ , … , x k ′ for k < n ) = { 1 F ~ ( x 0 ) , n = 0 , 0 , n > 0. (117) m_2^{\tilde{\mathcal{F}}(x_0),\tilde{\mathcal{F}}(x_0),\tilde{\mathcal{F}}(x_n)}
\left(\mathcal{T}^{x_0}_0(1)\otimes\mathcal{U}^{x_0,\ldots,x_n}_n\right)+\left(\text{terms in }\mathcal{U}^{x'_0,\ldots,x'_k}_k\text{ for }k<n\right)
=
\begin{cases}
1_{\tilde{\mathcal{F}}(x_0)}, & n=0,\\
0, & n>0.
\end{cases} \tag*{(117)} m 2 F ~ ( x 0 ) , F ~ ( x 0 ) , F ~ ( x n ) ( T 0 x 0 ( 1 ) ⊗ U n x 0 , … , x n ) + ( terms in U k x 0 ′ , … , x k ′ for k < n ) = { 1 F ~ ( x 0 ) , 0 , n = 0 , n > 0. (117)
But T 0 x 0 ( 1 ) = 1 F ~ ( x 0 ) \mathcal{T}^{x_0}_0(1)=1_{\tilde{\mathcal{F}}(x_0)} T 0 x 0 ( 1 ) = 1 F ~ ( x 0 ) , so m 2 F ~ ( x 0 ) , F ~ ( x 0 ) , F ~ ( x n ) ( T 0 x 0 ( 1 ) ⊗ U n x 0 , … , x n ) = U n x 0 , … , x n m_2^{\tilde{\mathcal{F}}(x_0),\tilde{\mathcal{F}}(x_0),\tilde{\mathcal{F}}(x_n)}(\mathcal{T}^{x_0}_0(1)\otimes\mathcal{U}^{x_0,\ldots,x_n}_n)=\mathcal{U}^{x_0,\ldots,x_n}_n m 2 F ~ ( x 0 ) , F ~ ( x 0 ) , F ~ ( x n ) ( T 0 x 0 ( 1 ) ⊗ U n x 0 , … , x n ) = U n x 0 , … , x n . Thus we can use (4.31) to determine U n x 0 , … , x n \mathcal{U}^{x_0,\ldots,x_n}_n U n x 0 , … , x n uniquely by induction on n = 0 , 1 , 2 , … n=0,1,2,\ldots n = 0 , 1 , 2 , … .
The A ∞ A_\infty A ∞ -equation (4.1) for n = 2 n=2 n = 2 and objects F ~ , G , F ~ \tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}} F ~ , G , F ~ applied to T ⊗ U \mathcal{T}\otimes\mathcal{U} T ⊗ U yields
0 = n 1 F ~ , F ~ ∘ n 2 F ~ , G , F ~ ( T ⊗ U ) − n 2 F ~ , F ~ ( n 1 F ~ , G ( T ) ⊗ U ) − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U ) ) = n 1 F ~ , F ~ ( 1 F ~ ) − 0 − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U ) ) = − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U ) ) . \begin{aligned}
0={}&n_1^{\tilde{\mathcal{F}},\tilde{\mathcal{F}}}\circ n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes\mathcal{U})-n_2^{\tilde{\mathcal{F}},\tilde{\mathcal{F}}} (n_1^{\tilde{\mathcal{F}},\mathcal{G}}(\mathcal{T})\otimes\mathcal{U})\\
&-n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{U}))\\
={}&n_1^{\tilde{\mathcal{F}},\tilde{\mathcal{F}}}(1_{\tilde{\mathcal{F}}})-0-n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{U}))\\
={}&-n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{U})).
\end{aligned} 0 = = = n 1 F ~ , F ~ ∘ n 2 F ~ , G , F ~ ( T ⊗ U ) − n 2 F ~ , F ~ ( n 1 F ~ , G ( T ) ⊗ U ) − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U )) n 1 F ~ , F ~ ( 1 F ~ ) − 0 − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U )) − n 2 F ~ , G , F ~ ( T ⊗ n 1 G , F ~ ( U )) .
Hence the uniqueness argument for U \mathcal{U} U above shows that n 1 G , F ~ ( U ) = 0 n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{U})=0 n 1 G , F ~ ( U ) = 0 .
Similarly, there is a unique S ∈ Hom A ∞ u n , 0 ( G , F ~ ) \mathcal{S}\in\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\mathcal{G},\tilde{\mathcal{F}}) S ∈ Hom A ∞ un , 0 ( G , F ~ ) such that S 0 x 0 : λ ↦ λ 1 G ( x 0 ) \mathcal{S}^{x_0}_0:\lambda\mapsto\lambda1_{\mathcal{G}(x_0)} S 0 x 0 : λ ↦ λ 1 G ( x 0 ) for all x 0 ∈ Obj A x_0\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A and n 1 G , F ~ ( S ) = 0 n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{S})=0 n 1 G , F ~ ( S ) = 0 and
n 2 G , F ~ , G ( S ⊗ T ) = 1 G in Hom A ∞ u n , 0 ( G , G ) . n_2^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G}}(\mathcal{S}\otimes\mathcal{T})=1_{\mathcal{G}}\quad\text{in }\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\mathcal{G},\mathcal{G}). n 2 G , F ~ , G ( S ⊗ T ) = 1 G in Hom A ∞ un , 0 ( G , G ) .
Now equation (4.1) for n = 3 n=3 n = 3 and objects G , F ~ , G , F ~ \mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}} G , F ~ , G , F ~ applied to S ⊗ T ⊗ U \mathcal{S}\otimes\mathcal{T}\otimes\mathcal{U} S ⊗ T ⊗ U yields
0 = n 1 G , F ~ ∘ n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ U ) + n 2 G , G , F ~ ( n 2 G , F ~ , G ( S ⊗ T ) ⊗ U ) − n 2 G , F ~ , F ~ ( S ⊗ n 2 F ~ , G , F ~ ( T ⊗ U ) ) + n 3 G , F ~ , G , F ~ ( n 1 G , F ~ ( S ) ⊗ T ⊗ U ) + n 3 G , F ~ , G , F ~ ( S ⊗ n 1 F ~ , G ( T ) ⊗ U ) + n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ n 1 G , F ~ ( U ) ) = n 1 G , F ~ ∘ n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ U ) + U − S + 0 + 0 + 0. (118) \begin{aligned}
0={}&n_1^{\mathcal{G},\tilde{\mathcal{F}}}\circ n_3^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{S}\otimes\mathcal{T}\otimes\mathcal{U})\\
&+n_2^{\mathcal{G},\mathcal{G},\tilde{\mathcal{F}}}\left(n_2^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G}}(\mathcal{S}\otimes\mathcal{T})\otimes\mathcal{U}\right)\\
&-n_2^{\mathcal{G},\tilde{\mathcal{F}},\tilde{\mathcal{F}}}\left(\mathcal{S}\otimes n_2^{\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{T}\otimes\mathcal{U})\right)\\
&+n_3^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}\left(n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{S})\otimes\mathcal{T}\otimes\mathcal{U}\right)\\
&+n_3^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}\left(\mathcal{S}\otimes n_1^{\tilde{\mathcal{F}},\mathcal{G}}(\mathcal{T})\otimes\mathcal{U}\right)\\
&+n_3^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}\left(\mathcal{S}\otimes\mathcal{T}\otimes n_1^{\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{U})\right)\\
={}&n_1^{\mathcal{G},\tilde{\mathcal{F}}}\circ n_3^{\mathcal{G},\tilde{\mathcal{F}},\mathcal{G},\tilde{\mathcal{F}}}(\mathcal{S}\otimes\mathcal{T}\otimes\mathcal{U})+\mathcal{U}-\mathcal{S}+0+0+0. \tag*{(118)}
\end{aligned} 0 = = n 1 G , F ~ ∘ n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ U ) + n 2 G , G , F ~ ( n 2 G , F ~ , G ( S ⊗ T ) ⊗ U ) − n 2 G , F ~ , F ~ ( S ⊗ n 2 F ~ , G , F ~ ( T ⊗ U ) ) + n 3 G , F ~ , G , F ~ ( n 1 G , F ~ ( S ) ⊗ T ⊗ U ) + n 3 G , F ~ , G , F ~ ( S ⊗ n 1 F ~ , G ( T ) ⊗ U ) + n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ n 1 G , F ~ ( U ) ) n 1 G , F ~ ∘ n 3 G , F ~ , G , F ~ ( S ⊗ T ⊗ U ) + U − S + 0 + 0 + 0. (118)
In the cohomology category H 0 ( Fun A ∞ u n ( A , B ) ) H^0(\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ un ( A , B )) we now have morphisms [ T ] : F ~ → G [\mathcal{T}]:\tilde{\mathcal{F}}\to\mathcal{G} [ T ] : F ~ → G and [ S ] , [ U ] : G → F ~ [\mathcal{S}],[\mathcal{U}]:\mathcal{G}\to\tilde{\mathcal{F}} [ S ] , [ U ] : G → F ~ with [ T ] ∘ [ U ] = 1 F ~ [\mathcal{T}]\circ[\mathcal{U}]=1_{\tilde{\mathcal{F}}} [ T ] ∘ [ U ] = 1 F ~ and [ S ] ∘ [ T ] = 1 G [\mathcal{S}]\circ[\mathcal{T}]=1_{\mathcal{G}} [ S ] ∘ [ T ] = 1 G . But (4.32) shows that [ S ] = [ U ] [\mathcal{S}]=[\mathcal{U}] [ S ] = [ U ] . Thus [ S ] [\mathcal{S}] [ S ] is the inverse morphism to [ T ] [\mathcal{T}] [ T ] , so [ T ] [\mathcal{T}] [ T ] is an isomorphism, and T \mathcal{T} T is an A ∞ A_\infty A ∞ -natural isomorphism. □
Remark 4.26. Both A ∞ A_\infty A ∞ -homotopies, and A ∞ A_\infty A ∞ -natural transformations, are defined for non-unital A ∞ A_\infty A ∞ -categories. However, the relation in Proposition 4.25 between A ∞ A_\infty A ∞ -homotopies and A ∞ A_\infty A ∞ -natural transformations only makes sense for unital A ∞ A_\infty A ∞ -categories.
The next definition follows [29, Def. 8.5] and [57, Def. 9.2.0.1].
Definition 4.27. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_\infty A ∞ -categories over R R R , and F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a unital A ∞ A_\infty A ∞ -functor. We call F \mathcal{F} F a (unital) homotopy equivalence if there exists a unital A ∞ A_\infty A ∞ -functor G : B → A \mathcal{G}:\mathcal{B}\to\mathcal{A} G : B → A and unital A ∞ A_\infty A ∞ -natural isomorphisms S : G ∘ F ⇒ id A \mathcal{S}:\mathcal{G}\circ\mathcal{F}\Rightarrow\operatorname{id}_{\mathcal{A}} S : G ∘ F ⇒ id A and T : F ∘ G ⇒ id B \mathcal{T}:\mathcal{F}\circ\mathcal{G}\Rightarrow\operatorname{id}_{\mathcal{B}} T : F ∘ G ⇒ id B . Then we also call A , B \mathcal{A},\mathcal{B} A , B (unital) homotopy equivalent .
Similarly, if A , B \mathcal{A},\mathcal{B} A , B are cohomologically unital A ∞ A_\infty A ∞ -categories over R R R , and F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B is a cohomologically unital A ∞ A_\infty A ∞ -functor, we call F \mathcal{F} F a (cohomologically unital) homotopy equivalence if there exists a cohomologically unital A ∞ A_\infty A ∞ -functor G : B → A \mathcal{G}:\mathcal{B}\to\mathcal{A} G : B → A and A ∞ A_\infty A ∞ -natural isomorphisms S : G ∘ F ⇒ id A \mathcal{S}:\mathcal{G}\circ\mathcal{F}\Rightarrow\operatorname{id}_{\mathcal{A}} S : G ∘ F ⇒ id A and T : F ∘ G ⇒ id B \mathcal{T}:\mathcal{F}\circ\mathcal{G}\Rightarrow\operatorname{id}_{\mathcal{B}} T : F ∘ G ⇒ id B . Then we also call A , B \mathcal{A},\mathcal{B} A , B (cohomologically unital) homotopy equivalent .
Compositions of homotopy equivalences are homotopy equivalences. Homotopy equivalence is an equivalence relation on unital, or cohomologically unital, A ∞ A_\infty A ∞ -categories.
The next definition follows [12, Def. 1.13] and [73, §1.4].
Definition 4.28. Let A , B \mathcal{A},\mathcal{B} A , B be unital, or cohomologically unital, A ∞ A_\infty A ∞ -categories over a commutative ring R R R , and F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a unital, or cohomologically unital, A ∞ A_\infty A ∞ -functor. We call F \mathcal{F} F a quasi-equivalence if:
(a) For all objects x , y x,y x , y in A \mathcal{A} A and k ∈ Z k\in\mathbb{Z} k ∈ Z , the following is an isomorphism
H k ( F 1 x , y ) : H k ( Hom A ( x , y ) ) ⟶ H k ( Hom B ( F ( x ) , F ( y ) ) ) . H^k(\mathcal{F}_1^{x,y}):H^k(\operatorname{Hom}_{\mathcal{A}}(x,y))\longrightarrow H^k(\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x),\mathcal{F}(y))). H k ( F 1 x , y ) : H k ( Hom A ( x , y )) ⟶ H k ( Hom B ( F ( x ) , F ( y ))) . (b) H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) H^0(\mathcal{F}_1):H^0(\mathcal{A})\to H^0(\mathcal{B}) H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) is an equivalence of categories.
All homotopy equivalences are quasi-equivalences. Theorem 4.45 below gives sufficient conditions for a quasi-equivalence to be a homotopy equivalence.
If R R R is a field, and in the unital case A , B \mathcal{A},\mathcal{B} A , B do not have zero objects (objects z z z with 1 z = 0 1_z=0 1 z = 0 ), then homotopy equivalences and quasi-equivalences coincide.
Remark 4.29. Homotopy equivalence and quasi-equivalence are two competing notions of when two (cohomologically) unital A ∞ A_\infty A ∞ -categories are ‘the same’. One could argue that for applications — e.g. in comparing two constructions of Fukaya category for a symplectic manifold — homotopy equivalence is better. However, there is a consensus in the mathematical community (see for example §4.10 and [11, 12, 24, 72, 73, 76, 88]) that to form localized (∞ \infty ∞ )-categories of A ∞ A_\infty A ∞ -categories, one should invert quasi-equivalences, not homotopy equivalences. One reason for this is that in the Tabuada model structure on dg-categories (see Theorem 4.35) the weak equivalences are quasi-equivalences.
One lesson from our results in §9.2 is that localizing homotopy equivalences of A ∞ A_\infty A ∞ -categories is also an interesting and fruitful thing to do.
Definitions 4.10–4.11 defined A ∞ A_\infty A ∞ -categories Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) , Fun A ∞ c u ( A , B ) \operatorname{Fun}^{\mathrm{cu}}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ cu ( A , B ) , Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) , and Definition 4.13 explained their functoriality under A ∞ A_\infty A ∞ -functors in A , B \mathcal{A},\mathcal{B} A , B . We now consider their functoriality under A ∞ A_\infty A ∞ -natural transformations and A ∞ A_\infty A ∞ -natural isomorphisms. The only reference the author can find to the content of Definition 4.30 in the literature is Lyubashenko [64, Prop. 8.4], who appears to claim that equivalence classes [ K ∗ ] [\mathcal{K}^{*}] [ K ∗ ] of K ∗ : ( G 1 ) ∗ ⇒ ( G 2 ) ∗ \mathcal{K}^{*}:(\mathcal{G}_{1})_{*}\Rightarrow(\mathcal{G}_{2})_{*} K ∗ : ( G 1 ) ∗ ⇒ ( G 2 ) ∗ in part (b) below are part of a strict 2-functor Fun A ∞ ( A , − ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A},-) Fun A ∞ ( A , − ) defined on 2-categorical enhancements of A ∞ Cat R , A ∞ Cat R c u A_{\infty}\operatorname{Cat}_{R},A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}} A ∞ Cat R , A ∞ Cat R cu .
Definition 4.30. (a) Let R R R be a commutative ring, and A 1 , A 2 , B \mathcal{A}_{1},\mathcal{A}_{2},\mathcal{B} A 1 , A 2 , B be A ∞ A_{\infty} A ∞ -categories over R R R , and E 1 , E 2 : A 1 → A 2 \mathcal{E}_{1},\mathcal{E}_{2}:\mathcal{A}_{1}\to\mathcal{A}_{2} E 1 , E 2 : A 1 → A 2 be A ∞ A_{\infty} A ∞ -functors, and K : E 1 ⇒ E 2 \mathcal{K}:\mathcal{E}_{1}\Rightarrow\mathcal{E}_{2} K : E 1 ⇒ E 2 be an A ∞ A_{\infty} A ∞ -natural transformation. Definition 4.13 constructed A ∞ A_{\infty} A ∞ -functors E 1 ∗ , E 2 ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) \mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*}:\operatorname{Fun}_{A_{\infty}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}(\mathcal{A}_{1},\mathcal{B}) E 1 ∗ , E 2 ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) . We will define an A ∞ A_{\infty} A ∞ -natural transformation K ∗ : E 1 ∗ ⇒ E 2 ∗ \mathcal{K}^{*}:\mathcal{E}_{1}^{*}\Rightarrow\mathcal{E}_{2}^{*} K ∗ : E 1 ∗ ⇒ E 2 ∗ . To do this, for all p ≥ 1 p\geq1 p ≥ 1 and A ∞ A_{\infty} A ∞ -functors F ~ 0 , F ~ 1 , … , F ~ p : A 2 → B \widetilde{\mathcal{F}}_{0},\widetilde{\mathcal{F}}_{1},\ldots,\widetilde{\mathcal{F}}_{p}:\mathcal{A}_{2}\to\mathcal{B} F 0 , F 1 , … , F p : A 2 → B , define a graded R R R -linear map of degree − p -p − p
( K ∗ ) p F ~ 0 , … , F ~ p : Hom A ∞ ( F ~ 0 , F ~ 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F ~ p − 1 , F ~ p ) ⟶ Hom A ∞ ( F ~ 0 ∘ E 1 , F ~ p ∘ E 2 ) (\mathcal{K}^{*})_{p}^{\widetilde{\mathcal{F}}_{0},\ldots,\widetilde{\mathcal{F}}_{p}}:\operatorname{Hom}_{A_{\infty}}(\widetilde{\mathcal{F}}_{0},\widetilde{\mathcal{F}}_{1})\otimes\cdots\otimes\operatorname{Hom}_{A_{\infty}}(\widetilde{\mathcal{F}}_{p-1},\widetilde{\mathcal{F}}_{p})\longrightarrow\operatorname{Hom}_{A_{\infty}}(\widetilde{\mathcal{F}}_{0}\circ\mathcal{E}_{1},\widetilde{\mathcal{F}}_{p}\circ\mathcal{E}_{2}) ( K ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 ∘ E 1 , F p ∘ E 2 )
by ( K ∗ ) p F ~ 0 , … , F ~ p = 0 (\mathcal{K}^{*})_{p}^{\widetilde{\mathcal{F}}_{0},\ldots,\widetilde{\mathcal{F}}_{p}}=0 ( K ∗ ) p F 0 , … , F p = 0 if p ≠ 1 p\ne1 p = 1 , and in a similar way to (4.20)
( ( K ∗ ) 1 F ~ 0 , F ~ 1 ( T ) ) n x 0 , … , x n = ∑ 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ j ≤ k : l i − 1 < l i i f i ≠ j ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 j − 1 ( l i − l i − 1 + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) ⋅ K k E 1 ( x l 0 ) , E 1 ( x l 1 ) , … , E 1 ( x l j − 1 ) , E 2 ( x l j ) , E 2 ( x l j + 1 ) , … , E 2 ( x l k ) ⋅ ( ( E 1 ) l 1 − l 0 x l 0 , … , x l 1 ⊗ ( E 1 ) l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ ( E 1 ) l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ K l j − l j − 1 x l j − 1 , … , x l j ⊗ ( E 2 ) l j + 1 − l j x l j , … , x l j + 1 ⊗ ⋯ ⊗ ( E 2 ) l k − l k − 1 x l k − 1 , … , x l k ) . (119) \begin{aligned}
((\mathcal{K}^{*})_{1}^{\widetilde{\mathcal{F}}_{0},\widetilde{\mathcal{F}}_{1}}(\mathcal{T}))_{n}^{x_{0},\ldots,x_{n}}
={}&\sum_{\substack{0=l_{0}\leq l_{1}\leq\cdots\leq l_{k}=n,\\1\leq j\leq k:\\l_{i-1}<l_{i}\ \mathrm{if}\ i\ne j}}
(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{j-1}(l_{i}-l_{i-1}+1)}\cdot(-1)^{\sum_{i=1}^{k}(l_{i}-l_{i-1}+1)(k-i)}\\
&\qquad\cdot\mathcal{K}_{k}^{\mathcal{E}_{1}(x_{l_{0}}),\mathcal{E}_{1}(x_{l_{1}}),\ldots,\mathcal{E}_{1}(x_{l_{j-1}}),\mathcal{E}_{2}(x_{l_{j}}),\mathcal{E}_{2}(x_{l_{j+1}}),\ldots,\mathcal{E}_{2}(x_{l_{k}})}\\
&\qquad\cdot\left((\mathcal{E}_{1})_{l_{1}-l_{0}}^{x_{l_{0}},\ldots,x_{l_{1}}}\otimes(\mathcal{E}_{1})_{l_{2}-l_{1}}^{x_{l_{1}},\ldots,x_{l_{2}}}\otimes\cdots\otimes(\mathcal{E}_{1})_{l_{j-1}-l_{j-2}}^{x_{l_{j-2}},\ldots,x_{l_{j-1}}}\right.\\
&\qquad\left.\otimes\mathcal{K}_{l_{j}-l_{j-1}}^{x_{l_{j-1}},\ldots,x_{l_{j}}}\otimes(\mathcal{E}_{2})_{l_{j+1}-l_{j}}^{x_{l_{j}},\ldots,x_{l_{j+1}}}\otimes\cdots\otimes(\mathcal{E}_{2})_{l_{k}-l_{k-1}}^{x_{l_{k-1}},\ldots,x_{l_{k}}}\right). \tag*{(119)}
\end{aligned} (( K ∗ ) 1 F 0 , F 1 ( T ) ) n x 0 , … , x n = 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , 1 ≤ j ≤ k : l i − 1 < l i if i = j ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 j − 1 ( l i − l i − 1 + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) ⋅ K k E 1 ( x l 0 ) , E 1 ( x l 1 ) , … , E 1 ( x l j − 1 ) , E 2 ( x l j ) , E 2 ( x l j + 1 ) , … , E 2 ( x l k ) ⋅ ( ( E 1 ) l 1 − l 0 x l 0 , … , x l 1 ⊗ ( E 1 ) l 2 − l 1 x l 1 , … , x l 2 ⊗ ⋯ ⊗ ( E 1 ) l j − 1 − l j − 2 x l j − 2 , … , x l j − 1 ⊗ K l j − l j − 1 x l j − 1 , … , x l j ⊗ ( E 2 ) l j + 1 − l j x l j , … , x l j + 1 ⊗ ⋯ ⊗ ( E 2 ) l k − l k − 1 x l k − 1 , … , x l k ) . (119)
We can prove using the identity n 1 E 1 , E 2 ( K ) = 0 \mathfrak{n}_{1}^{\mathcal{E}_{1},\mathcal{E}_{2}}(\mathcal{K})=0 n 1 E 1 , E 2 ( K ) = 0 that n 1 E 1 ∗ , E 2 ∗ ( K ∗ ) = 0 \mathfrak{n}_{1}^{\mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*}}(\mathcal{K}^{*})=0 n 1 E 1 ∗ , E 2 ∗ ( K ∗ ) = 0 , so K ∗ : E 1 ∗ ⇒ E 2 ∗ \mathcal{K}^{*}:\mathcal{E}_{1}^{*}\Rightarrow\mathcal{E}_{2}^{*} K ∗ : E 1 ∗ ⇒ E 2 ∗ is an A ∞ A_{\infty} A ∞ -natural transformation.
If A 1 , A 2 , B , E 1 , E 2 \mathcal{A}_{1},\mathcal{A}_{2},\mathcal{B},\mathcal{E}_{1},\mathcal{E}_{2} A 1 , A 2 , B , E 1 , E 2 are cohomologically unital then E 1 ∗ , E 2 ∗ \mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*} E 1 ∗ , E 2 ∗ restrict to E 1 ∗ , E 2 ∗ : Fun A ∞ c u ( A 2 , B ) → Fun A ∞ c u ( A 1 , B ) \mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*}:\operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A}_{1},\mathcal{B}) E 1 ∗ , E 2 ∗ : Fun A ∞ cu ( A 2 , B ) → Fun A ∞ cu ( A 1 , B ) , where Fun A ∞ c u ( A i , B ) ⊂ Fun A ∞ ( A i , B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A}_{i},\mathcal{B})\subset\operatorname{Fun}_{A_{\infty}}(\mathcal{A}_{i},\mathcal{B}) Fun A ∞ cu ( A i , B ) ⊂ Fun A ∞ ( A i , B ) is the full A ∞ A_{\infty} A ∞ -subcategory with objects cohomologically unital A ∞ A_{\infty} A ∞ -functors, and K ∗ \mathcal{K}^{*} K ∗ above restricts to an A ∞ A_{\infty} A ∞ -natural transformation of A ∞ A_{\infty} A ∞ -functors Fun A ∞ c u ( A 2 , B ) → Fun A ∞ c u ( A 1 , B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A}_{1},\mathcal{B}) Fun A ∞ cu ( A 2 , B ) → Fun A ∞ cu ( A 1 , B ) by taking F ~ 0 , … , F ~ p \widetilde{\mathcal{F}}_{0},\ldots,\widetilde{\mathcal{F}}_{p} F 0 , … , F p above to be cohomologically unital.
If A 1 , A 2 , B , E 1 , E 2 , K \mathcal{A}_{1},\mathcal{A}_{2},\mathcal{B},\mathcal{E}_{1},\mathcal{E}_{2},\mathcal{K} A 1 , A 2 , B , E 1 , E 2 , K are unital then E 1 ∗ , E 2 ∗ \mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*} E 1 ∗ , E 2 ∗ restrict to E 1 ∗ , E 2 ∗ : Fun A ∞ u n ( A 2 , B ) → Fun A ∞ u n ( A 1 , B ) \mathcal{E}_{1}^{*},\mathcal{E}_{2}^{*}:\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{1},\mathcal{B}) E 1 ∗ , E 2 ∗ : Fun A ∞ un ( A 2 , B ) → Fun A ∞ un ( A 1 , B ) . We replace Hom A ∞ ( ⋯ ) \operatorname{Hom}_{A_{\infty}}(\cdots) Hom A ∞ ( ⋯ ) by Hom u n ( ⋯ ) \operatorname{Hom}^{\mathrm{un}}(\cdots) Hom un ( ⋯ ) , and take F ~ 0 , … , F ~ p \widetilde{\mathcal{F}}_{0},\ldots,\widetilde{\mathcal{F}}_{p} F 0 , … , F p to be unital, and then K ∗ \mathcal{K}^{*} K ∗ above defines an A ∞ A_{\infty} A ∞ -natural transformation of unital A ∞ A_{\infty} A ∞ -functors Fun A ∞ u n ( A 2 , B ) → Fun A ∞ u n ( A 1 , B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{1},\mathcal{B}) Fun A ∞ un ( A 2 , B ) → Fun A ∞ un ( A 1 , B ) .
To see (4.33) defines an element of Hom u n ( F ~ 0 ∘ E 1 , F ~ p ∘ E 2 ) ⊂ Hom A ∞ ( F ~ 0 ∘ E 1 , F ~ 1 ∘ E 2 ) \operatorname{Hom}^{\mathrm{un}}(\widetilde{\mathcal{F}}_{0}\circ\mathcal{E}_{1},\widetilde{\mathcal{F}}_{p}\circ\mathcal{E}_{2})\subset\operatorname{Hom}_{A_{\infty}}(\widetilde{\mathcal{F}}_{0}\circ\mathcal{E}_{1},\widetilde{\mathcal{F}}_{1}\circ\mathcal{E}_{2}) Hom un ( F 0 ∘ E 1 , F p ∘ E 2 ) ⊂ Hom A ∞ ( F 0 ∘ E 1 , F 1 ∘ E 2 ) in this case, note that if 1 ≤ j ≤ n 1\leq j\leq n 1 ≤ j ≤ n and x j − 1 = x j x_{j-1}=x_{j} x j − 1 = x j , and we evaluate (4.33) on an element of the form a 1 ⊗ ⋯ ⊗ a j − 1 ⊗ 1 x j ⊗ a j + 1 ⊗ ⋯ ⊗ a n a_{1}\otimes\cdots\otimes a_{j-1}\otimes1_{x_{j}}\otimes a_{j+1}\otimes\cdots\otimes a_{n} a 1 ⊗ ⋯ ⊗ a j − 1 ⊗ 1 x j ⊗ a j + 1 ⊗ ⋯ ⊗ a n for a i ∈ Hom A ( x i − 1 , x i ) a_{i}\in\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_{i}) a i ∈ Hom A ( x i − 1 , x i ) then in each term in the sum in (4.33), either 1 x j 1_{x_{j}} 1 x j is an input in K n l x m l − 1 , … , x m l \mathcal{K}_{n_{l}}^{x_{m_{l-1}},\ldots,x_{m_{l}}} K n l x m l − 1 , … , x m l , so it gives zero as K \mathcal{K} K is unital, or 1 x j 1_{x_{j}} 1 x j is an input in ( E a ) n l x m l − 1 , … , x m l (\mathcal{E}_{a})_{n_{l}}^{x_{m_{l-1}},\ldots,x_{m_{l}}} ( E a ) n l x m l − 1 , … , x m l for a = 1 , 2 a=1,2 a = 1 , 2 and n l ≥ 2 n_{l}\geq2 n l ≥ 2 , so it gives zero as E a \mathcal{E}_{a} E a is unital, or 1 x j 1_{x_{j}} 1 x j is an input in ( E a ) 1 x j , x j (\mathcal{E}_{a})_{1}^{x_{j},x_{j}} ( E a ) 1 x j , x j for a = 1 , 2 a=1,2 a = 1 , 2 , and then ( E a ) 1 x j , x j ( 1 x j ) = 1 E a ( x j ) (\mathcal{E}_{a})_{1}^{x_{j},x_{j}}(1_{x_{j}})=1_{\mathcal{E}_{a}(x_{j})} ( E a ) 1 x j , x j ( 1 x j ) = 1 E a ( x j ) is an input in T k E 1 ( x m 0 ) , … , E 2 ( x m k ) \mathcal{T}_{k}^{\mathcal{E}_{1}(x_{m_{0}}),\ldots,\mathcal{E}_{2}(x_{m_{k}})} T k E 1 ( x m 0 ) , … , E 2 ( x m k ) , so it gives zero as T \mathcal{T} T lies in Hom A ∞ u n ( F ~ 0 , F ~ 1 ) \operatorname{Hom}^{\mathrm{un}}_{A_{\infty}}(\widetilde{\mathcal{F}}_{0},\widetilde{\mathcal{F}}_{1}) Hom A ∞ un ( F 0 , F 1 ) .
In the cohomologically unital, or unital, cases, we can also show that if K \mathcal{K} K is a (cohomologically) unital A ∞ A_{\infty} A ∞ -natural isomorphism then K ∗ \mathcal{K}^{*} K ∗ is also a (cohomologically) unital A ∞ A_{\infty} A ∞ -natural isomorphism. This is because then there is another A ∞ A_\infty A ∞ -natural transformation J ~ : E 2 ⇒ E 1 \widetilde{\mathcal{J}}:\mathcal{E}_2\Rightarrow\mathcal{E}_1 J : E 2 ⇒ E 1 such that [ J ~ ] [\widetilde{\mathcal{J}}] [ J ] is inverse to [ K ] [\mathcal{K}] [ K ] in H 0 ( Fun ∗ ( A 1 , A 2 ) ) H^0(\operatorname{Fun}_{*}(\mathcal{A}_1,\mathcal{A}_2)) H 0 ( Fun ∗ ( A 1 , A 2 )) , and we show that J ~ ∗ : E 2 ∗ ⇒ E 1 ∗ \widetilde{\mathcal{J}}^{*}:\mathcal{E}_2^{*}\Rightarrow\mathcal{E}_1^{*} J ∗ : E 2 ∗ ⇒ E 1 ∗ has [ J ~ ∗ ] [\widetilde{\mathcal{J}}^{*}] [ J ∗ ] inverse to [ K ∗ ] [\mathcal{K}^{*}] [ K ∗ ] in H 0 ( Fun ∗ ( Fun A ∞ ( A 2 , B ) , Fun A ∞ ( A 1 , B ) ) ) H^0(\operatorname{Fun}_{*}(\operatorname{Fun}_{A_\infty}(\mathcal{A}_2,\mathcal{B}),\operatorname{Fun}_{A_\infty}(\mathcal{A}_1,\mathcal{B}))) H 0 ( Fun ∗ ( Fun A ∞ ( A 2 , B ) , Fun A ∞ ( A 1 , B ))) .
(b) Let A , B 1 , B 2 \mathcal{A},\mathcal{B}_1,\mathcal{B}_2 A , B 1 , B 2 be A ∞ A_\infty A ∞ -categories over R R R , and G 1 , G 2 : B 1 → B 2 \mathcal{G}_1,\mathcal{G}_2:\mathcal{B}_1\to\mathcal{B}_2 G 1 , G 2 : B 1 → B 2 be A ∞ A_\infty A ∞ -functors, and L : G 1 ⇒ G 2 \mathcal{L}:\mathcal{G}_1\Rightarrow\mathcal{G}_2 L : G 1 ⇒ G 2 be an A ∞ A_\infty A ∞ -natural transformation. Definition 4.13 constructed A ∞ A_\infty A ∞ -functors ( G 1 ) ∗ , ( G 2 ) ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) (\mathcal{G}_1)_*,(\mathcal{G}_2)_*:\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}_1)\to\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}_2) ( G 1 ) ∗ , ( G 2 ) ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) . We will define an A ∞ A_\infty A ∞ -natural transformation L ∗ : ( G 1 ) ∗ ⇒ ( G 2 ) ∗ \mathcal{L}_*:(\mathcal{G}_1)_*\Rightarrow(\mathcal{G}_2)_* L ∗ : ( G 1 ) ∗ ⇒ ( G 2 ) ∗ . To do this, for all p ≥ 1 p\geq1 p ≥ 1 and A ∞ A_\infty A ∞ -functors F 0 , F 1 , … , F p : A → B 1 \mathcal{F}_0,\mathcal{F}_1,\ldots,\mathcal{F}_p:\mathcal{A}\to\mathcal{B}_1 F 0 , F 1 , … , F p : A → B 1 , in a very similar way to (4.21) define a graded R R R -linear map of degree − p -p − p
( L ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( G 1 ∘ F 0 , G 2 ∘ F p ) (\mathcal{L}_*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p:\operatorname{Hom}_{A_\infty}(\mathcal{F}_0,\mathcal{F}_1)\otimes\cdots\otimes\operatorname{Hom}_{A_\infty}(\mathcal{F}_{p-1},\mathcal{F}_p)\longrightarrow\operatorname{Hom}_{A_\infty}(\mathcal{G}_1\circ\mathcal{F}_0,\mathcal{G}_2\circ\mathcal{F}_p) ( L ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( G 1 ∘ F 0 , G 2 ∘ F p )
by
( ( L ∗ ) p F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) ) n x 0 , … , x n = ∑ 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n 1 ≤ a 1 < a 2 < ⋯ < a p ≤ k l j > l j − 1 , j ≠ 0 , a i ( − 1 ) ∑ i = 1 p ( k + 1 ) ( deg T i + 1 ) ( − 1 ) ∑ i = 1 p ∑ j = 1 a i − 1 ( l j − l j − 1 + 1 ) ( deg T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 p ( l i − l i − 1 + 1 ) ( k − i ) G k − l k x l k , … , x l n ( F l 1 − l 0 x l 0 , … , x l 1 ⊗ ⋯ ⊗ F l a 1 − l a 1 − 1 x l a 1 − 1 , x l a 1 ⊗ T 1 ; x l a 1 − 1 , … , x l a 1 ⊗ F l a 1 + 1 − l a 1 1 ; x l a 1 , … , x l a 1 + 1 ⊗ ⋯ ⊗ F l a 2 − l a 2 − 1 1 ; x l a 2 − 1 , … , x l a 2 ⊗ T 2 ; x l a 2 − 1 , … , x l a 2 ⊗ ⋯ ⊗ T p ; x l a p − 1 , … , x l a p ⊗ ⋯ ⊗ F l k − l k − 1 p ; x l k − 1 , … , x l k ) . \begin{aligned}
&((\mathcal{L}_*)^{\mathcal{F}_0,\ldots,\mathcal{F}_p}_p(T_1\otimes\cdots\otimes T_p))^{x_0,\ldots,x_n}_n\\
&=\sum_{\substack{0=l_0\leq l_1\leq\cdots\leq l_k=n\\1\leq a_1<a_2<\cdots<a_p\leq k\\l_j>l_{j-1},\,j\neq0,a_i}}
(-1)^{\sum_{i=1}^{p}(k+1)(\deg T_i+1)}
(-1)^{\sum_{i=1}^{p}\sum_{j=1}^{a_i-1}(l_j-l_{j-1}+1)(\deg T_i+1)}\\
&\qquad\cdot(-1)^{\sum_{i=1}^{p}(l_i-l_{i-1}+1)(k-i)}
\mathcal{G}^{x_{l_k},\ldots,x_{l_n}}_{k-l_k}
\bigl(\mathcal{F}^{x_{l_0},\ldots,x_{l_1}}_{l_1-l_0}\otimes\cdots\otimes
\mathcal{F}^{x_{l_{a_1-1}},x_{l_{a_1}}}_{l_{a_1}-l_{a_1-1}}\\
&\qquad\otimes T_{1;x_{l_{a_1-1}},\ldots,x_{l_{a_1}}}
\otimes\mathcal{F}^{1;x_{l_{a_1}},\ldots,x_{l_{a_1+1}}}_{l_{a_1+1}-l_{a_1}}
\otimes\cdots\otimes
\mathcal{F}^{1;x_{l_{a_2-1}},\ldots,x_{l_{a_2}}}_{l_{a_2}-l_{a_2-1}}\\
&\qquad\otimes T_{2;x_{l_{a_2-1}},\ldots,x_{l_{a_2}}}
\otimes\cdots\otimes T_{p;x_{l_{a_p-1}},\ldots,x_{l_{a_p}}}
\otimes\cdots\otimes
\mathcal{F}^{p;x_{l_{k-1}},\ldots,x_{l_k}}_{l_k-l_{k-1}}\bigr).
\end{aligned} (( L ∗ ) p F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) ) n x 0 , … , x n = 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n 1 ≤ a 1 < a 2 < ⋯ < a p ≤ k l j > l j − 1 , j = 0 , a i ∑ ( − 1 ) ∑ i = 1 p ( k + 1 ) ( d e g T i + 1 ) ( − 1 ) ∑ i = 1 p ∑ j = 1 a i − 1 ( l j − l j − 1 + 1 ) ( d e g T i + 1 ) ⋅ ( − 1 ) ∑ i = 1 p ( l i − l i − 1 + 1 ) ( k − i ) G k − l k x l k , … , x l n ( F l 1 − l 0 x l 0 , … , x l 1 ⊗ ⋯ ⊗ F l a 1 − l a 1 − 1 x l a 1 − 1 , x l a 1 ⊗ T 1 ; x l a 1 − 1 , … , x l a 1 ⊗ F l a 1 + 1 − l a 1 1 ; x l a 1 , … , x l a 1 + 1 ⊗ ⋯ ⊗ F l a 2 − l a 2 − 1 1 ; x l a 2 − 1 , … , x l a 2 ⊗ T 2 ; x l a 2 − 1 , … , x l a 2 ⊗ ⋯ ⊗ T p ; x l a p − 1 , … , x l a p ⊗ ⋯ ⊗ F l k − l k − 1 p ; x l k − 1 , … , x l k ) .
Again, we can prove using the identity n 1 B 1 , B 2 ( L ) = 0 n^{\mathcal{B}_1,\mathcal{B}_2}_1(\mathcal{L})=0 n 1 B 1 , B 2 ( L ) = 0 that n 1 B 1 ∗ , B 2 ∗ ( L ∗ ) = 0 n^{\mathcal{B}_1^*,\mathcal{B}_2^*}_1(\mathcal{L}_*)=0 n 1 B 1 ∗ , B 2 ∗ ( L ∗ ) = 0 , so L ∗ : G 1 ∗ ⇒ G 2 ∗ \mathcal{L}_*:\mathcal{G}_1^*\Rightarrow\mathcal{G}_2^* L ∗ : G 1 ∗ ⇒ G 2 ∗ is an A ∞ A_\infty A ∞ -natural transformation.
By the same reasoning as in (a), we can restrict L ∗ \mathcal{L}_* L ∗ to the cohomologically unital, and unital, cases, and if L \mathcal{L} L is a (cohomologically) unital A ∞ A_\infty A ∞ -natural isomorphism then L ∗ \mathcal{L}_* L ∗ is also a (cohomologically) unital A ∞ A_\infty A ∞ -natural isomorphism.
Corollary 4.31. (a) Let A , A 1 , A 2 , B , B 1 , B 2 \mathcal{A},\mathcal{A}_1,\mathcal{A}_2,\mathcal{B},\mathcal{B}_1,\mathcal{B}_2 A , A 1 , A 2 , B , B 1 , B 2 be unital A ∞ A_\infty A ∞ -categories over R R R , and E : A 1 → A 2 \mathcal{E}:\mathcal{A}_1\to\mathcal{A}_2 E : A 1 → A 2 , G : B 1 → B 2 \mathcal{G}:\mathcal{B}_1\to\mathcal{B}_2 G : B 1 → B 2 be unital homotopy equivalences. Then E ∗ : Fun A ∞ u n ( A 2 , B ) → Fun A ∞ u n ( A 1 , B ) \mathcal{E}^*:\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A}_2,\mathcal{B})\to\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A}_1,\mathcal{B}) E ∗ : Fun A ∞ un ( A 2 , B ) → Fun A ∞ un ( A 1 , B ) and G ∗ : Fun A ∞ u n ( A , B 1 ) → Fun A ∞ u n ( A , B 2 ) \mathcal{G}_*:\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}_1)\to\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B}_2) G ∗ : Fun A ∞ un ( A , B 1 ) → Fun A ∞ un ( A , B 2 ) from Definition 4.13 are unital homotopy equivalences.
(b) The analogue of (a) holds for the cohomologically unital case.
Proof. For the first part of (a), as E : A 1 → A 2 \mathcal{E}:\mathcal{A}_1\to\mathcal{A}_2 E : A 1 → A 2 is a unital homotopy equivalence there exists a unital A ∞ A_\infty A ∞ -functor F : A 2 → A 1 \mathcal{F}:\mathcal{A}_2\to\mathcal{A}_1 F : A 2 → A 1 and unital A ∞ A_\infty A ∞ -natural isomorphisms J : F ∘ E ⇒ id A 1 \mathcal{J}:\mathcal{F}\circ\mathcal{E}\Rightarrow\operatorname{id}_{\mathcal{A}_1} J : F ∘ E ⇒ id A 1 and K : E ∘ F ⇒ id A 2 \mathcal{K}:\mathcal{E}\circ\mathcal{F}\Rightarrow\operatorname{id}_{\mathcal{A}_2} K : E ∘ F ⇒ id A 2 . Then we have a unital A ∞ A_\infty A ∞ -functor F ∗ : Fun A ∞ ( A 1 , B ) → Fun A ∞ ( A 2 , B ) \mathcal{F}^*:\operatorname{Fun}_{A_\infty}(\mathcal{A}_1,\mathcal{B})\to\operatorname{Fun}_{A_\infty}(\mathcal{A}_2,\mathcal{B}) F ∗ : Fun A ∞ ( A 1 , B ) → Fun A ∞ ( A 2 , B ) and unital A ∞ A_\infty A ∞ -natural isomorphisms J ∗ : E ∗ ∘ F ∗ = ( F ∘ E ) ∗ ⇒ id Fun A ∞ ( A 1 , B ) \mathcal{J}^*:\mathcal{E}^*\circ\mathcal{F}^*=(\mathcal{F}\circ\mathcal{E})^*\Rightarrow\operatorname{id}_{\operatorname{Fun}_{A_\infty}(\mathcal{A}_1,\mathcal{B})} J ∗ : E ∗ ∘ F ∗ = ( F ∘ E ) ∗ ⇒ id Fun A ∞ ( A 1 , B ) and K ∗ : F ∗ ∘ E ∗ = ( E ∘ F ) ∗ ⇒ id Fun A ∞ ( A 2 , B ) \mathcal{K}^*:\mathcal{F}^*\circ\mathcal{E}^*=(\mathcal{E}\circ\mathcal{F})^*\Rightarrow\operatorname{id}_{\operatorname{Fun}_{A_\infty}(\mathcal{A}_2,\mathcal{B})} K ∗ : F ∗ ∘ E ∗ = ( E ∘ F ) ∗ ⇒ id Fun A ∞ ( A 2 , B ) . Hence E ∗ \mathcal{E}^* E ∗ is a unital homotopy equivalence. The rest of the proof is essentially the same.
A∞-categories and dg-categories
We generalize the relation between A ∞ A_{\infty} A ∞ -algebras and dg-algebras in § 3.5 .
Definition 4.32. Let R R R be a commutative ring. A differential graded category , or dg-category , D \mathcal{D} D over R R R consists of:
(a) A set Obj D \operatorname{Obj}\mathcal{D} Obj D of objects of D \mathcal{D} D . (More generally Obj D \operatorname{Obj}\mathcal{D} Obj D could be a proper class, but as in Remark 1.23 we restrict to U \mathcal{U} U -small sets in this book.)
(b) For all objects x , y ∈ Obj D x,y\in\operatorname{Obj}\mathcal{D} x , y ∈ Obj D , a Z \mathbb{Z} Z -graded R R R -module of morphisms
Hom D ( x , y ) = Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) , \operatorname{Hom}_{\mathcal{D}}(x,y)=\operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y), Hom D ( x , y ) = Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) , and an R R R -linear differential d : Hom d ( x , y ) ⟶ Hom d + 1 ( x , y ) d:\operatorname{Hom}^{d}(x,y)\longrightarrow\operatorname{Hom}^{d+1}(x,y) d : Hom d ( x , y ) ⟶ Hom d + 1 ( x , y ) with d 2 = 0 d^{2}=0 d 2 = 0 .
(c) For all x ∈ Obj D x\in\operatorname{Obj}\mathcal{D} x ∈ Obj D a unit 1 x 1_x 1 x (or 1 x D 1_x^{\mathcal{D}} 1 x D ) in Hom 0 ( x , x ) \operatorname{Hom}^{0}(x,x) Hom 0 ( x , x ) with d 1 x = 0 d1_x=0 d 1 x = 0 .
(d) For all x , y , z ∈ Obj D x,y,z\in\operatorname{Obj}\mathcal{D} x , y , z ∈ Obj D , an R R R -linear composition map graded of degree zero
μ x , y , z : Hom ( x , y ) ⊗ R Hom ( y , z ) ⟶ Hom ( x , z ) . \mu^{x,y,z}:\operatorname{Hom}(x,y)\otimes_R\operatorname{Hom}(y,z)\longrightarrow\operatorname{Hom}(x,z). μ x , y , z : Hom ( x , y ) ⊗ R Hom ( y , z ) ⟶ Hom ( x , z ) . This should satisfy for all w , x , y , z ∈ Obj D w,x,y,z\in\operatorname{Obj}\mathcal{D} w , x , y , z ∈ Obj D and a ∈ Hom k ( w , x ) a\in\operatorname{Hom}^{k}(w,x) a ∈ Hom k ( w , x ) , b ∈ Hom l ( x , y ) b\in\operatorname{Hom}^{l}(x,y) b ∈ Hom l ( x , y ) , and c ∈ Hom m ( y , z ) c\in\operatorname{Hom}^{m}(y,z) c ∈ Hom m ( y , z )
μ w , w , x ( 1 w ⊗ a ) = μ w , x , x ( a ⊗ 1 x ) = a , \mu^{w,w,x}(1_w\otimes a)=\mu^{w,x,x}(a\otimes1_x)=a, μ w , w , x ( 1 w ⊗ a ) = μ w , x , x ( a ⊗ 1 x ) = a , d ∘ μ w , x , y ( a ⊗ b ) = μ w , x , y ( ( d a ) ⊗ b ) + ( − 1 ) k μ w , x , y ( a ⊗ ( d b ) ) , d\circ\mu^{w,x,y}(a\otimes b)=\mu^{w,x,y}((da)\otimes b)+(-1)^{k}\mu^{w,x,y}(a\otimes(db)), d ∘ μ w , x , y ( a ⊗ b ) = μ w , x , y (( d a ) ⊗ b ) + ( − 1 ) k μ w , x , y ( a ⊗ ( d b )) , μ w , y , z ( μ w , x , y ( a ⊗ b ) ⊗ c ) = μ w , x , z ( a ⊗ μ x , y , z ( b ⊗ c ) ) . (120) \mu^{w,y,z}\left(\mu^{w,x,y}(a\otimes b)\otimes c\right)=\mu^{w,x,z}\left(a\otimes\mu^{x,y,z}(b\otimes c)\right). \tag*{(120)} μ w , y , z ( μ w , x , y ( a ⊗ b ) ⊗ c ) = μ w , x , z ( a ⊗ μ x , y , z ( b ⊗ c ) ) . (120)
Note that (unlike A ∞ A_{\infty} A ∞ -categories), D \mathcal{D} D is an honest category, with morphisms Hom ( x , y ) \operatorname{Hom}(x,y) Hom ( x , y ) (or alternatively Hom 0 ( x , y ) \operatorname{Hom}^{0}(x,y) Hom 0 ( x , y ) ), units 1 x 1_x 1 x , and composition μ x , y , z \mu^{x,y,z} μ x , y , z , which is associative by (120) .
Let D , E \mathcal{D},\mathcal{E} D , E be dg-categories over R R R . A dg-functor F : D ⟶ E \mathcal{F}:\mathcal{D}\longrightarrow\mathcal{E} F : D ⟶ E consists of a map F O b j : Obj D ⟶ Obj E \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{D}\longrightarrow\operatorname{Obj}\mathcal{E} F Obj : Obj D ⟶ Obj E , often written x ↦ F ( x ) x\mapsto\mathcal{F}(x) x ↦ F ( x ) rather than x ↦ F O b j ( x ) x\mapsto\mathcal{F}_{\mathrm{Obj}}(x) x ↦ F Obj ( x ) , and for all x , y ∈ Obj D x,y\in\operatorname{Obj}\mathcal{D} x , y ∈ Obj D an R R R -linear map graded of degree zero
F x , y : Hom D ( x , y ) ⟶ Hom E ( F ( x ) , F ( y ) ) , \mathcal{F}^{x,y}:\operatorname{Hom}_{\mathcal{D}}(x,y)\longrightarrow\operatorname{Hom}_{\mathcal{E}}(\mathcal{F}(x),\mathcal{F}(y)), F x , y : Hom D ( x , y ) ⟶ Hom E ( F ( x ) , F ( y )) ,
satisfying for all x , y , z ∈ Obj D x,y,z\in\operatorname{Obj}\mathcal{D} x , y , z ∈ Obj D
F x , y ( 1 x D ) = 1 F ( x ) E , d E ∘ F x , y = F x , y ∘ d D , \mathcal{F}^{x,y}(1_x^{\mathcal{D}})=1_{\mathcal{F}(x)}^{\mathcal{E}},\qquad d_{\mathcal{E}}\circ\mathcal{F}^{x,y}=\mathcal{F}^{x,y}\circ d_{\mathcal{D}}, F x , y ( 1 x D ) = 1 F ( x ) E , d E ∘ F x , y = F x , y ∘ d D ,
F x , z ∘ μ D x , y , z = μ E F ( x ) , F ( y ) , F ( z ) ∘ ( F x , y ⊗ F y , z ) . \mathcal{F}^{x,z}\circ\mu_{\mathcal{D}}^{x,y,z}=\mu_{\mathcal{E}}^{\mathcal{F}(x),\mathcal{F}(y),\mathcal{F}(z)}\circ(\mathcal{F}^{x,y}\otimes\mathcal{F}^{y,z}). F x , z ∘ μ D x , y , z = μ E F ( x ) , F ( y ) , F ( z ) ∘ ( F x , y ⊗ F y , z ) .
Again, dg-functors are honest functors of categories. Identity functors and compositions of functors, are defined in the obvious way. Then dg-categories over R R R and dg-functors form a category D G C a t R \mathbf{DGCat}_{R} DGCat R .
Any dg-category can be made into a unital A ∞ A_{\infty} A ∞ -category by defining m 1 x , y = d : Hom ( x , y ) ⟶ Hom ( x , y ) m_{1}^{x,y}=d:\operatorname{Hom}(x,y)\longrightarrow\operatorname{Hom}(x,y) m 1 x , y = d : Hom ( x , y ) ⟶ Hom ( x , y ) , and m 2 x , y , z = μ x , y , z m_{2}^{x,y,z}=\mu^{x,y,z} m 2 x , y , z = μ x , y , z , and m k x 0 , … , x k = 0 m_{k}^{x_{0},\ldots,x_{k}}=0 m k x 0 , … , x k = 0 for k ≥ 3 k\geq3 k ≥ 3 . This identifies dg-categories with unital A ∞ A_{\infty} A ∞ -categories with m k = 0 m_k=0 m k = 0 for k ≥ 3 k\geq3 k ≥ 3 .
Similarly, any dg-functor F : D → E \mathcal{F}:\mathcal{D}\to\mathcal{E} F : D → E can be made into a unital A ∞ A_{\infty} A ∞ -functor of A ∞ A_{\infty} A ∞ -categories by setting F 1 x , y = F x , y \mathcal{F}^{x,y}_{1}=\mathcal{F}^{x,y} F 1 x , y = F x , y and F k x 0 , … , x k = 0 \mathcal{F}^{x_{0},\ldots,x_{k}}_{k}=0 F k x 0 , … , x k = 0 for k ≥ 2 k\geq2 k ≥ 2 . This identifies dg-functors with unital A ∞ A_{\infty} A ∞ -functors F \mathcal{F} F (between A ∞ A_{\infty} A ∞ -categories coming from dg-categories) with F k = 0 \mathcal{F}_{k}=0 F k = 0 for k ≥ 2 k\geq2 k ≥ 2 . Thus we have an inclusion D G C a t R ⊂ A ∞ C a t R u n \mathbf{DGCat}_{R}\subset\mathbf{A_{\infty}Cat}^{\mathrm{un}}_{R} DGCat R ⊂ A ∞ Cat R un as a non-full subcategory.
We also write D G C a t R A ∞ \mathbf{DGCat}^{A_{\infty}}_{R} DGCat R A ∞ for the full subcategory of A ∞ C a t R u n \mathbf{A_{\infty}Cat}^{\mathrm{un}}_{R} A ∞ Cat R un whose objects are dg-categories, but whose morphisms are A ∞ A_{\infty} A ∞ -functors rather than dg-functors. Then D G C a t R ⊂ D G C a t R A ∞ ⊂ A ∞ C a t R u n \mathbf{DGCat}_{R}\subset\mathbf{DGCat}^{A_{\infty}}_{R}\subset\mathbf{A_{\infty}Cat}^{\mathrm{un}}_{R} DGCat R ⊂ DGCat R A ∞ ⊂ A ∞ Cat R un .
We call a dg-functor F : D → E \mathcal{F}:\mathcal{D}\to\mathcal{E} F : D → E a quasi-equivalence if it is a quasi-equivalence of A ∞ A_{\infty} A ∞ -categories, as in Definition 4.28.
Remark 4.33. Dg-categories have many applications (for example, to derived categories), and have received more attention than A ∞ A_{\infty} A ∞ -categories. Note that one can consider A ∞ A_{\infty} A ∞ -functors between dg-categories, and for some purposes these are better than dg-functors. For example, a claim of Kontsevich proved in [12, Th. C] says that if A , B \mathcal{A},\mathcal{B} A , B are h-projective dg-categories over R R R with split units, then the internal Hom object Hom ( A , B ) \operatorname{Hom}(\mathcal{A},\mathcal{B}) Hom ( A , B ) in D G C a t R \mathbf{DGCat}_{R} DGCat R is Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) . See [4, §1.8] for discussion of why A ∞ A_{\infty} A ∞ -categories and A ∞ A_{\infty} A ∞ -functors are important in the study of dg-categories.
Remark 4.34. If A \mathcal{A} A is a unital A ∞ A_{\infty} A ∞ -category over R R R , and B \mathcal{B} B is a dg-category over R R R , then Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) in Definition 4.11 is a dg-category over R R R . This follows from (4.16)–(4.17), as m k B = 0 m^{\mathcal{B}}_{k}=0 m k B = 0 for k ≥ 3 k\geq3 k ≥ 3 implies that the A ∞ A_{\infty} A ∞ -operations n p \mathfrak{n}_{p} n p in Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) have n p = 0 \mathfrak{n}_{p}=0 n p = 0 for p ≥ 3 p\geq3 p ≥ 3 .
Tabuada [87] proves:
Theorem 4.35. Let R R R be a commutative ring. There is a combinatorial model category structure on the category D G C a t R \mathbf{DGCat}_{R} DGCat R of dg-categories over R R R , whose weak equivalences are quasi-equivalences. All objects in D G C a t R \mathbf{DGCat}_{R} DGCat R are fibrant.
By properties of ∞ \infty ∞ -category localizations of model categories, and using Lurie [61, Prop. 1.3.4.22] to deduce presentability, this implies:
Corollary 4.36. The ∞ \infty ∞ -category D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]_{\infty} DGCat R [ QE dg − 1 ] ∞ localizing quasi-equivalences in D G C a t R \mathbf{DGCat}_{R} DGCat R , as in §2.6, has all small homotopy limits and homotopy colimits. Also D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]_{\infty} DGCat R [ QE dg − 1 ] ∞ is presentable in the sense of Lurie [60, §5], [63, §9.5].
Remark 4.37. For more information on model categories, see Hovey [44]. In §2.6 we discussed ∞ \infty ∞ -category localizations C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]_{\infty} C [ W − 1 ] ∞ of a category with weak equivalences ( C , W ) (\mathcal{C},\mathcal{W}) ( C , W ) . If C \mathcal{C} C is a model category and the weak equivalences W \mathcal{W} W come from the model category structure, then the theory of model categories provides tools for describing C [ W − 1 ] ∞ \mathcal{C}[\mathcal{W}^{-1}]_{\infty} C [ W − 1 ] ∞ and Ho ( C [ W − 1 ] ∞ ) \operatorname{Ho}(\mathcal{C}[\mathcal{W}^{-1}]_{\infty}) Ho ( C [ W − 1 ] ∞ ) . So Theorem 4.35 helps us understand D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]_{\infty} DGCat R [ QE dg − 1 ] ∞ . We return to this in §4.10 and §9.2.
In contrast to Theorem 4.35, Canónaco–Ornaghi–Stellari [11, §1.5] show that A ∞ C a t R u n \mathbf{A_{\infty}Cat}^{\mathrm{un}}_{R} A ∞ Cat R un has no model structure, as not all equalizers exist in A ∞ C a t R u n \mathbf{A_{\infty}Cat}^{\mathrm{un}}_{R} A ∞ Cat R un .
The next theorem is primarily due to Toën [90], although [90] predates the theory of ∞ \infty ∞ -categories, and we have restated it in ∞ \infty ∞ -categorical language. See also Canonaco–Stellari [13], who give a new proof of the results of [90], and Tanaka [88, §3], who gives an alternative ∞ \infty ∞ -categorical proof of the existence of ⊗ ∞ \otimes_\infty ⊗ ∞ , Hom \operatorname{Hom} Hom using results of Haugseng and Gepner–Haugseng.
Theorem 4.38. (a) There is a symmetric monoidal structure ⊗ ∞ \otimes^\infty ⊗ ∞ on the ∞ \infty ∞ -category D G C a t R [ Q d g − 1 ] ∞ \mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty DGCat R [ Q dg − 1 ] ∞ . On objects C \mathcal{C} C , D \mathcal{D} D it is given by
C ⊗ ∞ D ≃ C ⊗ L D ≃ C ′ ⊗ D ′ , \mathcal{C}\otimes^\infty\mathcal{D}\simeq\mathcal{C}\otimes^{\mathbb{L}}\mathcal{D}\simeq\mathcal{C}'\otimes\mathcal{D}', C ⊗ ∞ D ≃ C ⊗ L D ≃ C ′ ⊗ D ′ , where ⊗ : D G C a t R × D G C a t R → D G C a t R \otimes:\mathrm{DGCat}_R\times\mathrm{DGCat}_R\to\mathrm{DGCat}_R ⊗ : DGCat R × DGCat R → DGCat R is the classical tensor product of dg-categories, and ⊗ L \otimes^{\mathbb{L}} ⊗ L is its left derived tensor product, and C ′ \mathcal{C}' C ′ , D ′ \mathcal{D}' D ′ are cofibrant replacements for C \mathcal{C} C , D \mathcal{D} D in the Tabuada model structure.
(b) The monoidal structure ⊗ ∞ \otimes^\infty ⊗ ∞ is closed. That is, there exist internal mapping objects H o m ( C , D ) \mathcal{H}\!om(\mathcal{C},\mathcal{D}) H o m ( C , D ) in D G C a t R [ Q d g − 1 ] ∞ \mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty DGCat R [ Q dg − 1 ] ∞ such that for any other object B \mathcal{B} B , we have homotopy equivalences of mapping spaces
Hom D G C a t R [ Q d g − 1 ] ∞ ( B , H o m ( C , D ) ) \operatorname{Hom}_{\mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty}\bigl(\mathcal{B},\mathcal{H}\!om(\mathcal{C},\mathcal{D})\bigr) Hom DGCat R [ Q dg − 1 ] ∞ ( B , H o m ( C , D ) ) ≃ Hom D G C a t R [ Q d g − 1 ] ∞ ( B ⊗ ∞ C , D ) . \simeq\operatorname{Hom}_{\mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty}\bigl(\mathcal{B}\otimes^\infty\mathcal{C},\mathcal{D}\bigr). ≃ Hom DGCat R [ Q dg − 1 ] ∞ ( B ⊗ ∞ C , D ) . (c) The mapping object H o m ( C , D ) \mathcal{H}\!om(\mathcal{C},\mathcal{D}) H o m ( C , D ) may be described explicitly as a dg-category as the full subcategory of ‘right quasi-representable objects’ in the dg-category of C ⊗ ∞ D o p \mathcal{C}\otimes^\infty\mathcal{D}^{\mathrm{op}} C ⊗ ∞ D op -modules (dg-functors from C ⊗ ∞ D o p \mathcal{C}\otimes^\infty\mathcal{D}^{\mathrm{op}} C ⊗ ∞ D op to complexes of R R R -modules).
(d) For all objects C \mathcal{C} C , D \mathcal{D} D in D G C a t R [ Q d g − 1 ] ∞ \mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty DGCat R [ Q dg − 1 ] ∞ , we have a homotopy equivalence
Hom D G C a t R [ Q d g − 1 ] ∞ ( C , D ) ≃ N d g ( H o m ( C , D ) ) K a n , \operatorname{Hom}_{\mathrm{DGCat}_R[\mathcal{Q}^{-1}_{\mathrm{dg}}]_\infty}(\mathcal{C},\mathcal{D})\simeq\mathcal{N}_{\mathrm{dg}}\bigl(\mathcal{H}\!om(\mathcal{C},\mathcal{D})\bigr)^{\mathrm{Kan}}, Hom DGCat R [ Q dg − 1 ] ∞ ( C , D ) ≃ N dg ( H o m ( C , D ) ) Kan , where N d g ( − ) \mathcal{N}_{\mathrm{dg}}(-) N dg ( − ) is the dg-nerve N d g : D G C a t R → C a t ∞ ∞ \mathcal{N}_{\mathrm{dg}}:\mathrm{DGCat}_R\to\mathrm{Cat}^{\infty}_\infty N dg : DGCat R → Cat ∞ ∞ defined by Lurie [61, §1.3.1], an ∞ \infty ∞ -category, and ( ⋯ ) K a n (\cdots)^{\mathrm{Kan}} ( ⋯ ) Kan is the maximal Kan complex inside it.
Unital A ∞ A_\infty A ∞ -categories with split units
To have split units is a condition we can impose on a unital A ∞ A_\infty A ∞ -category. It becomes important when we want to compare unital A ∞ A_\infty A ∞ -categories with dg-categories, or with cohomologically unital A ∞ A_\infty A ∞ -categories. It is also part of a notion of ‘cofibrant’ unital A ∞ A_\infty A ∞ -category, as explained in §4.7. The next definition follows Tanaka [88, Def. 3.33], Ornaghi [73, Def. 1.13] and Canonaco–Ornaghi–Stellari [12, Def. 3.4].
Definition 4.39. We say that a unital A ∞ A_\infty A ∞ -category A \mathcal{A} A over a commutative ring R R R has split units if for all objects x x x in A \mathcal{A} A the map R → Hom 0 ( x , x ) R\to\operatorname{Hom}^0(x,x) R → Hom 0 ( x , x ) , λ ↦ λ 1 x \lambda\mapsto\lambda1_x λ ↦ λ 1 x , is injective and splits as a map of R R R -modules (not necessarily as a map of complexes).
To have split units is a homotopy-invariant property:
Proposition 4.40. Let F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a unital homotopy equivalence of unital A ∞ A_{\infty} A ∞ -categories over R R R . Then A \mathcal{A} A has split units if and only if B \mathcal{B} B has split units.
Proof. Suppose B \mathcal{B} B has split units, and let x ∈ Obj A x\in\operatorname{Obj}_{\mathcal{A}} x ∈ Obj A . Consider the R R R -linear morphism
F 1 x , x : Hom A 0 ( x , x ) ⟶ Hom B 0 ( F ( x ) , F ( x ) ) = R ⋅ 1 F ( x ) B ⊕ Hom B 0 ( F ( x ) , F ( x ) ) ′ . \mathcal{F}^{x,x}_{1}:\operatorname{Hom}^{0}_{\mathcal{A}}(x,x)\longrightarrow\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x),\mathcal{F}(x))=R\cdot1^{\mathcal{B}}_{\mathcal{F}(x)}\oplus\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x),\mathcal{F}(x))'. F 1 x , x : Hom A 0 ( x , x ) ⟶ Hom B 0 ( F ( x ) , F ( x )) = R ⋅ 1 F ( x ) B ⊕ Hom B 0 ( F ( x ) , F ( x ) ) ′ . where the right hand side is split as B \mathcal{B} B has split units. As F 1 x , x ( 1 x A ) = 1 F ( x ) B \mathcal{F}^{x,x}_{1}(1^{\mathcal{A}}_{x})=1^{\mathcal{B}}_{\mathcal{F}(x)} F 1 x , x ( 1 x A ) = 1 F ( x ) B , using F 1 x , x \mathcal{F}^{x,x}_{1} F 1 x , x we see that the map R → Hom A 0 ( x , x ) R\to\operatorname{Hom}^{0}_{\mathcal{A}}(x,x) R → Hom A 0 ( x , x ) , λ ↦ λ 1 x A \lambda\mapsto\lambda1^{\mathcal{A}}_{x} λ ↦ λ 1 x A is injective, and we have a splitting
Hom A 0 ( x , x ) = R ⋅ 1 x A ⊕ ( F 1 x , x ) − 1 ( Hom B 0 ( F ( x ) , F ( x ) ) ′ ) . \operatorname{Hom}^{0}_{\mathcal{A}}(x,x)=R\cdot1^{\mathcal{A}}_{x}\oplus(\mathcal{F}^{x,x}_{1})^{-1}\bigl(\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x),\mathcal{F}(x))'\bigr). Hom A 0 ( x , x ) = R ⋅ 1 x A ⊕ ( F 1 x , x ) − 1 ( Hom B 0 ( F ( x ) , F ( x ) ) ′ ) . Thus A \mathcal{A} A has split units. Applying the same argument to a quasi-inverse G : B → A \mathcal{G}:\mathcal{B}\to\mathcal{A} G : B → A for F \mathcal{F} F , we see that if A \mathcal{A} A has split units then so does B \mathcal{B} B . □
Remark 4.41. Some natural unital A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A contain a ‘zero object’ z ∈ Obj A z\in\operatorname{Obj}_{\mathcal{A}} z ∈ Obj A with Hom ( z , z ) = 0 \operatorname{Hom}(z,z)=0 Hom ( z , z ) = 0 , and Hom ( x , z ) = Hom ( z , x ) = 0 \operatorname{Hom}(x,z)=\operatorname{Hom}(z,x)=0 Hom ( x , z ) = Hom ( z , x ) = 0 for all x x x in Obj A \operatorname{Obj}_{\mathcal{A}} Obj A . Then 1 z = 0 1_z=0 1 z = 0 , so the map R → Hom 0 ( z , z ) R\to\operatorname{Hom}^{0}(z,z) R → Hom 0 ( z , z ) , λ ↦ λ 1 z \lambda\mapsto\lambda1_z λ ↦ λ 1 z is not injective, and A \mathcal{A} A does not have split units, and nor does any unital A ∞ A_{\infty} A ∞ -category B \mathcal{B} B homotopy equivalent to A \mathcal{A} A . This holds even if R R R is a field.
Here is Canonaco–Ornaghi–Stellari [12, Lem. 4.1].
Lemma 4.42. Let F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a cohomologically unital functor between unital A ∞ A_{\infty} A ∞ -categories A , B \mathcal{A},\mathcal{B} A , B over a commutative ring R R R . If A \mathcal{A} A has split units then there exists a unital A ∞ A_{\infty} A ∞ -functor F ′ : A → B \mathcal{F}':\mathcal{A}\to\mathcal{B} F ′ : A → B and a cohomologically unital A ∞ A_{\infty} A ∞ -natural isomorphism S : F ⇒ F ′ \mathcal{S}:\mathcal{F}\Rightarrow\mathcal{F}' S : F ⇒ F ′ .
Example 4.43. Let A , B \mathcal{A},\mathcal{B} A , B be the unital dg-categories over the commutative ring R = Z 4 R=\mathbb{Z}_4 R = Z 4 with one object ∗ \ast ∗ and, regarding Z 2 \mathbb{Z}_2 Z 2 as a module over Z 4 \mathbb{Z}_4 Z 4 ,
Hom A ( ∗ , ∗ ) = Z 2 ⟨ e A ⟩ ⊕ Z 2 ⟨ 1 ∗ A ⟩ , Hom B ( ∗ , ∗ ) = Z 4 ⟨ e B ⟩ ⊕ Z 4 ⟨ 1 ∗ B ⟩ , \operatorname{Hom}_{\mathcal{A}}(\ast,\ast)=\mathbb{Z}_2\langle e_{\mathcal{A}}\rangle\oplus\mathbb{Z}_2\langle1^{\mathcal{A}}_{\ast}\rangle,\qquad\operatorname{Hom}_{\mathcal{B}}(\ast,\ast)=\mathbb{Z}_4\langle e_{\mathcal{B}}\rangle\oplus\mathbb{Z}_4\langle1^{\mathcal{B}}_{\ast}\rangle, Hom A ( ∗ , ∗ ) = Z 2 ⟨ e A ⟩ ⊕ Z 2 ⟨ 1 ∗ A ⟩ , Hom B ( ∗ , ∗ ) = Z 4 ⟨ e B ⟩ ⊕ Z 4 ⟨ 1 ∗ B ⟩ , where e A , e B e_{\mathcal{A}},e_{\mathcal{B}} e A , e B have degree − 1 -1 − 1 and m 1 A ; ∗ , ∗ ( e A ) = 1 ∗ A m^{\mathcal{A};\ast,\ast}_{1}(e_{\mathcal{A}})=1^{\mathcal{A}}_{\ast} m 1 A ; ∗ , ∗ ( e A ) = 1 ∗ A , m 1 B ; ∗ , ∗ ( e B ) = 1 ∗ B m^{\mathcal{B};\ast,\ast}_{1}(e_{\mathcal{B}})=1^{\mathcal{B}}_{\ast} m 1 B ; ∗ , ∗ ( e B ) = 1 ∗ B . Then H ∗ ( Hom A ( ∗ , ∗ ) ) = H ∗ ( Hom B ( ∗ , ∗ ) ) = 0 H^{\ast}(\operatorname{Hom}_{\mathcal{A}}(\ast,\ast))=H^{\ast}(\operatorname{Hom}_{\mathcal{B}}(\ast,\ast))=0 H ∗ ( Hom A ( ∗ , ∗ )) = H ∗ ( Hom B ( ∗ , ∗ )) = 0 . There is a cohomologically unital A ∞ A_{\infty} A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B with F ( ∗ ) = ∗ \mathcal{F}(\ast)=\ast F ( ∗ ) = ∗ and F k ∗ , … , ∗ = 0 \mathcal{F}^{\ast,\ldots,\ast}_{k}=0 F k ∗ , … , ∗ = 0 for all k ≥ 1 k\ge1 k ≥ 1 . However, there are no unital A ∞ A_{\infty} A ∞ -functors F ′ : A → B \mathcal{F}':\mathcal{A}\to\mathcal{B} F ′ : A → B , as we cannot have F 1 ′ ∗ , ∗ ( 1 ∗ A ) = 1 ∗ B \mathcal{F}'^{\ast,\ast}_{1}(1^{\mathcal{A}}_{\ast})=1^{\mathcal{B}}_{\ast} F 1 ′ ∗ , ∗ ( 1 ∗ A ) = 1 ∗ B , since any Z 4 \mathbb{Z}_4 Z 4 -linear morphism Z 2 → Z 4 \mathbb{Z}_2\to\mathbb{Z}_4 Z 2 → Z 4 is zero. Here A \mathcal{A} A does not have split units. This shows that the condition that A \mathcal{A} A has split units in Lemma 4.42 is necessary.
As in Canonaco–Ornaghi–Stellari [11, Rem. 1.24 & Cor. 2.6], [12, Rem. 5.4], the next result follows from Lefèvre-Hasegawa [57, Lem. 8.2.1.3] and Lemma 4.42. Lefèvre-Hasegawa takes R R R to be a field, but his argument works for general R R R provided A \mathcal{A} A has split units. Here Func A ∞ u ( A , B ) \operatorname{Func}^{u}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Func A ∞ u ( A , B ) is unital, not merely cohomologically unital, as A , B \mathcal{A},\mathcal{B} A , B are unital.
Proposition 4.44. Let R R R be a commutative ring, and A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over R R R , where A \mathcal{A} A has split units. Then the inclusion
inc : Fun A ∞ u n ( A , B ) ⟶ Fun A ∞ c u ( A , B ) \operatorname{inc}:\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\longrightarrow\operatorname{Fun}_{A_{\infty}}^{\mathrm{cu}}(\mathcal{A},\mathcal{B}) inc : Fun A ∞ un ( A , B ) ⟶ Fun A ∞ cu ( A , B ) is a quasi-equivalence of cohomologically unital A ∞ A_{\infty} A ∞ -categories.
Lyubashenko [64, Th. 8.8] proves part (a) of the next theorem (see also Lefèvre-Hasegawa [57, Th. 9.2.0.4], Seidel [84, Th. 2.9], or Fukaya [29, Th. 8.6] when R R R is a field, or for free R R R -modules). We can deduce part (b) from (a) and Proposition 4.44 as follows. Given unital F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B satisfying (i),(ii), part (a) gives a cohomologically unital quasi-inverse G ′ : B → A \mathcal{G}':\mathcal{B}\to\mathcal{A} G ′ : B → A . Proposition 4.44 in Fun A ∞ ∗ ( B , A ) \operatorname{Fun}_{A_{\infty}}^{*}(\mathcal{B},\mathcal{A}) Fun A ∞ ∗ ( B , A ) shows G ′ \mathcal{G}' G ′ is homotopic to unital G : B → A \mathcal{G}:\mathcal{B}\to\mathcal{A} G : B → A . Then we have A ∞ A_{\infty} A ∞ -natural isomorphisms S : G ∘ F ⇒ id A \mathcal{S}:\mathcal{G}\circ\mathcal{F}\Rightarrow\operatorname{id}_{\mathcal{A}} S : G ∘ F ⇒ id A and T : F ∘ G ⇒ id B \mathcal{T}:\mathcal{F}\circ\mathcal{G}\Rightarrow\operatorname{id}_{\mathcal{B}} T : F ∘ G ⇒ id B , and using Proposition 4.44 in Fun A ∞ ∗ ( A , A ) \operatorname{Fun}_{A_{\infty}}^{*}(\mathcal{A},\mathcal{A}) Fun A ∞ ∗ ( A , A ) and Fun A ∞ ∗ ( B , B ) \operatorname{Fun}_{A_{\infty}}^{*}(\mathcal{B},\mathcal{B}) Fun A ∞ ∗ ( B , B ) , we can replace these by unital A ∞ A_{\infty} A ∞ -natural isomorphisms, so F \mathcal{F} F is a unital homotopy equivalence.
Theorem 4.45. (a) Let F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a cohomologically unital A ∞ A_{\infty} A ∞ -functor of cohomologically unital A ∞ A_{\infty} A ∞ -categories A , B \mathcal{A},\mathcal{B} A , B over R R R . Then F \mathcal{F} F is a cohomologically unital homotopy equivalence if and only if:
(i) For all objects x , y x,y x , y in A \mathcal{A} A , the following is a chain homotopy equivalence of chain complexes
F 1 x , y : ( Hom A ( x , y ) , m 1 A ; x , y ) ⟶ ( Hom B ( F ( x ) , F ( y ) ) , m 1 B ; F ( x ) , F ( y ) ) . (121) \mathcal{F}_{1}^{x,y}:\left(\operatorname{Hom}_{\mathcal{A}}(x,y),m_{1}^{\mathcal{A};x,y}\right)\longrightarrow\left(\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x),\mathcal{F}(y)),m_{1}^{\mathcal{B};\mathcal{F}(x),\mathcal{F}(y)}\right). \tag*{(121)} F 1 x , y : ( Hom A ( x , y ) , m 1 A ; x , y ) ⟶ ( Hom B ( F ( x ) , F ( y )) , m 1 B ; F ( x ) , F ( y ) ) . (121) If R R R is a field, this holds if H ∗ ( F 1 x , y ) H^{*}(\mathcal{F}_{1}^{x,y}) H ∗ ( F 1 x , y ) is an isomorphism on cohomology.
(ii) H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) H^{0}(\mathcal{F}_{1}):H^{0}(\mathcal{A})\to H^{0}(\mathcal{B}) H 0 ( F 1 ) : H 0 ( A ) → H 0 ( B ) is an equivalence of categories.
(b) Suppose instead that A , B , F \mathcal{A},\mathcal{B},\mathcal{F} A , B , F are unital, and A , B \mathcal{A},\mathcal{B} A , B have split units. Then F \mathcal{F} F is a unital homotopy equivalence if and only if (i),(ii) hold.
The next theorem follows from Canonaco–Ornaghi–Stellari [12, Prop. 3.5]. (See [11, Prop. 2.1] and [4, Cor. 1.4] for related results.)
Theorem 4.46. Let R R R be a commutative ring. For every A ∈ A ∞ Cat R u n \mathcal{A}\in A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∈ A ∞ Cat R un there exists a unital quasi-equivalence F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B in the sense of Definition 4.28 with B ∈ DGCat R ⊂ A ∞ Cat R u n \mathcal{B}\in\operatorname{DGCat}_{R}\subset A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} B ∈ DGCat R ⊂ A ∞ Cat R un , where we may take F O b j : Obj A → Obj B \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{A}\to\operatorname{Obj}\mathcal{B} F Obj : Obj A → Obj B to be a bijection. If A \mathcal{A} A has split units, we may take F \mathcal{F} F to be a unital homotopy equivalence in the sense of Definition 4.27, and then B \mathcal{B} B also has split units by Proposition 4.40.
H-projective A ∞ A_{\infty} A ∞ -categories
As in Remark 4.37, A ∞ Cat R u n A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un and A ∞ Cat R c u A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}} A ∞ Cat R cu are not model categories. However, the authors of [11,12,24,72,73,76,88] treat A ∞ Cat R u n A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , A ∞ Cat R c u A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}} A ∞ Cat R cu as though they were model categories, in which weak equivalences are quasi-equivalences, and all objects are fibrant. This is true of the closely related category DGCat R \operatorname{DGCat}_{R} DGCat R by Theorem 4.35.
When working in model categories, to compute things like ∞ \infty ∞ -category mapping spaces, one needs ‘cofibrant replacements’. In [12, 72, 73, 88] the substitute for cofibrant objects in A ∞ C a t R c u \mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R cu are h-projective cohomologically unital A ∞ \mathrm{A}_{\infty} A ∞ -categories. Although this is not explicit in [12, 72, 73, 88], the appropriate substitute for cofibrant objects in A ∞ C a t R u n \mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un are h-projective unital A ∞ \mathrm{A}_{\infty} A ∞ -categories with split units , as in §4.6. The next definition follows Ornaghi [73, §6.1] and Tanaka [88, Rem.s 1.6, 1.7, Def. 3.6 & §3.11].
Definition 4.47. Let R R R be a commutative ring. A complex of R R R -modules ( A , d ) (\mathcal{A},\mathrm{d}) ( A , d ) is called h-projective (short for homotopically projective ) if whenever ( B , d ) (\mathcal{B},\mathrm{d}) ( B , d ) is a complex of R R R -modules with H ∗ ( B , d ) = 0 H^{*}(\mathcal{B},\mathrm{d})=0 H ∗ ( B , d ) = 0 then the morphism complex ( Hom ( A , B ) , d ) (\operatorname{Hom}(\mathcal{A},\mathcal{B}),\mathrm{d}) ( Hom ( A , B ) , d ) has H ∗ ( Hom ( A , B ) , d ) = 0 H^{*}(\operatorname{Hom}(\mathcal{A},\mathcal{B}),\mathrm{d})=0 H ∗ ( Hom ( A , B ) , d ) = 0 .
If ( A , d ) (\mathcal{A},\mathrm{d}) ( A , d ) is a complex of free R R R -modules it is automatically h-projective. If R R R is a field then all complexes ( A , d ) (\mathcal{A},\mathrm{d}) ( A , d ) are h-projective. An A ∞ \mathrm{A}_{\infty} A ∞ -category or dg-category A \mathcal{A} A over R R R is called h-projective if for all x , y ∈ Obj A x,y\in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , the complex ( Hom ( x , y ) , m 1 x , y ) (\operatorname{Hom}(x,y),\mathfrak{m}^{x,y}_{1}) ( Hom ( x , y ) , m 1 x , y ) is h-projective. Oh–Tanaka [72, Def. 1.2] call h-projective A ∞ \mathrm{A}_{\infty} A ∞ -categories cofibrant . Note that A \mathcal{A} A is automatically h-projective if R R R is a field, or if all Hom A ( x , y ) \operatorname{Hom}_{\mathcal{A}}(x,y) Hom A ( x , y ) are free R R R -modules.
Define full subcategories
D G C a t R s u , D G C a t R h p , D G C a t R h s ⊂ D G C a t R , \mathrm{DGCat}^{\mathrm{su}}_{R},\mathrm{DGCat}^{\mathrm{hp}}_{R},\mathrm{DGCat}^{\mathrm{hs}}_{R}\subset\mathrm{DGCat}_{R}, DGCat R su , DGCat R hp , DGCat R hs ⊂ DGCat R ,
D G C a t R A ∞ , s u , D G C a t R A ∞ , h p , D G C a t R A ∞ , h s ⊂ D G C a t R A ∞ , \mathrm{DGCat}^{\mathrm{A}_{\infty},\mathrm{su}}_{R},\mathrm{DGCat}^{\mathrm{A}_{\infty},\mathrm{hp}}_{R},\mathrm{DGCat}^{\mathrm{A}_{\infty},\mathrm{hs}}_{R}\subset\mathrm{DGCat}^{\mathrm{A}_{\infty}}_{R}, DGCat R A ∞ , su , DGCat R A ∞ , hp , DGCat R A ∞ , hs ⊂ DGCat R A ∞ ,
A ∞ C a t R u n , s u , A ∞ C a t R u n , h p , A ∞ C a t R u n , h s ⊂ A ∞ C a t R u n , \mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\mathrm{su}}_{R},\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\mathrm{hp}}_{R},\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\mathrm{hs}}_{R}\subset\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}, A ∞ Cat R un , su , A ∞ Cat R un , hp , A ∞ Cat R un , hs ⊂ A ∞ Cat R un ,
A ∞ C a t R c u , h p ⊂ A ∞ C a t R c u (122) \mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{cu},\mathrm{hp}}_{R}\subset\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} \tag*{(122)} A ∞ Cat R cu , hp ⊂ A ∞ Cat R cu (122)
where ‘su’ means objects have split units, ‘hp’ means objects are h-projective, and ‘hs’ means objects are h-projective with split units.
The following proposition comes from Tanaka [88, Prop 3.8].
Proposition 4.48. Let f : ( A , d ) → ( B , d ) f:(\mathcal{A},\mathrm{d})\to(\mathcal{B},\mathrm{d}) f : ( A , d ) → ( B , d ) be a morphism of h-projective complexes over a commutative ring R R R such that H k ( f ) : H k ( A , d ) → H k ( B , d ) H^{k}(f):H^{k}(\mathcal{A},\mathrm{d})\to H^{k}(\mathcal{B},\mathrm{d}) H k ( f ) : H k ( A , d ) → H k ( B , d ) is an isomorphism for all k ∈ Z k\in\mathbb{Z} k ∈ Z . Then f f f is a chain homotopy equivalence.
As in Ornaghi [73, Cor. 2.5], Theorem 4.45 and Proposition 4.48 yield:
Corollary 4.49. (a) Let R R R be a commutative ring and F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a quasi-equivalence of h-projective cohomologically unital A ∞ \mathrm{A}_{\infty} A ∞ -categories over R R R , as in Definition 4.28. Then F \mathcal{F} F is a cohomologically unital homotopy equivalence, as in Definition 4.27.
(b) Let F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B be a unital quasi-equivalence of h-projective unital A ∞ \mathrm{A}_{\infty} A ∞ -categories over R R R with split units, as in Definitions 4.27–4.28. Then F \mathcal{F} F is a unital homotopy equivalence, as in Definition 4.27.
To be h-projective is a homotopy-invariant property.
Proposition 4.51. Let R R R be a commutative ring and F : A → B \mathcal{F} : \mathcal{A} \to\mathcal{B} F : A → B be a (cohomologically) unital homotopy equivalence of (cohomologically) unital A ∞ A_{\infty} A ∞ -categories over R R R . Then A \mathcal{A} A is h-projective if and only if B \mathcal{B} B is h-projective.
Proof. Note that if ϕ : ( C , d ) → ( D , d ) \phi: (C,\mathrm{d}) \to(D,\mathrm{d}) ϕ : ( C , d ) → ( D , d ) is a chain homotopy equivalence of complexes of R R R -modules then ( C , d ) (C,\mathrm{d}) ( C , d ) is h-projective if and only if ( D , d ) (D,\mathrm{d}) ( D , d ) is h-projective. Suppose B \mathcal{B} B is h-projective. By Theorem 4.45 (a), equation (121) is a chain homotopy equivalence. As B \mathcal{B} B is h-projective, the right hand side of (121) is h-projective, so the left hand side is h-projective for all x , y ∈ Obj A x,y \in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , and thus A \mathcal{A} A is h-projective. Applying the same argument to a quasi-inverse G : B → A \mathcal{G} : \mathcal{B} \to\mathcal{A} G : B → A for F \mathcal{F} F , we see that if A \mathcal{A} A is h-projective then so is B \mathcal{B} B .
Ornaghi [73, Th.s B′, 5.1 & 5.2] (see also Tanaka [88, Rem. 1.8]) proves:
Theorem 4.52. Let A \mathcal{A} A be a unital A ∞ A_{\infty} A ∞ -category over a commutative ring R R R . Then there exists an h-projective unital A ∞ A_{\infty} A ∞ -category A ′ \mathcal{A}' A ′ over R R R and a unital quasi-equivalence Φ : A ′ → A \Phi: \mathcal{A}' \to\mathcal{A} Φ : A ′ → A . We call Φ \Phi Φ an h-projective resolution of A \mathcal{A} A .
We may take A ′ \mathcal{A}' A ′ to have the same objects as A \mathcal{A} A , and to have split units. If A \mathcal{A} A is a dg-category, the above holds with A ′ \mathcal{A}' A ′ a dg-category and Φ \Phi Φ a dg-functor.
The simplicial nerve construction
Example 2.11 defined the nerve functor N : Cat 1 → Cat ∞ 1 \mathcal{N} : \operatorname{Cat}^{1} \to\operatorname{Cat}^{1}_{\infty} N : Cat 1 → Cat ∞ 1 from categories to ∞ \infty ∞ -categories. Faonte [23, Def. 2.8] defines a functor N A ∞ : A ∞ Cat R u n → Cat ∞ 1 \mathcal{N}_{A_{\infty}} : A_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} \to\operatorname{Cat}^{1}_{\infty} N A ∞ : A ∞ Cat R un → Cat ∞ 1 , called the simplicial nerve , from the category of unital A ∞ A_{\infty} A ∞ -categories over R R R to the ordinary category of ∞ \infty ∞ -categories Cat ∞ 1 \operatorname{Cat}^{1}_{\infty} Cat ∞ 1 from Definition 2.9. (Note that this is different to the ∞ \infty ∞ -category of ∞ \infty ∞ -categories Cat ∞ \operatorname{Cat}_{\infty} Cat ∞ in §2.4.)
Faonte works with R R R a field, but his construction works just as well with R R R a commutative ring. As in [23, Cor. 2.10], the restriction N A ∞ ∣ DGCat R \mathcal{N}_{A_{\infty}}|_{\operatorname{DGCat}_{R}} N A ∞ ∣ DGCat R of N A ∞ \mathcal{N}_{A_{\infty}} N A ∞ to the subcategory DGCat R ⊂ A ∞ Cat R u n \operatorname{DGCat}_{R} \subset A_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} DGCat R ⊂ A ∞ Cat R un of dg-categories over R R R from Definition 4.32 is the dg-nerve N d g : DGCat R → Cat ∞ 1 \mathcal{N}_{\mathrm{dg}} : \operatorname{DGCat}_{R} \to\operatorname{Cat}^{1}_{\infty} N dg : DGCat R → Cat ∞ 1 defined by Lurie [61, §1.3.1]. Le Grignou [58] studies the simplicial nerve using operads.
Definition 4.53. Let R R R be a commutative ring. For each N ≥ 0 N \ge0 N ≥ 0 , define a unital A ∞ A_{\infty} A ∞ -category C [ N ] Δ \mathcal{C}^{\Delta}_{[N]} C [ N ] Δ over R R R to have objects Obj C [ N ] Δ = { 0 , 1 , … , N } \operatorname{Obj}\mathcal{C}^{\Delta}_{[N]} = \{0,1,\ldots,N\} Obj C [ N ] Δ = { 0 , 1 , … , N } and Hom spaces
Hom ( i , j ) = { R ⋅ ω i j , i ≤ j , 0 , i > j , \operatorname{Hom}(i,j)=
\begin{cases}
R\cdot\omega_{ij}, & i\le j,\\
0, & i>j,
\end{cases} Hom ( i , j ) = { R ⋅ ω ij , 0 , i ≤ j , i > j , with deg ω i j = 0 \deg\omega_{ij}=0 deg ω ij = 0 , and units 1 i = ω i i 1_i=\omega_{ii} 1 i = ω ii , and maps m k i 0 , … , i k m^{i_0,\ldots,i_k}_k m k i 0 , … , i k given by
m 2 i 0 , i 1 , i 2 ( ω i 0 i 1 ⊗ ω i 1 i 2 ) = ω i 0 i 2 , 0 ≤ i 0 ≤ i 1 ≤ i 2 ≤ N , m k i 0 , … , i k = 0 , k ≠ 2. (123) m^{i_0,i_1,i_2}_2(\omega_{i_0i_1}\otimes\omega_{i_1i_2})=\omega_{i_0i_2},\quad0\le i_0\le i_1\le i_2\le N,\qquad m^{i_0,\ldots,i_k}_k=0,\quad k\ne2. \tag*{(123)} m 2 i 0 , i 1 , i 2 ( ω i 0 i 1 ⊗ ω i 1 i 2 ) = ω i 0 i 2 , 0 ≤ i 0 ≤ i 1 ≤ i 2 ≤ N , m k i 0 , … , i k = 0 , k = 2. (123) If α : [ N ] → [ M ] \alpha: [N]\to[M] α : [ N ] → [ M ] is a morphism in Δ \Delta Δ in Definition 2.1, so that α : { 0 , … , N } → { 0 , … , M } \alpha: \{0,\ldots,N\}\to\{0,\ldots,M\} α : { 0 , … , N } → { 0 , … , M } is an order-preserving function, we define a unital A ∞ A_{\infty} A ∞ -functor C α Δ : C [ N ] Δ → C [ M ] Δ \mathcal{C}^{\Delta}_{\alpha} : \mathcal{C}^{\Delta}_{[N]} \to\mathcal{C}^{\Delta}_{[M]} C α Δ : C [ N ] Δ → C [ M ] Δ by Obj C α Δ = α \operatorname{Obj}\mathcal{C}^{\Delta}_{\alpha}=\alpha Obj C α Δ = α , and to have maps
( C α Δ ) 1 i 0 , i 1 : ω i 0 i 1 ↦ ω α ( i 0 ) α ( i 1 ) , 0 ≤ i 0 ≤ i 1 ≤ N , ( C α Δ ) k i 0 , … , i k = 0 , k ≠ 1. (\mathcal{C}^{\Delta}_{\alpha})^{i_0,i_1}_1:\omega_{i_0i_1}\mapsto\omega_{\alpha(i_0)\alpha(i_1)},\quad0\le i_0\le i_1\le N,\qquad(\mathcal{C}^{\Delta}_{\alpha})^{i_0,\ldots,i_k}_k=0,\quad k\ne1. ( C α Δ ) 1 i 0 , i 1 : ω i 0 i 1 ↦ ω α ( i 0 ) α ( i 1 ) , 0 ≤ i 0 ≤ i 1 ≤ N , ( C α Δ ) k i 0 , … , i k = 0 , k = 1.
This defines a functor C Δ : Δ → A ∞ C a t R u n \mathcal{C}^{\Delta} : \Delta\to\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R} C Δ : Δ → A ∞ Cat R un mapping [ N ] ↦ C [ N ] Δ [N] \mapsto\mathcal{C}^{\Delta}_{[N]} [ N ] ↦ C [ N ] Δ on objects and α ↦ C α Δ \alpha\mapsto\mathcal{C}^{\Delta}_{\alpha} α ↦ C α Δ on morphisms. In fact C [ N ] Δ \mathcal{C}^{\Delta}_{[N]} C [ N ] Δ is a dg-category, and C α Δ \mathcal{C}^{\Delta}_{\alpha} C α Δ is a dg-functor, so C Δ \mathcal{C}^{\Delta} C Δ is a functor Δ → D G C a t R ⊂ A ∞ C a t R u n \Delta\to\mathbf{DGCat}_{R} \subset\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R} Δ → DGCat R ⊂ A ∞ Cat R un .
Noting that S S e t \mathbf{SSet} SSet is the category of functors S : Δ o p → S e t S : \Delta^{\mathrm{op}} \to\mathbf{Set} S : Δ op → Set , define a functor N ∞ : A ∞ C a t R u n → S S e t \mathcal{N}_{\infty} : \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R} \to\mathbf{SSet} N ∞ : A ∞ Cat R un → SSet by N ∞ ( − ) = Hom A ∞ C a t R u n ( C Δ , − ) \mathcal{N}_{\infty}(-) = \operatorname{Hom}_{\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}}(\mathcal{C}^{\Delta}, -) N ∞ ( − ) = Hom A ∞ Cat R un ( C Δ , − ) . That is, N ∞ \mathcal{N}_{\infty} N ∞ maps an object A \mathcal{A} A to the functor N ∞ ( A ) : Δ o p → S e t \mathcal{N}_{\infty}(\mathcal{A}) : \Delta^{\mathrm{op}} \to\mathbf{Set} N ∞ ( A ) : Δ op → Set , N ∞ ( A ) ( − ) = Hom A ∞ C a t R u n ( C Δ ( − ) , A ) \mathcal{N}_{\infty}(\mathcal{A})(-) = \operatorname{Hom}_{\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R} }(\mathcal{C}^{\Delta}(-),\mathcal{A}) N ∞ ( A ) ( − ) = Hom A ∞ Cat R un ( C Δ ( − ) , A ) , and N ∞ \mathcal{N}_{\infty} N ∞ maps a morphism F : A → B \mathcal{F} : \mathcal{A} \to\mathcal{B} F : A → B to the natural transformation N ∞ ( F ) : N ∞ ( A ) ⇒ N ∞ ( B ) \mathcal{N}_{\infty}(\mathcal{F}) : \mathcal{N}_{\infty}(\mathcal{A}) \Rightarrow\mathcal{N}_{\infty}(\mathcal{B}) N ∞ ( F ) : N ∞ ( A ) ⇒ N ∞ ( B ) , N ∞ ( F ) = − ∘ F : Hom A ∞ C a t R u n ( C Δ ( − ) , A ) → Hom A ∞ C a t R u n ( C Δ ( − ) , B ) \mathcal{N}_{\infty}(\mathcal{F}) = - \circ\mathcal{F} : \operatorname{Hom}_{\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}}(\mathcal{C}^{\Delta}(-),\mathcal{A}) \to\operatorname{Hom}_{\mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}}(\mathcal{C}^{\Delta}(-),\mathcal{B}) N ∞ ( F ) = − ∘ F : Hom A ∞ Cat R un ( C Δ ( − ) , A ) → Hom A ∞ Cat R un ( C Δ ( − ) , B ) .
Faonte shows [23, Prop. 2.15] that N ∞ \mathcal{N}_{\infty} N ∞ maps to C a t ∞ 1 ⊂ S S e t \mathbf{Cat}_{\infty}^{1} \subset\mathbf{SSet} Cat ∞ 1 ⊂ SSet .
The next proposition, which is easy to prove, shows that we should regard N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) as an ∞ \infty ∞ -categorical enhancement of H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) .
Proposition 4.54. Let A \mathcal{A} A be a unital A ∞ A_{\infty} A ∞ -category over R R R . Then there is a strict isomorphism of ordinary categories H 0 ( A ) ≅ Ho ( N ∞ ( A ) ) H^{0}(\mathcal{A}) \cong\operatorname{Ho}(\mathcal{N}_{\infty}(\mathcal{A})) H 0 ( A ) ≅ Ho ( N ∞ ( A )) , where H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) is the cohomology category of A \mathcal{A} A from Definition 4.7, and Ho ( N ∞ ( A ) ) \operatorname{Ho}(\mathcal{N}_{\infty}(\mathcal{A})) Ho ( N ∞ ( A )) is the homotopy category from Definition 2.12 of N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) above. This extends to a natural isomorphism H 0 ≅ Ho ∘ N ∞ H^{0} \cong\operatorname{Ho} \circ\mathcal{N}_{\infty} H 0 ≅ Ho ∘ N ∞ of functors A ∞ C a t R u n → C a t 1 \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R} \to\mathbf{Cat}^{1} A ∞ Cat R un → Cat 1 .
This implies that homotopy equivalences f : x → y f : x \to y f : x → y in A \mathcal{A} A in the sense of Definition 4.8 are the same thing as equivalences in the ∞ \infty ∞ -category N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) in the sense of Definition 2.12.
Faonte [23, Props. 2.9 & 2.11] describes the ∞ \infty ∞ -category N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) explicitly. We have rewritten it for compatibility with our sign and order conventions.
Proposition 4.55. Let A \mathcal{A} A be a unital A ∞ A_{\infty} A ∞ -category over R R R , so that N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) is an ∞ \infty ∞ -category. Then
(a) An N N N -simplex in N ∞ ( A ) \mathcal{N}_{\infty}(\mathcal{A}) N ∞ ( A ) is equivalent to data ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) \left((x_i)_{i=0}^{N},(f^{i_0,\ldots,i_k}_{k})_{k\geq1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) , where x 0 , … , x N x_0,\ldots,x_N x 0 , … , x N are objects in A \mathcal{A} A , and for all k ≥ 1 k\geq1 k ≥ 1 and 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i k ≤ N 0\leq i_0\leq i_1\leq\cdots\leq i_k\leq N 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i k ≤ N , we have f k i 0 , … , i k ∈ Hom A 1 − k ( x i 0 , x i k ) f^{i_0,\ldots,i_k}_{k}\in\operatorname{Hom}^{1-k}_{\mathcal{A}}(x_{i_0},x_{i_k}) f k i 0 , … , i k ∈ Hom A 1 − k ( x i 0 , x i k ) , satisfying
f 1 i 0 i 0 = 1 x i 0 , f k + 1 i 0 , i 1 , … , i j − 1 , i j , i j , i j + 1 , … , i k = 0 , 0 ≤ j ≤ k , k ≠ 1 , (124) f^{i_0i_0}_{1}=\mathbb{1}_{x_{i_0}},\qquad f^{i_0,i_1,\ldots,i_{j-1},i_j,i_j,i_{j+1},\ldots,i_k}_{k+1}=0,\quad0\leq j\leq k,\ k\ne1, \tag*{(124)} f 1 i 0 i 0 = 1 x i 0 , f k + 1 i 0 , i 1 , … , i j − 1 , i j , i j , i j + 1 , … , i k = 0 , 0 ≤ j ≤ k , k = 1 , (124) and for all n ≥ 1 n\geq1 n ≥ 1 and 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i n ≤ N 0\leq i_0\leq i_1\leq\cdots\leq i_n\leq N 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i n ≤ N
∑ 0 < j < n ( − 1 ) j − 1 f n − 1 i 0 , i 1 , … , i j − 1 , i j + 1 , … , i n = ∑ k ≥ 1 , 0 = l 0 < l 1 < ⋯ < l k = n ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) m k A ; x l 0 , x l 1 , … , x l k ∘ ( f l 1 − l 0 i l 0 , … , i l 1 ⊗ f l 2 − l 1 i l 1 , … , i l 2 ⊗ ⋯ ⊗ f l k − l k − 1 i l k − 1 , … , i l k ) . (125) \sum_{0<j<n}(-1)^{j-1}f^{i_0,i_1,\ldots,i_{j-1},i_{j+1},\ldots,i_n}_{n-1}
=
\sum_{\substack{k\geq1,\ 0=l_0<l_1<\cdots<l_k=n}}(-1)^{\sum_{i=1}^{k}(l_i-l_{i-1}+1)(k-i)}m^{\mathcal{A};x_{l_0},x_{l_1},\ldots,x_{l_k}}_{k}\circ\left(f^{i_{l_0},\ldots,i_{l_1}}_{l_1-l_0}\otimes f^{i_{l_1},\ldots,i_{l_2}}_{l_2-l_1}\otimes\cdots\otimes f^{i_{l_{k-1}},\ldots,i_{l_k}}_{l_k-l_{k-1}}\right). \tag*{(125)} 0 < j < n ∑ ( − 1 ) j − 1 f n − 1 i 0 , i 1 , … , i j − 1 , i j + 1 , … , i n = k ≥ 1 , 0 = l 0 < l 1 < ⋯ < l k = n ∑ ( − 1 ) ∑ i = 1 k ( l i − l i − 1 + 1 ) ( k − i ) m k A ; x l 0 , x l 1 , … , x l k ∘ ( f l 1 − l 0 i l 0 , … , i l 1 ⊗ f l 2 − l 1 i l 1 , … , i l 2 ⊗ ⋯ ⊗ f l k − l k − 1 i l k − 1 , … , i l k ) . (125) (b) If α : [ N ~ ] → [ N ] \alpha: [\widetilde{N}] \to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , so α : { 0 , … , N ~ } → { 0 , … , N } \alpha: \{0,\ldots,\widetilde{N}\}\to\{0,\ldots,N\} α : { 0 , … , N } → { 0 , … , N } is a nondecreasing function, then N ∞ ( A ) ( α ) \mathcal{N}_{\infty}(\mathcal{A})(\alpha) N ∞ ( A ) ( α ) maps ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) \left((x_i)_{i=0}^{N},(f^{i_0,\ldots,i_k}_{k})_{k\geq1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) to ( ( x ~ i ) i = 0 N ~ , ( f ~ k i 0 , … , i k ) k ≥ 1 ) \left((\widetilde{x}_i)_{i=0}^{\widetilde{N}},(\widetilde{f}^{i_0,\ldots,i_k}_{k})_{k\geq1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) , where x ~ i = x α ( i ) \widetilde{x}_i=x_{\alpha(i)} x i = x α ( i ) and f ~ k i 0 , … , i k = f k α ( i 0 ) , … , α ( i k ) \widetilde{f}^{i_0,\ldots,i_k}_{k}=f^{\alpha(i_0),\ldots,\alpha(i_k)}_{k} f k i 0 , … , i k = f k α ( i 0 ) , … , α ( i k ) for all k ≥ 1 k\geq1 k ≥ 1 and 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i k ≤ N ~ 0\leq i_0\leq i_1\leq\cdots\leq i_k\leq\widetilde{N} 0 ≤ i 0 ≤ i 1 ≤ ⋯ ≤ i k ≤ N .
Observe in (a) that if i j − 1 = i j i_{j-1}=i_j i j − 1 = i j for any j = 1 , … , k j=1,\ldots,k j = 1 , … , k then f k i 0 , … , i k f_k^{i_0,\ldots,i_k} f k i 0 , … , i k is uniquely determined by (4.38). Thus we can restrict to data f k i 0 , … , i k f_k^{i_0,\ldots,i_k} f k i 0 , … , i k for 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N 0\le i_0<i_1<\cdots<i_k\le N 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N . This gives the following way to rewrite Proposition 4.55, which will be useful in Chapter 9. In (4.40), the last two cases are deduced from (4.38).
Corollary 4.56. Let A \mathcal{A} A be a unital A ∞ A_\infty A ∞ -category over R R R , so that N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) is an ∞ \infty ∞ -category. Then
(a) An N N N -simplex in N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) is equivalent to data ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) \left((x_i)_{i=0}^N,(f_k^{i_0,\ldots,i_k})_{k\ge1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) , where x 0 , … , x N x_0,\ldots,x_N x 0 , … , x N are objects in A \mathcal{A} A , and for all k ≥ 1 k\ge1 k ≥ 1 and 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N 0\le i_0<i_1<\cdots<i_k\le N 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N , we have f k i 0 , … , i k ∈ Hom A 1 − k ( x i 0 , x i k ) f_k^{i_0,\ldots,i_k}\in\operatorname{Hom}_{\mathcal{A}}^{1-k}(x_{i_0},x_{i_k}) f k i 0 , … , i k ∈ Hom A 1 − k ( x i 0 , x i k ) , satisfying (4.39) for all 1 ≤ n ≤ N 1\le n\le N 1 ≤ n ≤ N and $0\le i_0<i_1<\cdots<i_n\le N.
(b) If α : [ N ~ ] → [ N ] \alpha:[\widetilde{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , so α : { 0 , … , N ~ } → { 0 , … , N } \alpha:\{0,\ldots,\widetilde{N}\}\to\{0,\ldots,N\} α : { 0 , … , N } → { 0 , … , N } is a nondecreasing function, then N A ∞ ( A ) ( α ) \mathcal{N}_{A_\infty}(\mathcal{A})(\alpha) N A ∞ ( A ) ( α ) maps ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) \left((x_i)_{i=0}^N,(f_k^{i_0,\ldots,i_k})_{k\ge1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) to ( ( x ~ i ) i = 0 N ~ , ( f ~ k i 0 , … , i k ) k ≥ 1 ) \left((\widetilde{x}_i)_{i=0}^{\widetilde{N}},(\widetilde{f}_k^{i_0,\ldots,i_k})_{k\ge1}\right) ( ( x i ) i = 0 N , ( f k i 0 , … , i k ) k ≥ 1 ) , where x ~ i = x α ( i ) \widetilde{x}_i=x_{\alpha(i)} x i = x α ( i ) and for all 1 ≤ k ≤ N 1\le k\le N 1 ≤ k ≤ N and 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N ~ 0\le i_0<i_1<\cdots<i_k\le\widetilde{N} 0 ≤ i 0 < i 1 < ⋯ < i k ≤ N we have
f ~ k i 0 , … , i k = { f k α ( i 0 ) , … , α ( i k ) if α ∣ { i 0 , … , i k } is injective , 1 x i if k = 1 and α ( i 0 ) = α ( i 1 ) = i , 0 otherwise. (126) \widetilde{f}_k^{i_0,\ldots,i_k}=
\begin{cases}
f_k^{\alpha(i_0),\ldots,\alpha(i_k)} & \text{if }\alpha|_{\{i_0,\ldots,i_k\}}\text{ is injective},\\
\mathbbm{1}_{x_i} & \text{if }k=1\text{ and }\alpha(i_0)=\alpha(i_1)=i,\\
0 & \text{otherwise.}
\end{cases} \tag*{(126)} f k i 0 , … , i k = ⎩ ⎨ ⎧ f k α ( i 0 ) , … , α ( i k ) 1 x i 0 if α ∣ { i 0 , … , i k } is injective , if k = 1 and α ( i 0 ) = α ( i 1 ) = i , otherwise. (126)
Remark 4.57. The simplicial nerve construction does not work nicely for nonunital A ∞ A_\infty A ∞ -categories, because we need units in (4.38) and (4.40).
Theorem 4.58. Let A \mathcal{A} A be a unital A ∞ A_\infty A ∞ -category over a commutative ring R R R , and x x x an object in A \mathcal{A} A . Write N A ∞ ( A ) K a n \mathcal{N}_{A_\infty}(\mathcal{A})^{\mathrm{Kan}} N A ∞ ( A ) Kan for the maximal Kan complex in N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) , as in Definition 2.14, and N A ∞ ( A ) x K a n \mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}} N A ∞ ( A ) x Kan for the connected component of N A ∞ ( A ) K a n \mathcal{N}_{A_\infty}(\mathcal{A})^{\mathrm{Kan}} N A ∞ ( A ) Kan containing x x x . Then the homotopy groups of N A ∞ ( A ) x K a n \mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}} N A ∞ ( A ) x Kan are given by
π 1 ( N A ∞ ( A ) x K a n ) ≅ { [ f ] ∣ [ f ] : x → x is an isomorphism in H 0 ( A ) } , (127) \pi_1\left(\mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}}\right)\cong\{[f]\mid[f]:x\to x\text{ is an isomorphism in }H^0(\mathcal{A})\}, \tag*{(127)} π 1 ( N A ∞ ( A ) x Kan ) ≅ {[ f ] ∣ [ f ] : x → x is an isomorphism in H 0 ( A )} , (127)
with group operation composition in H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) , and for n ≥ 2 n\ge2 n ≥ 2
π n ( N A ∞ ( A ) x K a n ) ≅ H 1 − n ( Hom A ( x , x ) , m 1 x , x ) , (128) \pi_n\left(\mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}}\right)\cong H^{1-n}\left(\operatorname{Hom}_{\mathcal{A}}(x,x),m_1^{x,x}\right), \tag*{(128)} π n ( N A ∞ ( A ) x Kan ) ≅ H 1 − n ( Hom A ( x , x ) , m 1 x , x ) , (128)
with group operation addition in cohomology.
Proof. As N A ∞ ( A ) x K a n \mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}} N A ∞ ( A ) x Kan is a Kan complex, elements of π 1 ( N A ∞ ( A ) x K a n ) \pi_1\left(\mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}}\right) π 1 ( N A ∞ ( A ) x Kan ) are homotopy classes of morphisms f : x → x f:x\to x f : x → x in N A ∞ ( A ) x K a n \mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}} N A ∞ ( A ) x Kan . Such f f f are morphisms f : x → x f:x\to x f : x → x in N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) which are equivalences, and hence morphisms f : x → x f:x\to x f : x → x in A \mathcal{A} A for which [ f ] : x → x [f]:x\to x [ f ] : x → x is an isomorphism in H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) .
Two such morphisms f , f ′ : x → x f,f':x\to x f , f ′ : x → x are homotopic if there exists a 2-simplex η \eta η in N A ∞ ( A ) x K a n \mathcal{N}_{A_\infty}(\mathcal{A})_x^{\mathrm{Kan}} N A ∞ ( A ) x Kan of the form
x → f ′ x id x ↓ ↓ x → f x \begin{CD}
x @>{f'}>> x\\
@V{\operatorname{id}_x}VV @VV{}V\\
x @>{f}>> x
\end{CD} x id x ↓ ⏐ x f ′ f x ↓ ⏐ x
By Corollary 4.56(a), we can show such η \eta η is equivalent to η 2 012 ∈ Hom A − 1 ( x , x ) \eta^{012}_{2} \in\operatorname{Hom}^{-1}_{\mathcal A}(x,x) η 2 012 ∈ Hom A − 1 ( x , x ) satisfying f ′ = f + m 1 x , x ∘ η 2 012 f' = f + m^{x,x}_{1} \circ\eta^{012}_{2} f ′ = f + m 1 x , x ∘ η 2 012 . Thus f , f ′ f,f' f , f ′ are homotopic if and only if [ f ] = [ f ′ ] [f] = [f'] [ f ] = [ f ′ ] in Ho ( A ) \operatorname{Ho}(\mathcal A) Ho ( A ) . This proves (4.41) as sets. To see that the group operation is composition in H 0 ( A ) H^{0}(\mathcal A) H 0 ( A ) , note that for all isomorphisms f , f ′ : x → x f,f':x \to x f , f ′ : x → x in A \mathcal A A there is a 2-simplex ζ \zeta ζ in N A ∞ ( A ) x Kan \mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_{x} N A ∞ ( A ) x Kan as follows, with ζ 2 012 = 0 \zeta^{012}_{2}=0 ζ 2 012 = 0 .
Figure 1 (PDF p. 89).
Now let n ≥ 2 n \ge2 n ≥ 2 . Then elements of π n ( N A ∞ ( A ) x Kan ) \pi_n(\mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_x) π n ( N A ∞ ( A ) x Kan ) are homotopy classes of n n n -simplices S : Δ n → N A ∞ ( A ) x Kan S:\Delta^n \to\mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_x S : Δ n → N A ∞ ( A ) x Kan such that S ∣ ∂ Δ n S|_{\partial\Delta^n} S ∣ ∂ Δ n is constant at x x x . By Corollary 4.56(a), S S S is equivalent to data ( x i ) i = 0 n , ( S k i 0 , … , i k ) k ≥ 1 (x_i)_{i=0}^{n},(S^{i_0,\ldots,i_k}_{k})_{k\ge1} ( x i ) i = 0 n , ( S k i 0 , … , i k ) k ≥ 1 in which x i = x x_i=x x i = x for all i i i , and as S ∣ ∂ Δ n ≡ x S|_{\partial\Delta^n}\equiv x S ∣ ∂ Δ n ≡ x , the only possibly nonzero S k i 0 , … , i k S^{i_0,\ldots,i_k}_{k} S k i 0 , … , i k is S n 01 ⋯ n S^{01\cdots n}_{n} S n 01 ⋯ n in Hom A 1 − n ( x , x ) \operatorname{Hom}^{1-n}_{\mathcal A}(x,x) Hom A 1 − n ( x , x ) . Then (4.39) becomes m 1 x , x ( S n 01 ⋯ n ) = 0 m^{x,x}_{1}(S^{01\cdots n}_{n})=0 m 1 x , x ( S n 01 ⋯ n ) = 0 , so we have a cohomology class [ S n 01 ⋯ n ] ∈ H 1 − n ( Hom A ( x , x ) , m 1 x , x ) [S^{01\cdots n}_{n}] \in H^{1-n}(\operatorname{Hom}_{\mathcal A}(x,x),m^{x,x}_{1}) [ S n 01 ⋯ n ] ∈ H 1 − n ( Hom A ( x , x ) , m 1 x , x ) .
Two such S , S ′ S,S' S , S ′ are homotopic if there is an ( n + 1 ) (n+1) ( n + 1 ) -simplex T T T in N A ∞ ( A ) x Kan \mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_x N A ∞ ( A ) x Kan such that two prescribed Δ n \Delta^n Δ n -faces of T T T are S , S ′ S,S' S , S ′ , and the rest of T ∣ ∂ Δ n + 1 T|_{\partial\Delta^{n+1}} T ∣ ∂ Δ n + 1 is constant at x x x . By Corollary 4.56(a), we find that T T T is equivalent to T n + 1 01 ⋯ n + 1 T^{01\cdots n+1}_{n+1} T n + 1 01 ⋯ n + 1 in Hom A − n ( x , x ) \operatorname{Hom}^{-n}_{\mathcal A}(x,x) Hom A − n ( x , x ) satisfying m 1 x , x ( T n + 1 01 ⋯ n + 1 ) = S n ′ 01 ⋯ n − S n 01 ⋯ n m^{x,x}_{1}(T^{01\cdots n+1}_{n+1})=S'^{01\cdots n}_{n}-S^{01\cdots n}_{n} m 1 x , x ( T n + 1 01 ⋯ n + 1 ) = S n ′ 01 ⋯ n − S n 01 ⋯ n . Thus S , S ′ S,S' S , S ′ are homotopic if and only if [ S n 01 ⋯ n ] = [ S n ′ 01 ⋯ n ] [S^{01\cdots n}_{n}]=[S'^{01\cdots n}_{n}] [ S n 01 ⋯ n ] = [ S n ′ 01 ⋯ n ] in H 1 − n ( Hom A ( x , x ) , m 1 x , x ) H^{1-n}(\operatorname{Hom}_{\mathcal A}(x,x),m^{x,x}_{1}) H 1 − n ( Hom A ( x , x ) , m 1 x , x ) .
This proves (4.42) as sets. To see the group operation is addition, [ S n 01 ⋯ n ] ⋅ [ S n ′ 01 ⋯ n ] = [ S n 01 ⋯ n + S n ′ 01 ⋯ n ] [S^{01\cdots n}_{n}]\cdot[S'^{01\cdots n}_{n}]=[S^{01\cdots n}_{n}+S'^{01\cdots n}_{n}] [ S n 01 ⋯ n ] ⋅ [ S n ′ 01 ⋯ n ] = [ S n 01 ⋯ n + S n ′ 01 ⋯ n ] , consider an ( n + 1 ) (n+1) ( n + 1 ) -simplex T T T in N A ∞ ( A ) x Kan \mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_x N A ∞ ( A ) x Kan which has three prescribed Δ n \Delta^n Δ n -faces S , S ′ , S + S ′ S,S',S+S' S , S ′ , S + S ′ , and the rest of T ∣ ∂ Δ n + 1 T|_{\partial\Delta^{n+1}} T ∣ ∂ Δ n + 1 is constant at x x x , and T n + 1 01 ⋯ n + 1 = 0 T^{01\cdots n+1}_{n+1}=0 T n + 1 01 ⋯ n + 1 = 0 , and check that (4.39) holds.
Note that the next corollary only needs F \mathcal F F to be a quasi-equivalence, not a homotopy equivalence.
Corollary 4.59. Suppose F : A → B \mathcal F:\mathcal A \to\mathcal B F : A → B is a unital quasi-equivalence of unital A ∞ A_\infty A ∞ -categories over a commutative ring R R R . Then the following is a homotopy equivalence:
N A ∞ ( F ) Kan : N A ∞ ( A ) Kan ⟶ N A ∞ ( B ) Kan . \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}}:\mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}} \longrightarrow\mathcal N_{\mathcal A_\infty}(\mathcal B)^{\operatorname{Kan}}. N A ∞ ( F ) Kan : N A ∞ ( A ) Kan ⟶ N A ∞ ( B ) Kan . Proof. Connected components of N A ∞ ( A ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}} N A ∞ ( A ) Kan and N A ∞ ( B ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal B)^{\operatorname{Kan}} N A ∞ ( B ) Kan correspond to isomorphism classes of objects in H 0 ( A ) H^{0}(\mathcal A) H 0 ( A ) and H 0 ( B ) H^{0}(\mathcal B) H 0 ( B ) , so N A ∞ ( F ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}} N A ∞ ( F ) Kan induces a 1-1 correspondence of connected components as H 0 ( F ) : H 0 ( A ) → H 0 ( B ) H^{0}(\mathcal F):H^{0}(\mathcal A)\to H^{0}(\mathcal B) H 0 ( F ) : H 0 ( A ) → H 0 ( B ) is an equivalence of categories.
Let x ∈ Obj A x \in\operatorname{Obj}\mathcal A x ∈ Obj A , so that N A ∞ ( F ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}} N A ∞ ( F ) Kan maps N A ∞ ( A ) x Kan → N A ∞ ( B ) F ( x ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_{x} \to\mathcal N_{\mathcal A_\infty}(\mathcal B)^{\operatorname{Kan}}_{\mathcal F(x)} N A ∞ ( A ) x Kan → N A ∞ ( B ) F ( x ) Kan . Then N A ∞ ( F ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}} N A ∞ ( F ) Kan gives an isomorphism π 1 ( N A ∞ ( A ) x Kan ) → π 1 ( N A ∞ ( B ) F ( x ) Kan ) \pi_1(\mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_{x}) \to\pi_1(\mathcal N_{\mathcal A_\infty}(\mathcal B)^{\operatorname{Kan}}_{\mathcal F(x)}) π 1 ( N A ∞ ( A ) x Kan ) → π 1 ( N A ∞ ( B ) F ( x ) Kan ) by (4.41), as H 0 ( F ) H^{0}(\mathcal F) H 0 ( F ) is an equivalence of categories. Also N A ∞ ( F ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}} N A ∞ ( F ) Kan induces an isomorphism π n ( N A ∞ ( A ) x Kan ) → π n ( N A ∞ ( B ) F ( x ) Kan ) \pi_n(\mathcal N_{\mathcal A_\infty}(\mathcal A)^{\operatorname{Kan}}_{x}) \to\pi_n(\mathcal N_{\mathcal A_\infty}(\mathcal B)^{\operatorname{Kan}}_{\mathcal F(x)}) π n ( N A ∞ ( A ) x Kan ) → π n ( N A ∞ ( B ) F ( x ) Kan ) for n ≥ 2 n\ge2 n ≥ 2 by (4.42), as F 1 x , x : ( Hom A ( x , x ) , m 1 x , x ) → ( Hom B ( F ( x ) , F ( x ) ) , m 1 F ( x ) , F ( x ) ) \mathcal F^{x,x}_{1}:(\operatorname{Hom}_{\mathcal A}(x,x),m^{x,x}_{1})\to(\operatorname{Hom}_{\mathcal B}(\mathcal F(x),\mathcal F(x)),m^{\mathcal F(x),\mathcal F(x)}_{1}) F 1 x , x : ( Hom A ( x , x ) , m 1 x , x ) → ( Hom B ( F ( x ) , F ( x )) , m 1 F ( x ) , F ( x ) ) is a quasi-isomorphism. Hence N A ∞ ( F ) Kan \mathcal N_{\mathcal A_\infty}(\mathcal F)^{\operatorname{Kan}} N A ∞ ( F ) Kan is a weak homotopy equivalence, so it is a homotopy equivalence by the Whitehead Theorem.
Pretriangulated A ∞ A_\infty A ∞ -categories and the twist functor
Definition 4.60. As in Gelfand and Manin [35], a triangulated category T \mathcal{T} T is an additive category equipped with a ‘translation functor’ and a class of ‘distinguished triangles’ satisfying some axioms. They appear in many places in mathematics. Derived categories D b C D^b\mathcal{C} D b C of abelian categories C \mathcal{C} C , such as derived categories of coherent sheaves D c o h b X D^b_{\mathrm{coh}}X D coh b X , are important examples.
Let R R R be a commutative ring, and A \mathcal{A} A an A ∞ A_\infty A ∞ -category over R R R , which should be weakly cohomologically unital, or cohomologically unital, or unital. There is a notion of when A \mathcal{A} A is pretriangulated , defined in Bespalov–Lyubashenko–Manzyuk [4] and Seidel [84, §3h] (who calls them triangulated ). They are defined directly in [4, 84] in terms of the existence of translations and distinguished triangles. We will not write this out, but instead give an indirect definition in Definition 4.62 using the twist functor.
If A \mathcal{A} A is a pretriangulated A ∞ A_\infty A ∞ -category, its cohomology category H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) is triangulated. Note that although to be triangulated is additional structure on a category, to be pretriangulated is a property of an A ∞ A_\infty A ∞ -category. We write
A ∞ △ Cat R u n ⊂ A ∞ Cat R u n , A ∞ △ Cat R w c ⊂ A ∞ Cat R w c , A ∞ △ Cat R c u ⊂ A ∞ Cat R c u , A^\triangle_\infty\operatorname{Cat}^{\mathrm{un}}_R \subset A_\infty\operatorname{Cat}^{\mathrm{un}}_R,\quad A^\triangle_\infty\operatorname{Cat}^{\mathrm{wc}}_R \subset A_\infty\operatorname{Cat}^{\mathrm{wc}}_R,\quad A^\triangle_\infty\operatorname{Cat}^{\mathrm{cu}}_R \subset A_\infty\operatorname{Cat}^{\mathrm{cu}}_R, A ∞ △ Cat R un ⊂ A ∞ Cat R un , A ∞ △ Cat R wc ⊂ A ∞ Cat R wc , A ∞ △ Cat R cu ⊂ A ∞ Cat R cu ,
for the full subcategories whose objects are pretriangulated. For the subcategories A ∞ Cat R u n , s u , … , A ∞ Cat R c u , h p A_\infty\operatorname{Cat}^{\mathrm{un,su}}_R,\ldots,A_\infty\operatorname{Cat}^{\mathrm{cu,hp}}_R A ∞ Cat R un , su , … , A ∞ Cat R cu , hp in (122) , we also write A ∞ △ Cat R u n , s u , … , A ∞ △ Cat R c u , h p A^\triangle_\infty\operatorname{Cat}^{\mathrm{un,su}}_R,\ldots,A^\triangle_\infty\operatorname{Cat}^{\mathrm{cu,hp}}_R A ∞ △ Cat R un , su , … , A ∞ △ Cat R cu , hp for the full subcategories of these categories whose objects are pretriangulated.
The next theorem is proved by Faonte [23, Th. 3.18] for pretriangulated dg-categories over a field, and by Ornaghi [75, Th. 5.1] in the general case.
Theorem 4.61. Let A \mathcal{A} A be a pretriangulated unital A ∞ A_\infty A ∞ -category over a commutative ring R R R . Then the simplicial nerve N A ∞ ( A ) \mathcal{N}_{A_\infty}(\mathcal{A}) N A ∞ ( A ) from Definition 4.53 is a stable ∞ \infty ∞ -category in the sense of §2.7. The strict isomorphism H 0 ( A ) ≅ Ho ( N A ∞ ( A ) ) H^0(\mathcal{A}) \cong\operatorname{Ho}(\mathcal{N}_{A_\infty}(\mathcal{A})) H 0 ( A ) ≅ Ho ( N A ∞ ( A )) in Proposition 4.54 is an isomorphism of triangulated categories.
Definition 4.62. Let R R R be a commutative ring. The twist functor is a functor
Tw : A ∞ Cat R ⟶ A ∞ Cat R , \operatorname{Tw}: A_\infty\operatorname{Cat}_R \longrightarrow A_\infty\operatorname{Cat}_R, Tw : A ∞ Cat R ⟶ A ∞ Cat R ,
defined in Seidel [84, §3], Keller [51, §7.6], Lefèvre-Hasegawa [57, §7.2], and Bespalov–Lyubashenko–Manzyuk [4, §12]. We will construct it explicitly in §5.8, using filtered A ∞ A_\infty A ∞ -categories; doing it this way will make it easy to generalize Tw \operatorname{Tw} Tw to ∞ \infty ∞ -categories in Chapter 9. Here we explain it informally, and review its properties.
The twist functor preserved the properties of being weakly cohomologically unital, cohomologically unital, and unital, and if A \mathcal{A} A is at least weakly cohomologically unital then Tw A \operatorname{Tw}\mathcal{A} Tw A is pretriangulated. Thus Tw \operatorname{Tw} Tw restricts to functors
Tw : A ∞ Cat R w c ⟶ A ∞ △ Cat R w c , Tw : A ∞ Cat R c u ⟶ A ∞ △ Cat R c u , \operatorname{Tw}: A_\infty\operatorname{Cat}^{\mathrm{wc}}_R \longrightarrow A^\triangle_\infty\operatorname{Cat}^{\mathrm{wc}}_R,\qquad\operatorname{Tw}: A_\infty\operatorname{Cat}^{\mathrm{cu}}_R \longrightarrow A^\triangle_\infty\operatorname{Cat}^{\mathrm{cu}}_R, Tw : A ∞ Cat R wc ⟶ A ∞ △ Cat R wc , Tw : A ∞ Cat R cu ⟶ A ∞ △ Cat R cu ,
Tw : A ∞ Cat R u n ⟶ A ∞ △ Cat R u n . (129) \operatorname{Tw}: A_\infty\operatorname{Cat}^{\mathrm{un}}_R \longrightarrow A^\triangle_\infty\operatorname{Cat}^{\mathrm{un}}_R. \tag*{(129)} Tw : A ∞ Cat R un ⟶ A ∞ △ Cat R un . (129)
For each A ∞ A_{\infty} A ∞ -category A \mathcal{A} A there is an inclusion inc A : A ↪ Tw A \operatorname{inc}_{\mathcal{A}}:\mathcal{A}\hookrightarrow\operatorname{Tw}\mathcal{A} inc A : A ↪ Tw A , defined explicitly in § 5.8 , which is a strict isomorphism with a full A ∞ A_{\infty} A ∞ -subcategory, so that H 0 ( inc A ) : H 0 ( A ) ↪ H 0 ( Tw A ) H^{0}(\operatorname{inc}_{\mathcal{A}}):H^{0}(\mathcal{A})\hookrightarrow H^{0}(\operatorname{Tw}\mathcal{A}) H 0 ( inc A ) : H 0 ( A ) ↪ H 0 ( Tw A ) is full and faithful. Taken over all A \mathcal{A} A these form a natural transformation inc : Id ⇒ Tw \operatorname{inc}:\operatorname{Id}\Rightarrow\operatorname{Tw} inc : Id ⇒ Tw of functors A ∞ Cat R → A ∞ Cat R A_{\infty}\operatorname{Cat}_{R}\to A_{\infty}\operatorname{Cat}_{R} A ∞ Cat R → A ∞ Cat R . Composing on the left or right with Tw \operatorname{Tw} Tw gives two natural transformations id Tw ∗ inc , inc ∗ id Tw : Tw ⇒ Tw ∘ Tw \operatorname{id}_{\operatorname{Tw}}\ast\operatorname{inc},\operatorname{inc}\ast\operatorname{id}_{\operatorname{Tw}}:\operatorname{Tw}\Rightarrow\operatorname{Tw}\circ\operatorname{Tw} id Tw ∗ inc , inc ∗ id Tw : Tw ⇒ Tw ∘ Tw . For each A \mathcal{A} A these give A ∞ A_{\infty} A ∞ -functors Tw ( inc A ) , inc Tw ( A ) : Tw A ↪ Tw Tw A \operatorname{Tw}(\operatorname{inc}_{\mathcal{A}}),\operatorname{inc}_{\operatorname{Tw}(\mathcal{A})}:\operatorname{Tw}\mathcal{A}\hookrightarrow\operatorname{Tw}\operatorname{Tw}\mathcal{A} Tw ( inc A ) , inc Tw ( A ) : Tw A ↪ Tw Tw A . These are full and faithful embeddings, strictly A ∞ A_{\infty} A ∞ -naturally isomorphic to each other, and they are homotopy equivalences for unital A \mathcal{A} A [4, Prop. 12.11]. A quasi-inverse Tw Tw A → Tw A \operatorname{Tw}\operatorname{Tw}\mathcal{A}\to\operatorname{Tw}\mathcal{A} Tw Tw A → Tw A is constructed in [4, §12.2.3].
As in [84, Lem. 3.33] and [4, Prop. 12.10], a weakly cohomological A ∞ A_{\infty} A ∞ -category A \mathcal{A} A is pretriangulated if and only if inc A : A ↪ Tw A \operatorname{inc}_{\mathcal{A}}:\mathcal{A}\hookrightarrow\operatorname{Tw}\mathcal{A} inc A : A ↪ Tw A is a quasi-equivalence (this happens provided inc A \operatorname{inc}_{\mathcal{A}} inc A is essentially surjective, and then inc A \operatorname{inc}_{\mathcal{A}} inc A is also a homotopy equivalence). Thus, we can define A \mathcal{A} A to be pretriangulated if inc A : A ↪ Tw A \operatorname{inc}_{\mathcal{A}}:\mathcal{A}\hookrightarrow\operatorname{Tw}\mathcal{A} inc A : A ↪ Tw A is a homotopy equivalence. If A , B \mathcal{A},\mathcal{B} A , B are homotopy equivalent then A \mathcal{A} A is pretriangulated if and only if B \mathcal{B} B is.
The definition of Tw A \operatorname{Tw}\mathcal{A} Tw A is similar to the definition of the bounded derived category D b C D^{b}\mathcal{C} D b C of an abelian category C \mathcal{C} C . Objects of Tw A \operatorname{Tw}\mathcal{A} Tw A are twisted complexes in A \mathcal{A} A . Basically, we make Tw A \operatorname{Tw}\mathcal{A} Tw A by starting from A \mathcal{A} A and adding many more objects: shifts x [ k ] x[k] x [ k ] for x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A and k ∈ Z k\in\mathbb{Z} k ∈ Z , direct sums x ⊕ y x\oplus y x ⊕ y of objects x , y x,y x , y , and iterated extensions of objects x 1 , … , x k x_{1},\ldots,x_{k} x 1 , … , x k , to make Tw A \operatorname{Tw}\mathcal{A} Tw A closed under shifts [ k ] [k] [ k ] and extensions in exact triangles.
Remark 4.63. Here is an important application of the twist functor. If ( X , ω ) (X,\omega) ( X , ω ) is a symplectic manifold, one can try to define the Fukaya category F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) , a cohomologically unital A ∞ A_{\infty} A ∞ -category with objects Lagrangian submanifolds L L L in ( X , ω ) (X,\omega) ( X , ω ) (plus extra data). Then the derived Fukaya category is D b F ( X , ω ) = H 0 ( Tw ( F ( X , ω ) ) ) D^{b}\mathcal{F}(X,\omega)=H^{0}(\operatorname{Tw}(\mathcal{F}(X,\omega))) D b F ( X , ω ) = H 0 ( Tw ( F ( X , ω ))) , a triangulated category.
Kontsevich’s Homological Mirror Symmetry Conjecture [53] explains Mirror Symmetry of Calabi–Yau manifolds X , X ˇ X,\check{X} X , X ˇ in terms of equivalences of triangulated categories D b F ( X , ω ) ≃ D b coh X ˇ D^{b}\mathcal{F}(X,\omega)\simeq D^{b}\operatorname{coh}\check{X} D b F ( X , ω ) ≃ D b coh X ˇ , D b F ( X ˇ , ω ˇ ) ≃ D b coh X D^{b}\mathcal{F}(\check{X},\check{\omega})\simeq D^{b}\operatorname{coh}X D b F ( X ˇ , ω ˇ ) ≃ D b coh X .
∞-categories of A∞-categories as localizations
We now discuss work by Canonaco–Ornaghi–Stellari [11, 12], Faonte [24], Oh–Tanaka [72], Ornaghi [73], Tanaka [88], and Pascaleff [76], which studies ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories (or dg-categories) defined by starting with the category A ∞ Cat R u n A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un (or some variant), and localizing the quasi-equivalences Q E u n ⊂ Mor ( A ∞ Cat R u n ) QE_{\mathrm{un}}\subset\operatorname{Mor}(A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}) Q E un ⊂ Mor ( A ∞ Cat R un ) from Definition 4.28, using ∞ \infty ∞ -category localization from § 2.6 , to get an ∞ \infty ∞ -category A ∞ Cat R u n [ Q E u n − 1 ] A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[QE_{\mathrm{un}}^{-1}] A ∞ Cat R un [ Q E un − 1 ] . So for them (unlike us) the definition of their ∞ \infty ∞ -category is trivial, and they are concerned with its properties.
The next theorem is proved by Canonaco–Ornaghi–Stellari [12, Th. A & Rem. 3.13]. Some of the equivalences in (130) were proved earlier by Pascaleff [76, Cor. 5.2] when R R R is a field, and by Tanaka [88, Th. 1.1].
Theorem 4.64. For any commutative ring R R R , the inclusions of subcategories
DGCat R ↪ DGCat R A ∞ ↪ A ∞ Cat R u n ↪ A ∞ Cat R c u ↪ A ∞ Cat R w c \operatorname{DGCat}_{R} \hookrightarrow\operatorname{DGCat}_{R}^{A_{\infty}} \hookrightarrow A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} \hookrightarrow A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}} \hookrightarrow A_{\infty}\operatorname{Cat}_{R}^{\mathrm{wc}} DGCat R ↪ DGCat R A ∞ ↪ A ∞ Cat R un ↪ A ∞ Cat R cu ↪ A ∞ Cat R wc induce equivalences of ∞ \infty ∞ -categories
DGCat R [ Q E d g − 1 ] ∞ ≃ DGCat R A ∞ [ Q E d g , A ∞ − 1 ] ∞ ≃ A ∞ Cat R u n [ Q E u n − 1 ] ∞ ≃ \operatorname{DGCat}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_{\infty} \simeq\operatorname{DGCat}_{R}^{A_{\infty}}[\mathcal{Q}\mathcal{E}_{\mathrm{dg},A_{\infty}}^{-1}]_{\infty} \simeq A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} \simeq DGCat R [ Q E dg − 1 ] ∞ ≃ DGCat R A ∞ [ Q E dg , A ∞ − 1 ] ∞ ≃ A ∞ Cat R un [ Q E un − 1 ] ∞ ≃ A ∞ Cat R c u [ Q E c u − 1 ] ∞ ≃ A ∞ Cat R w c [ Q E w c − 1 ] ∞ . (130) A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}}[\mathcal{Q}\mathcal{E}_{\mathrm{cu}}^{-1}]_{\infty} \simeq A_{\infty}\operatorname{Cat}_{R}^{\mathrm{wc}}[\mathcal{Q}\mathcal{E}_{\mathrm{wc}}^{-1}]_{\infty}. \tag*{(130)} A ∞ Cat R cu [ Q E cu − 1 ] ∞ ≃ A ∞ Cat R wc [ Q E wc − 1 ] ∞ . (130)
Combining Theorem 4.64 with Corollary 4.36 and Theorem 4.38 yields:
Corollary 4.65. (a) The ∞ \infty ∞ -category A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ has all small homotopy limits and homotopy colimits.
(b) The ∞ \infty ∞ -category A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ is presentable.
(c) There is a symmetric monoidal structure ⊗ ∞ \otimes^{\infty} ⊗ ∞ on A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ .
(d) The monoidal structure ⊗ ∞ \otimes^{\infty} ⊗ ∞ is closed, so there exist internal mapping objects Hom ( C , D ) \operatorname{Hom}(\mathcal{C},\mathcal{D}) Hom ( C , D ) in A ∞ Cat R u n [ Q E u n − 1 ] ∞ A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ with homotopy equivalences
Hom A ∞ Cat R u n [ Q E u n − 1 ] ∞ ( B , Hom ( C , D ) ) \operatorname{Hom}_{A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}\left(\mathcal{B},\operatorname{Hom}(\mathcal{C},\mathcal{D})\right)} Hom A ∞ Cat R un [ Q E un − 1 ] ∞ ( B , Hom ( C , D ) ) ≃ Hom A ∞ Cat R u n [ Q E u n − 1 ] ∞ ( B ⊗ ∞ C , D ) . \simeq\operatorname{Hom}_{A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}\left(\mathcal{B}\otimes^{\infty}\mathcal{C},\mathcal{D}\right).} ≃ Hom A ∞ Cat R un [ Q E un − 1 ] ∞ ( B ⊗ ∞ C , D ) .
The analogues hold for A ∞ Cat R c u [ Q E c u − 1 ] ∞ A_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu}}[\mathcal{Q}\mathcal{E}_{\mathrm{cu}}^{-1}]_{\infty} A ∞ Cat R cu [ Q E cu − 1 ] ∞ .
We noted in §4.7 that the authors of [11,12,24,72,73,76,88] treat A ∞ Cat R u n A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un as though it was a model category, in which weak equivalences are quasi-equivalences, and cofibrant objects are h-projective unital A ∞ A_{\infty} A ∞ -categories with split units. The next three theorems use this idea, since Theorems 4.66 –4.67 are like using cofibrant replacements in a model category, and Theorem 4.68 is an analogue of Remark 2.35(d) for model categories.
Here is Canonaco–Ornaghi–Stellari [12, Th. C], which proves a claim by Kontsevich. A version was claimed earlier by Faonte [24, Th. 1.2] for R R R a field of characteristic zero. However, Canonaco et al. [11, p. 4] point out errors in [24]. In particular, [24, Th. 1.2] is false as stated, as it has no cofibrant replacement of C \mathcal{C} C by C ′ \mathcal{C}' C ′ , but this is necessary even for R R R a field if C \mathcal{C} C has zero objects.
Theorem 4.66. Let C , D \mathcal{C},\mathcal{D} C , D be dg-categories over a commutative ring R R R . Then as in Theorem 4.38 (b) there is an internal mapping object Hom ( C , D ) \operatorname{Hom}(\mathcal{C},\mathcal{D}) Hom ( C , D ) in the ∞ \infty ∞ -category DGCat R [ Q E d g − 1 ] ∞ \operatorname{DGCat}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_{\infty} DGCat R [ Q E dg − 1 ] ∞ , which is explicitly described in Theorem 4.38 (b).
We can give an alternative description of Hom ( C , D ) \operatorname{Hom}(\mathcal{C},\mathcal{D}) Hom ( C , D ) as
Hom ( C , D ) ≃ Fun A ∞ u n ( C ′ , D ) . \operatorname{Hom}(\mathcal{C},\mathcal{D}) \simeq\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{C}',\mathcal{D}). Hom ( C , D ) ≃ Fun A ∞ un ( C ′ , D ) . Here Φ : C ′ → C \Phi:\mathcal{C}'\to\mathcal{C} Φ : C ′ → C is a quasi-equivalence in DGCat R \operatorname{DGCat}_{R} DGCat R such that C ′ \mathcal{C}' C ′ is h-projective with split units, which exists by Theorem 4.52 , and Fun A ∞ u n ( C ′ , D ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{C}',\mathcal{D}) Fun A ∞ un ( C ′ , D ) is the A ∞ A_{\infty} A ∞ -category from §4.3, which is a dg-category in this case as C ′ , D \mathcal{C}',\mathcal{D} C ′ , D are dg-categories. Hence by Theorem 4.38 (d) we have a homotopy equivalence
Hom DGCat R [ Q E d g − 1 ] ∞ ( C , D ) ≃ N A ∞ ( Fun A ∞ u n ( C ′ , D ) ) K a n , \operatorname{Hom}_{\operatorname{DGCat}_{R}[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_{\infty}}(\mathcal{C},\mathcal{D}) \simeq\mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{C}',\mathcal{D})\right)^{\mathrm{Kan}}, Hom DGCat R [ Q E dg − 1 ] ∞ ( C , D ) ≃ N A ∞ ( Fun A ∞ un ( C ′ , D ) ) Kan , where N A ∞ ( − ) \mathcal{N}_{A_{\infty}}(-) N A ∞ ( − ) is as in §4.8, and coincides with N d g ( − ) \mathcal{N}_{\mathrm{dg}}(-) N dg ( − ) for dg-categories.
The next theorem is taken from Tanaka [88, Th.s 3.36 & 3.37]. It generalizes Theorem 4.66 to A ∞ A_{\infty} A ∞ -categories.
Theorem 4.67. Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R , and let Φ : A ′ → A \Phi:\mathcal{A}'\to\mathcal{A} Φ : A ′ → A be an h h h -projective resolution of A \mathcal{A} A , as in Theorem 4.52 , such that A ′ \mathcal{A}' A ′ has split units, as in Definition 4.39. Then the internal mapping object Hom ( A , B ) \operatorname{Hom}(\mathcal{A},\mathcal{B}) Hom ( A , B ) in A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}[Q\!E^{-1}_{\mathrm{un}}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ given by Corollary 4.65 (d) satisfies
Hom ( A , B ) ≃ Fun A ∞ u n ( A ′ , B ) , \operatorname{Hom}(\mathcal{A},\mathcal{B}) \simeq\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A}',\mathcal{B}), Hom ( A , B ) ≃ Fun A ∞ un ( A ′ , B ) , where Fun A ∞ u n ( A ′ , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A}',\mathcal{B}) Fun A ∞ un ( A ′ , B ) is as in §4.3. Also, we have a homotopy equivalence
Hom A ∞ C a t R u n [ Q E u n − 1 ] ∞ ( A , B ) ≃ N A ∞ ( Fun A ∞ u n ( A ′ , B ) ) K a n , (131) \operatorname{Hom}_{A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[Q\!E^{-1}_{\mathrm{un}}]_{\infty}}(\mathcal{A},\mathcal{B}) \simeq\mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A}',\mathcal{B})\right)^{\mathrm{Kan}}, \tag*{(131)} Hom A ∞ Cat R un [ Q E un − 1 ] ∞ ( A , B ) ≃ N A ∞ ( Fun A ∞ un ( A ′ , B ) ) Kan , (131) where Hom A ∞ C a t R u n [ Q E u n − 1 ] ∞ ( A , B ) \operatorname{Hom}_{A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[Q\!E^{-1}_{\mathrm{un}}]_{\infty}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un [ Q E un − 1 ] ∞ ( A , B ) is the mapping space as in §2.3 in the ∞ \infty ∞ -category A ∞ C a t R u n [ Q E u n − 1 ] ∞ A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[Q\!E^{-1}_{\mathrm{un}}]_{\infty} A ∞ Cat R un [ Q E un − 1 ] ∞ , and N A ∞ ( Fun A ∞ u n ( A ′ , B ) ) K a n \mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A}',\mathcal{B})\right)^{\mathrm{Kan}} N A ∞ ( Fun A ∞ un ( A ′ , B ) ) Kan is the maximal Kan complex in the simplicial nerve of the A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( A ′ , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A}',\mathcal{B}) Fun A ∞ un ( A ′ , B ) , as in §4.8.
If a dg-category A \mathcal{A} A is cofibrant in the Tabuada model structure from Theorem 4.35 then it is h h h -projective with split units. Because of this, by the argument in Tanaka [88, Cons. 3.20 & Rem.s 3.21 & 3.22] we can show:
Theorem 4.68. Let R R R be a commutative ring. Then the inclusion of subcategories D G C a t R h s ↪ D G C a t R \mathbf{DGCat}^{\mathrm{hs}}_{R}\hookrightarrow\mathbf{DGCat}_{R} DGCat R hs ↪ DGCat R from Definition 4.47 induces an equivalence of ∞ \infty ∞ -categories
D G C a t R h s [ H E d g , h s − 1 ] ∞ ≃ D G C a t R [ Q E d g − 1 ] ∞ . (132) \mathbf{DGCat}^{\mathrm{hs}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{dg,hs}}]_{\infty}\simeq\mathbf{DGCat}_{R}[Q\!E^{-1}_{\mathrm{dg}}]_{\infty}. \tag*{(132)} DGCat R hs [ H E dg , hs − 1 ] ∞ ≃ DGCat R [ Q E dg − 1 ] ∞ . (132)
Remark 4.69. If it were true (it is not) that A ∞ C a t R u n , A ∞ C a t R c u A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}, A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R un , A ∞ Cat R cu were model categories in which weak equivalences are quasi-equivalences, and cofibrant objects in A ∞ C a t R u n A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un are h h h -projective with split units, and cofibrant objects in A ∞ C a t R c u A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R cu are h h h -projective, then by Remark 2.35 (d) and Corollary 4.49 we could deduce that the inclusions of subcategories A ∞ C a t R u n , h s ↪ A ∞ C a t R u n A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R}\hookrightarrow A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un , hs ↪ A ∞ Cat R un and A ∞ C a t R c u , h p ↪ A ∞ C a t R c u A_{\infty}\mathrm{Cat}^{\mathrm{cu,hp}}_{R}\hookrightarrow A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R cu , hp ↪ A ∞ Cat R cu from Definition 4.47 induce equivalences
A ∞ C a t R u n , h s [ H E u n , h s − 1 ] ∞ ≃ A ∞ C a t R u n [ Q E u n − 1 ] ∞ , (133) A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{un,hs}}]_{\infty}\simeq A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[Q\!E^{-1}_{\mathrm{un}}]_{\infty}, \tag*{(133)} A ∞ Cat R un , hs [ H E un , hs − 1 ] ∞ ≃ A ∞ Cat R un [ Q E un − 1 ] ∞ , (133) A ∞ C a t R c u , h p [ H E c u , h p − 1 ] ∞ ≃ A ∞ C a t R c u [ Q E c u − 1 ] ∞ . (134) A_{\infty}\mathrm{Cat}^{\mathrm{cu,hp}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{cu,hp}}]_{\infty}\simeq A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R}[Q\!E^{-1}_{\mathrm{cu}}]_{\infty}. \tag*{(134)} A ∞ Cat R cu , hp [ H E cu , hp − 1 ] ∞ ≃ A ∞ Cat R cu [ Q E cu − 1 ] ∞ . (134)
Following the philosophy of [11,12,24,72,73,76,88], one might expect (133) –(134) to be true even though A ∞ C a t R u n , A ∞ C a t R c u A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}, A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R} A ∞ Cat R un , A ∞ Cat R cu are not model categories. The author cannot find a proof of these in the literature. But we will prove (133) ourselves in Theorem 9.6 below.
Faonte's ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) - and ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories of A ∞ A_{\infty} A ∞ -categories
We now review some ideas due to Faonte [24]. We note two things: firstly, Faonte works over a field K K K of characteristic zero. However, the parts of [24] we discuss work over a general commutative ring R R R , and we will state them like this. Secondly, Canonaco–Ornaghi–Stellari [11, Intro] point out errors in [24], which affect [24, §A.4] and [24, Th. 1.2], which is false as stated.
Definition 4.70. Let A , B , C \mathcal{A},\mathcal{B},\mathcal{C} A , B , C be unital A ∞ A_{\infty} A ∞ -categories over R R R . In Remark 4.20 we explained how work of Lyubashenko [64] defines an ‘A ∞ A_{\infty} A ∞ -bifunctor’
κ A B C u n : Fun A ∞ u n ( A , B ) × Fun A ∞ u n ( B , C ) ⟶ Fun A ∞ u n ( A , C ) , \kappa^{\mathrm{un}}_{\mathcal{A}\mathcal{B}\mathcal{C}}:\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\times\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{B},\mathcal{C})\longrightarrow\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{C}), κ A BC un : Fun A ∞ un ( A , B ) × Fun A ∞ un ( B , C ) ⟶ Fun A ∞ un ( A , C ) , which acts on objects as ( F , G ) ↦ G ∘ F (\mathcal{F},\mathcal{G})\mapsto\mathcal{G}\circ\mathcal{F} ( F , G ) ↦ G ∘ F .
Faonte [24, §3.4, Def. 3.18] uses κ A B C u n \kappa^{\mathrm{un}}_{\mathcal{A}\mathcal{B}\mathcal{C}} κ A BC un (actually, the corresponding functor of counital dg-categories, as in (112) ) to construct an explicit ∞ \infty ∞ -functor
λ A B C : N ∞ ( Fun A ∞ u n ( A , B ) ) × N ∞ ( Fun A ∞ u n ( B , C ) ) ⟶ N ∞ ( Fun A ∞ u n ( A , C ) ) . (135) \lambda_{\mathcal{A}\mathcal{B}\mathcal{C}}:\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\times\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{B},\mathcal{C})\bigr)\longrightarrow\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{C})\bigr). \tag*{(135)} λ A BC : N ∞ ( Fun A ∞ un ( A , B ) ) × N ∞ ( Fun A ∞ un ( B , C ) ) ⟶ N ∞ ( Fun A ∞ un ( A , C ) ) . (135) There are no arbitrary choices involved, λ A B C \lambda_{\mathcal{A}\mathcal{B}\mathcal{C}} λ A BC is defined as an explicit map on N N N -simplices. Restricting to maximal Kan complexes ( − ) K a n (-)^{\mathrm{Kan}} ( − ) Kan gives a morphism
λ A B C K a n : N ∞ ( Fun A ∞ u n ( A , B ) ) K a n × N ∞ ( Fun A ∞ u n ( B , C ) ) K a n ⟶ N ∞ ( Fun A ∞ u n ( A , C ) ) K a n . (136) \lambda^{\mathrm{Kan}}_{\mathcal{A}\mathcal{B}\mathcal{C}}:\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)^{\mathrm{Kan}}\times\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{B},\mathcal{C})\bigr)^{\mathrm{Kan}}\longrightarrow\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{C})\bigr)^{\mathrm{Kan}}. \tag*{(136)} λ A BC Kan : N ∞ ( Fun A ∞ un ( A , B ) ) Kan × N ∞ ( Fun A ∞ un ( B , C ) ) Kan ⟶ N ∞ ( Fun A ∞ un ( A , C ) ) Kan . (136) Note that spaces N ∞ ( Fun A ∞ u n ( A , B ) ) K a n \mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)^{\mathrm{Kan}} N ∞ ( Fun A ∞ un ( A , B ) ) Kan are models for mapping spaces in ∞ \infty ∞ -categories in Theorems 4.66 and 4.67 .
Faonte [24, Prop. 3.19] proves:
Proposition 4.71. Let A , B , C , D \mathcal{A},\mathcal{B},\mathcal{C},\mathcal{D} A , B , C , D be unital A ∞ A_{\infty} A ∞ -categories over R R R . Then the ∞ \infty ∞ -functors λ A B C \lambda_{\mathcal{A}\mathcal{B}\mathcal{C}} λ A BC in (135) satisfy the identities
λ A B D ∘ ( id ⋯ ( A , B ) × λ B C D ) = λ A C D ∘ ( λ A B C × id ⋯ ( C , D ) ) : \lambda_{\mathcal{A}\mathcal{B}\mathcal{D}}\circ\bigl(\operatorname{id}_{\cdots(\mathcal{A},\mathcal{B})}\times\lambda_{\mathcal{B}\mathcal{C}\mathcal{D}}\bigr)=\lambda_{\mathcal{A}\mathcal{C}\mathcal{D}}\circ\bigl(\lambda_{\mathcal{A}\mathcal{B}\mathcal{C}}\times\operatorname{id}_{\cdots(\mathcal{C},\mathcal{D})}\bigr): λ A B D ∘ ( id ⋯ ( A , B ) × λ BC D ) = λ A C D ∘ ( λ A BC × id ⋯ ( C , D ) ) : N ∞ ( Fun A ∞ u n ( A , B ) ) × N ∞ ( Fun A ∞ u n ( B , C ) ) × N ∞ ( Fun A ∞ u n ( C , D ) ) ⟶ N ∞ ( Fun A ∞ u n ( A , D ) ) , (137) \mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\times\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{B},\mathcal{C})\bigr)\times\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C},\mathcal{D})\bigr)\longrightarrow\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{D})\bigr), \tag*{(137)} N ∞ ( Fun A ∞ un ( A , B ) ) × N ∞ ( Fun A ∞ un ( B , C ) ) × N ∞ ( Fun A ∞ un ( C , D ) ) ⟶ N ∞ ( Fun A ∞ un ( A , D ) ) , (137) λ A A B ∣ { id A } × N ∞ ( Fun A ∞ u n ( A , B ) ) = id N ∞ ( Fun A ∞ u n ( A , B ) ) : \lambda_{\mathcal{A}\mathcal{A}\mathcal{B}}\big|_{\{\operatorname{id}_{\mathcal{A}}\}\times\mathcal{N}_{\infty}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))}=\operatorname{id}_{\mathcal{N}_{\infty}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))}: λ AA B { id A } × N ∞ ( Fun A ∞ un ( A , B )) = id N ∞ ( Fun A ∞ un ( A , B )) : ∗ × N ∞ ( Fun A ∞ u n ( A , B ) ) ≅ N ∞ ( Fun A ∞ u n ( A , B ) ) ⟶ N ∞ ( Fun A ∞ u n ( A , B ) ) , (138) *\times\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\cong\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\longrightarrow\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr), \tag*{(138)} ∗ × N ∞ ( Fun A ∞ un ( A , B ) ) ≅ N ∞ ( Fun A ∞ un ( A , B ) ) ⟶ N ∞ ( Fun A ∞ un ( A , B ) ) , (138) λ A B B ∣ N ∞ ( Fun A ∞ u n ( A , B ) ) × { id B } = id N ∞ ( Fun A ∞ u n ( A , B ) ) : \lambda_{\mathcal{A}\mathcal{B}\mathcal{B}}\big|_{\mathcal{N}_{\infty}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))\times\{\operatorname{id}_{\mathcal{B}}\}}=\operatorname{id}_{\mathcal{N}_{\infty}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))}: λ A BB N ∞ ( Fun A ∞ un ( A , B )) × { id B } = id N ∞ ( Fun A ∞ un ( A , B )) : N ∞ ( Fun A ∞ u n ( A , B ) ) × ∗ ≅ N ∞ ( Fun A ∞ u n ( A , B ) ) ⟶ N ∞ ( Fun A ∞ u n ( A , B ) ) . (139) \mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\times*\cong\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr)\longrightarrow\mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\bigr). \tag*{(139)} N ∞ ( Fun A ∞ un ( A , B ) ) × ∗ ≅ N ∞ ( Fun A ∞ un ( A , B ) ) ⟶ N ∞ ( Fun A ∞ un ( A , B ) ) . (139) Here for (138) we restrict to the 0 0 0 -simplex id A \operatorname{id}_{\mathcal{A}} id A in N ∞ ( Fun A ∞ u n ( A , A ) ) \mathcal{N}_{\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{A})\bigr) N ∞ ( Fun A ∞ un ( A , A ) ) , regarded as a point ∗ * ∗ , and similarly for (139) . Note that (4.51)–(4.53) hold not just up to homotopy, but strictly as maps of simplicial sets.
Recall as in [55, Def. 1.2.39], [60, §1.1.4] that a simplicial category S \mathcal{S} S is a category enriched in simplicial sets. That is, for objects X , Y X,Y X , Y in S \mathcal{S} S , Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) is a simplicial set. Composition maps λ X Y Z : Hom S ( X , Y ) × Hom S ( Y , Z ) ⟶ Hom S ( X , Z ) \lambda_{XYZ}:\operatorname{Hom}_{\mathcal{S}}(X,Y)\times\operatorname{Hom}_{\mathcal{S}}(Y,Z)\longrightarrow\operatorname{Hom}_{\mathcal{S}}(X,Z) λ X Y Z : Hom S ( X , Y ) × Hom S ( Y , Z ) ⟶ Hom S ( X , Z ) are maps of simplicial sets. The identity id X \operatorname{id}_{X} id X is a 0 0 0 -simplex in Hom S ( X , X ) \operatorname{Hom}_{\mathcal{S}}(X,X) Hom S ( X , X ) . Associativity and identity properties hold strictly in maps of simplicial sets. Write S C a t \mathbf{SCat} SCat for the category of simplicial categories.
Proposition 4.71 implies that we can make unital A ∞ A_{\infty} A ∞ -categories over R R R into a simplicial category S \mathcal{S} S in two different ways:
We can take Hom S ( A , B ) = N A ∞ ( Fun A ∞ u n ( A , B ) ) K a n \operatorname{Hom}_{\mathcal{S}}(\mathcal{A},\mathcal{B})=\mathcal{N}_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B}))^{\mathrm{Kan}} Hom S ( A , B ) = N A ∞ ( Fun A ∞ un ( A , B ) ) Kan , with composition maps (4.50) and identities id A ∈ N A ∞ ( Fun A ∞ u n ( A , A ) ) K a n \operatorname{id}_{\mathcal{A}}\in\mathcal{N}_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{A}))^{\mathrm{Kan}} id A ∈ N A ∞ ( Fun A ∞ un ( A , A ) ) Kan . We will write A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan for this simplicial category.
We can take Hom S ( A , B ) = N A ∞ ( Fun A ∞ u n ( A , B ) ) \operatorname{Hom}_{\mathcal{S}}(\mathcal{A},\mathcal{B})=\mathcal{N}_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})) Hom S ( A , B ) = N A ∞ ( Fun A ∞ un ( A , B )) , with composition maps (4.49) and identities id A ∈ N A ∞ ( Fun A ∞ u n ( A , A ) ) \operatorname{id}_{\mathcal{A}}\in\mathcal{N}_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{A})) id A ∈ N A ∞ ( Fun A ∞ un ( A , A )) . We will write A ∞ S C a t R u n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un}}_{R} A ∞ SCat R un for this simplicial category, so A ∞ S C a t R u n , K a n ⊂ A ∞ S C a t R u n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R}\subset\mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un}}_{R} A ∞ SCat R un , Kan ⊂ A ∞ SCat R un .
Considered as higher categories, 2-morphisms in A ∞ S C a t R u n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un}}_{R} A ∞ SCat R un are A ∞ A_{\infty} A ∞ -natural transformations as in §4.4, but 2-morphisms in A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan are A ∞ A_{\infty} A ∞ -natural isomorphisms. We discuss (i),(ii) in the next two remarks.
Remark 4.72. (a) As in Faonte [24, §3.1], we should regard a simplicial category S \mathcal{S} S with all Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) Kan complexes, such as A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan , as a kind of ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category. Here n n n -morphisms in S \mathcal{S} S for n ≥ 1 n\ge1 n ≥ 1 are ( n − 1 ) (n-1) ( n − 1 ) -morphisms in Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) , so taking Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) a Kan complex means that n n n -morphisms in S \mathcal{S} S are invertible for n ≥ 2 n\ge2 n ≥ 2 , as we want in an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category.
(b) Bergner [3] defined a model structure on S C a t \mathbf{SCat} SCat . There is a Quillen equivalence between S C a t \mathbf{SCat} SCat with the Bergner model structure and S S e t \mathbf{SSet} SSet with the Joyal model structure. This induces the first in a chain of equivalences of ∞ \infty ∞ -categories, which give different models for the ∞ \infty ∞ -category of ∞ \infty ∞ -categories:
S C a t [ W B e r g − 1 ] ∞ ≃ S S e t [ W J o y − 1 ] ∞ ≃ C a t ∞ 1 [ J o y − 1 ] ∞ ≃ C a t ∞ . \mathbf{SCat}[W_{\mathrm{Berg}}^{-1}]^{\infty}\simeq\mathbf{SSet}[W_{\mathrm{Joy}}^{-1}]^{\infty}\simeq\mathbf{Cat}_{\infty}^{1}[\mathrm{Joy}^{-1}]^{\infty}\simeq\mathcal{C}\mathrm{at}_{\infty}. SCat [ W Berg − 1 ] ∞ ≃ SSet [ W Joy − 1 ] ∞ ≃ Cat ∞ 1 [ Joy − 1 ] ∞ ≃ C at ∞ .
The fibrant objects in the Bergner model structure are simplicial categories whose Hom spaces are Kan complexes, such as A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan .
(c) Faonte works with R R R a field K K K of characteristic zero. When C , D \mathcal{C},\mathcal{D} C , D are dg-categories, he claims [24, §4] that the Hom spaces N A ∞ ( Fun A ∞ u n ( C , D ) ) K a n \mathcal{N}_{A_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C},\mathcal{D}))^{\mathrm{Kan}} N A ∞ ( Fun A ∞ un ( C , D ) ) Kan in A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan are homotopic to those in D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]^{\infty} DGCat R [ QE dg − 1 ] ∞ . In fact this may be false if C \mathcal{C} C has zero objects, because of the error in [24, Th. 1.2] we noted before Theorem 4.66. He then claims without proof [24, §4] that the full simplicial subcategory D G S C a t R K a n ⊂ A ∞ S C a t R u n , K a n \mathbf{DGSCat}^{\mathrm{Kan}}_{R}\subset\mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} DGSCat R Kan ⊂ A ∞ SCat R un , Kan with objects dg-categories should be equivalent as an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category to D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]^{\infty} DGCat R [ QE dg − 1 ] ∞ .
(d) The following observation may be new. As in §1.6.2, because of our Theorems 9.1(a), 9.4(d), 9.7 below, the author expects that Faonte’s construction of A ∞ S C a t R u n , K a n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un},\mathrm{Kan}}_{R} A ∞ SCat R un , Kan works for general rings R R R , and gives an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category which is a model for A ∞ C a t R u n [ H E − 1 ] ∞ \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}[\mathcal{HE}^{-1}]^{\infty} A ∞ Cat R un [ HE − 1 ] ∞ , not A ∞ C a t R u n [ Q E − 1 ] ∞ \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}[\mathcal{QE}^{-1}]^{\infty} A ∞ Cat R un [ QE − 1 ] ∞ or D G C a t R [ Q E d g − 1 ] ∞ \mathbf{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]^{\infty} DGCat R [ QE dg − 1 ] ∞ , and should be equivalent as an ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category to our A ∞ C a t R u n , ∞ \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ .
Remark 4.73. (a) As in Faonte [24, §3.4], we should regard a simplicial category S \mathcal{S} S with all Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) ∞ \infty ∞ -categories, such as A ∞ S C a t R u n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un}}_{R} A ∞ SCat R un , as a kind of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category. Here n n n -morphisms in S \mathcal{S} S for n ≥ 1 n\ge1 n ≥ 1 are ( n − 1 ) (n-1) ( n − 1 ) -morphisms in Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) , so taking Hom S ( X , Y ) \operatorname{Hom}_{\mathcal{S}}(X,Y) Hom S ( X , Y ) an ∞ \infty ∞ -category means that n n n -morphisms in S \mathcal{S} S are invertible for n ≥ 3 n\ge3 n ≥ 3 , as we want in an ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category. We should think of 2-morphisms in A ∞ S C a t R u n \mathbf{A}_{\infty}\mathbf{SCat}^{\mathrm{un}}_{R} A ∞ SCat R un as A ∞ A_{\infty} A ∞ -natural transformations, which need not be invertible unless they are A ∞ A_{\infty} A ∞ -natural isomorphisms. (b) In [24, Def. 3.20], Faonte applies a construction to convert A ∞ S C a t R u n A_{\infty}\mathrm{SCat}_{R}^{\mathrm{un}} A ∞ SCat R un into a ‘fibrant Segal category in the model category (SSet, Joy)’, which is a widely accepted notion of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category.
(c) Any ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category has an associated ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category, in which one restricts to only invertible 2-morphisms. The ( ∞ , 1 ) (\infty,1) ( ∞ , 1 ) -category associated to A ∞ S C a t R u n A_{\infty}\mathrm{SCat}_{R}^{\mathrm{un}} A ∞ SCat R un is A ∞ S C a t R u n , K a n A_{\infty}\mathrm{SCat}_{R}^{\mathrm{un},\mathrm{Kan}} A ∞ SCat R un , Kan .
(d) In Theorem 7.17 we will construct an ∞ \infty ∞ -bicategory ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) of unital A ∞ A_{\infty} A ∞ -categories, in the sense of § 2.8 . As in Remark 7.18(d), we expect this to be closely connected to Faonte’s ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category.
Filtered A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories
As we explain in §5.1, curved A ∞ A_{\infty} A ∞ -algebras and curved A ∞ A_{\infty} A ∞ -categories are defined as for A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories in §3.2 and § 4.1 , but allowing operations m k m_k m k , m x 0 , … , x k m_{x_0,\ldots,x_k} m x 0 , … , x k for k ≥ 0 k \ge0 k ≥ 0 rather than k ≥ 1 k \ge1 k ≥ 1 . To define morphisms, composition, homotopies, … \ldots … , equations such as (66) , (69) , (71) , (94) , … \ldots … , become infinite sums. So the theory must be set in a context in which convergence of infinite sums makes sense. This leads to filtered A ∞ A_{\infty} A ∞ -algebras and filtered A ∞ A_{\infty} A ∞ -categories, written using complete filtered modules with a notion of convergence.
There are three main notions of filtered A ∞ A_{\infty} A ∞ -algebra and filtered A ∞ A_{\infty} A ∞ -category in the literature, due to Fukaya et al. [28–31, 33], Positselski [78] and De Deken and Lowen [18], reviewed in §5.1. We will give definitions of filtered A ∞ A_{\infty} A ∞ -algebras and filtered A ∞ A_{\infty} A ∞ -categories which are new, but generalize all three of these.
Literature review on curved and filtered A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories
Definition 5.1. Let R R R be a commutative ring. A curved A ∞ A_{\infty} A ∞ -algebra, or weak A ∞ A_{\infty} A ∞ -algebra, ( A , m ) (A,m) ( A , m ) over R R R consists of:
(a) A Z \mathbb{Z} Z -graded R R R -module A = ⨁ d ∈ Z A d A=\bigoplus_{d\in\mathbb{Z}} A^d A = ⨁ d ∈ Z A d ; and
(b) Graded R R R -linear maps m k : A ⊗ k → A m_k:A^{\otimes k}\to A m k : A ⊗ k → A of degree 2 − k 2-k 2 − k for k ≥ 0 k\ge0 k ≥ 0 . When k = 0 k=0 k = 0 we take m 0 m_0 m 0 to be an R R R -linear map m 0 : R → A 2 m_0:R\to A^2 m 0 : R → A 2 . Write m = ( m k ) k ≥ 0 m=(m_k)_{k\ge0} m = ( m k ) k ≥ 0 . These must satisfy the equation in maps A ⊗ n → A A^{\otimes n}\to A A ⊗ n → A for all n ≥ 0 n\ge0 n ≥ 0
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (140) \sum_{\substack{i,j,k\ge0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}\circ\left(\operatorname{id}_{A}^{\otimes i}\otimes m_j\otimes\operatorname{id}_{A}^{\otimes k}\right)=0. \tag*{(140)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (140) This is the same as an A ∞ A_{\infty} A ∞ -algebra in Definition 3.2, except that we have maps m k m_k m k for k ≥ 0 k\ge0 k ≥ 0 rather than k ≥ 1 k\ge1 k ≥ 1 , and in (140) we allow j ≥ 0 j\ge0 j ≥ 0 rather than j ≥ 1 j\ge1 j ≥ 1 in (62) . We say that ( A , m ) (A,m) ( A , m ) is strict if m 0 = 0 m_0=0 m 0 = 0 . Then ( A , m ) (A,m) ( A , m ) is an ordinary A ∞ A_{\infty} A ∞ -algebra. We define units 1 ∈ A 0 1\in A^0 1 ∈ A 0 for curved A ∞ A_{\infty} A ∞ -algebras ( A , m ) (A,m) ( A , m ) as for A ∞ A_{\infty} A ∞ -algebras in (65) . We call a unital curved A ∞ A_{\infty} A ∞ -algebra ( A , m ) (A,m) ( A , m ) a curved dg-algebra or weak dg-algebra if m k = 0 m_k=0 m k = 0 for k ≥ 3 k\ge3 k ≥ 3 .
Definition 5.2. Let R R R be a commutative ring. A curved A ∞ A_{\infty} A ∞ -category, or weak A ∞ A_{\infty} A ∞ -category, A \mathcal{A} A over R R R consists of:
(a) A set, or proper class, Obj A \operatorname{Obj}\mathcal{A} Obj A of objects of A \mathcal{A} A .
(b) For all objects x , y ∈ Obj A x,y \in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , a Z \mathbb{Z} Z -graded R R R -module
Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) . \operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y). Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) . We also write Hom A ( x , y ) \operatorname{Hom}_{\mathcal{A}}(x,y) Hom A ( x , y ) when we want to identify the A ∞ A_{\infty} A ∞ -category A \mathcal{A} A .
(c) For all k ≥ 0 k\geq0 k ≥ 0 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) m_k^{x_0,\ldots,x_k}:\operatorname{Hom}(x_0,x_1)\otimes_R\operatorname{Hom}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}(x_{k-1},x_k)
\longrightarrow\operatorname{Hom}(x_0,x_k) m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) of degree 2 − k 2-k 2 − k . For all n ≥ 0 n\geq0 n ≥ 0 and objects x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n these must satisfy
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (141) \sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}\circ\left(\operatorname{id}_{\operatorname{Hom}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{i-1},x_i)}\otimes m_j^{x_i,\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{n-1},x_n)}\right)=0. \tag*{(141)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (141) This is the same as an A ∞ A_{\infty} A ∞ -category in Definition 4.1, except that we have maps m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k for k ≥ 0 k\geq0 k ≥ 0 rather than k ≥ 1 k\geq1 k ≥ 1 , and in (5.2) we allow j ≥ 0 j\geq0 j ≥ 0 rather than j ≥ 1 j\geq1 j ≥ 1 in (4.2). We define unital curved A ∞ A_{\infty} A ∞ -categories and identities 1 x 1_x 1 x as in Definition 4.5. We call a unital curved A ∞ A_{\infty} A ∞ -category A \mathcal{A} A a curved dg-category or weak dg-category if m k x 0 , … , x k = 0 m_k^{x_0,\ldots,x_k}=0 m k x 0 , … , x k = 0 for all k ≥ 3 k\geq3 k ≥ 3 and x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k .
Remark 5.3. (a) Curved A ∞ A_{\infty} A ∞ -algebras/dg-algebras/A ∞ A_{\infty} A ∞ -categories/dg-categories occur naturally in many places in Mathematics and Physics. For example, curved A ∞ A_{\infty} A ∞ -algebras appear in the study of partial differential operators in Calderbank–Diemer [10, §6], of Poisson geometry in Cattaneo–Felder [14, §3.1], of differential forms on loop spaces in Getzler–Jones–Petrack [36, 37], of algebraic Chern–Weil theory in Lazarev–Schedler [56] and Nikolov–Zahariev [71], of Koszul algebras in Floystad [25] and Positselski [77], and of derived categories in Nicolás [70] and Positselski [78], and in deformation theory in de Kleijn–Wierstra [19]. Curved A ∞ A_{\infty} A ∞ -categories appear in the Lagrangian Floer theory of Fukaya–Oh–Ohta–Ono [33].
Curved dg-algebras and dg-categories appear in the study of noncommutative geometry in Block–Daenzer [6], Kajiura [48], Schwarz [82,83], and Tang [89], and of Landau–Ginzburg models in Caldararu–Tu [9] and Kapustin–Li [49, 50].
(b) If ( A , m ) (A,m) ( A , m ) is an A ∞ A_{\infty} A ∞ -algebra then (3.4) for k = 1 k=1 k = 1 gives m 1 ∘ m 1 = 0 m_1\circ m_1=0 m 1 ∘ m 1 = 0 , so ( A , m 1 ) (A,m_1) ( A , m 1 ) is a complex and we can form its cohomology H ∗ ( A ) H^*(A) H ∗ ( A ) as in (3.5).
However, if ( A , m ) (A,m) ( A , m ) is only a curved A ∞ A_{\infty} A ∞ -algebra then (5.1) for k = 1 k=1 k = 1 yields
m 1 ∘ m 1 = − m 2 ∘ ( m 0 ( 1 ) ⊗ id A ) + m 2 ∘ ( id A ⊗ m 0 ( 1 ) ) . m_1\circ m_1=-m_2\circ\left(m_0(1)\otimes\operatorname{id}_A\right)+m_2\circ\left(\operatorname{id}_A\otimes m_0(1)\right). m 1 ∘ m 1 = − m 2 ∘ ( m 0 ( 1 ) ⊗ id A ) + m 2 ∘ ( id A ⊗ m 0 ( 1 ) ) .
So we may no longer have m 1 ∘ m 1 = 0 m_1\circ m_1=0 m 1 ∘ m 1 = 0 , and we cannot form the cohomology H ∗ ( A ) H^*(A) H ∗ ( A ) . We regard m 0 m_0 m 0 as the obstruction to ( A , m 1 ) (A,m_1) ( A , m 1 ) being a complex.
(c) Part (b) means that definitions involving H ∗ ( A , m 1 ) H^*(A,m_1) H ∗ ( A , m 1 ) , such as cohomological units and quasi-isomorphisms, do not make sense for curved A ∞ A_\infty A ∞ -algebras and curved A ∞ A_\infty A ∞ -categories.
(d) The obvious way to define morphisms of curved A ∞ A_\infty A ∞ -algebras is to follow Definition 3.8, but defining f k f_k f k for k ≥ 0 k \ge0 k ≥ 0 rather than k ≥ 1 k \ge1 k ≥ 1 . However, we immediately run into a problem: equation (66) then involves an infinite sum, as on the right hand side we should sum over decompositions n = n 1 + ⋯ + n k n=n_1+\cdots+n_k n = n 1 + ⋯ + n k with k ≥ 0 k \ge0 k ≥ 0 and n i ≥ 0 n_i \ge0 n i ≥ 0 , rather than n i ≥ 1 n_i \ge1 n i ≥ 1 . The same applies to compositions in (69) and homotopies in (71) . Because of this, it is natural to study curved A ∞ A_\infty A ∞ -algebras and curved A ∞ A_\infty A ∞ -categories in settings in which we have notions of convergence of infinite sums . Parts (e)–(g) describe three cases in the literature.
(e) There is a massive project by Fukaya–Oh–Ohta–Ono over 2000–2025, including [28–31, 33], which constructs Lagrangian Floer cohomology and Fukaya categories for symplectic manifolds ( X , ω ) (X,\omega) ( X , ω ) . They define notions of filtered A ∞ A_\infty A ∞ -algebra [33, §3.2] and filtered A ∞ A_\infty A ∞ -category [29, §1], [31, §2], where [33] associates a filtered A ∞ A_\infty A ∞ -algebra to each Lagrangian L L L in ( X , ω ) (X,\omega) ( X , ω ) . These involve complete filtered modules over filtered Novikov rings such as Λ K n o v \Lambda^{\mathrm{nov}}_K Λ K nov and Λ K 0 , n o v \Lambda^{0,\mathrm{nov}}_K Λ K 0 , nov in Example 5.9 below, so convergence of infinite sums in such modules makes sense.
(f) Positselski [78] studies curved A ∞ A_\infty A ∞ -algebras ( A , m ) (A,\mathfrak m) ( A , m ) which are complete modules over a pro-Artinian local ring R R R with (maximal) ideal i ⊂ R \mathfrak i \subset R i ⊂ R , such that m 0 ( 1 ) ∈ i A 2 ⊂ A 2 \mathfrak m_0(1) \in\mathfrak i A^2 \subset A^2 m 0 ( 1 ) ∈ i A 2 ⊂ A 2 . An infinite sum ∑ i ∈ I a i \sum_{i\in I} a_i ∑ i ∈ I a i in A d A^{d} A d is convergent if its projection to A d / i N A d A^{d}/\mathfrak i^N A^{d} A d / i N A d has finitely many nonzero terms for all N ≥ 1 N \ge1 N ≥ 1 . Filtered A ∞ A_\infty A ∞ -algebras over Λ K 0 , n o v \Lambda^{0,\mathrm{nov}}_K Λ K 0 , nov in Fukaya et al. [33] fit into Positselski’s framework, but filtered A ∞ A_\infty A ∞ -algebras over Λ K n o v \Lambda^{\mathrm{nov}}_K Λ K nov do not.
(g) De Deken and Lowen [18] define a notion of filtered A ∞ A_\infty A ∞ -category (which they call filtered cA ∞ A_\infty A ∞ -categories , where the ‘c’ stands for ‘curved’), intended for applications in noncommutative geometry. It is different to that in Fukaya [29, §1], [31, §2]. They fix a totally ordered monoid ( Γ , ≤ ) (\Gamma,\le) ( Γ , ≤ ) and a complete Γ \Gamma Γ -filtered commutative ring R R R , and their filtered A ∞ A_\infty A ∞ -categories involve complete Γ \Gamma Γ -filtered R R R -modules. Convergence of infinite sums in such modules makes sense. They require that 0 ∈ Γ 0\in\Gamma 0 ∈ Γ is the minimal element in the order Γ \Gamma Γ . Because of this, filtered A ∞ A_\infty A ∞ -algebras over Λ K 0 , n o v \Lambda^{0,\mathrm{nov}}_K Λ K 0 , nov in Fukaya et al. [33] fit into De Deken–Lowen’s framework with ( Γ , ≤ ) = ( [ 0 , ∞ ) , ≤ ) (\Gamma,\le)=([0,\infty),\le) ( Γ , ≤ ) = ([ 0 , ∞ ) , ≤ ) , but filtered A ∞ A_\infty A ∞ -algebras over K K K with ( Γ , ≤ ) = ( R , ≤ ) (\Gamma,\le)=(\mathbb R,\le) ( Γ , ≤ ) = ( R , ≤ ) do not.
(h) My notion of filtered A ∞ A_\infty A ∞ -algebra and A ∞ A_\infty A ∞ -category in §5.3–§5.4 are new, but include (e)–(g) as special cases. Although my primary motivation comes from Symplectic Geometry, I hope this work will be useful in other areas too, as in (a). As a sample application, in §5.8 I will write the ‘twist construction’ for A ∞ A_\infty A ∞ -categories from §4.9 in terms of filtered A ∞ A_\infty A ∞ -categories.
(i) Here are some references on filtered A ∞ A_\infty A ∞ -theory which are not very relevant to our project. Biran–Cornea–Shelukhin [5, §2] discuss ‘weakly filtered’ A ∞ A_\infty A ∞ -categories and A ∞ A_\infty A ∞ -functors. Fukaya [32, §4] defines ‘filtered A ∞ A_\infty A ∞ -functors with energy loss’. In both [5, 32], the A ∞ A_\infty A ∞ -categories and A ∞ A_\infty A ∞ -functors are ordinary (not curved) A ∞ A_\infty A ∞ -categories A \mathcal A A and A ∞ A_\infty A ∞ -functors F \mathcal F F , with m 0 A = 0 \mathfrak m_0^{\mathcal A}=0 m 0 A = 0 and f 0 = 0 \mathfrak f_0=0 f 0 = 0 .
They are equipped with R \mathbb{R} R -filtrations, but the A ∞ A_{\infty} A ∞ -maps in A \mathcal{A} A and F ~ \widetilde{\mathcal{F}} F need not preserve filtrations, they may decrease filtrations by the ‘energy cost’ or ‘energy loss’, a measure of the ‘distance’ between two A ∞ A_{\infty} A ∞ -categories.
Complete filtered rings and modules
Definition 5.4. Let Γ \Gamma Γ be a commutative monoid, with identity 0 ∈ Γ 0 \in\Gamma 0 ∈ Γ and monoid operation + + + , and ⪯ \preceq ⪯ be a partial order on Γ \Gamma Γ . We will use the following notation: we write γ ≺ δ \gamma\prec\delta γ ≺ δ to mean γ ⪯ δ \gamma\preceq\delta γ ⪯ δ and γ ≠ δ \gamma\ne\delta γ = δ , and we write γ ⪰ δ \gamma\succeq\delta γ ⪰ δ and γ ≻ δ \gamma\succ\delta γ ≻ δ to mean δ ⪯ γ \delta\preceq\gamma δ ⪯ γ and δ ≺ γ \delta\prec\gamma δ ≺ γ respectively. For γ ∈ Γ \gamma\in\Gamma γ ∈ Γ and n ∈ N n \in\mathbb{N} n ∈ N we write n γ = γ + ⋯ + γ ⏟ n copies n\gamma= \underbrace{\gamma+\cdots+\gamma}_{n\text{ copies}} nγ = n copies γ + ⋯ + γ , with 0 γ = 0 0\gamma=0 0 γ = 0 . If γ ∈ Γ \gamma\in\Gamma γ ∈ Γ with γ ≻ 0 \gamma\succ0 γ ≻ 0 , we write γ ≫ 0 \gamma\gg0 γ ≫ 0 if for all δ ∈ Γ \delta\in\Gamma δ ∈ Γ there exists n ≫ 0 n \gg0 n ≫ 0 in N \mathbb{N} N with n γ ⪰ δ n\gamma\succeq\delta nγ ⪰ δ .
We call ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) a partially-ordered monoid , or po-monoid , if:
(a) If α ⪯ β \alpha\preceq\beta α ⪯ β and γ ⪯ δ \gamma\preceq\delta γ ⪯ δ in Γ \Gamma Γ then α + γ ⪯ β + δ \alpha+\gamma\preceq\beta+\delta α + γ ⪯ β + δ .
(b) If γ , δ ∈ Γ \gamma,\delta\in\Gamma γ , δ ∈ Γ there is a unique least upper bound lub ( γ , δ ) \operatorname{lub}(\gamma,\delta) lub ( γ , δ ) and a unique greatest lower bound glb ( γ , δ ) \operatorname{glb}(\gamma,\delta) glb ( γ , δ ) , such that γ ⪯ lub ( γ , δ ) \gamma\preceq\operatorname{lub}(\gamma,\delta) γ ⪯ lub ( γ , δ ) , δ ⪯ lub ( γ , δ ) \delta\preceq\operatorname{lub}(\gamma,\delta) δ ⪯ lub ( γ , δ ) , γ ⪰ glb ( γ , δ ) \gamma\succeq\operatorname{glb}(\gamma,\delta) γ ⪰ glb ( γ , δ ) , δ ⪰ glb ( γ , δ ) \delta\succeq\operatorname{glb}(\gamma,\delta) δ ⪰ glb ( γ , δ ) , and if ϵ ∈ Γ \epsilon\in\Gamma ϵ ∈ Γ with γ ⪯ ϵ \gamma\preceq\epsilon γ ⪯ ϵ , δ ⪯ ϵ \delta\preceq\epsilon δ ⪯ ϵ then lub ( γ , δ ) ⪯ ϵ \operatorname{lub}(\gamma,\delta) \preceq\epsilon lub ( γ , δ ) ⪯ ϵ , and if ζ ∈ Γ \zeta\in\Gamma ζ ∈ Γ with γ ⪰ ζ \gamma\succeq\zeta γ ⪰ ζ , δ ⪰ ζ \delta\succeq\zeta δ ⪰ ζ then glb ( γ , δ ) ⪰ ζ \operatorname{glb}(\gamma,\delta) \succeq\zeta glb ( γ , δ ) ⪰ ζ .
By induction on the size of the set, least upper bounds and greatest lower bounds also exist for all nonempty finite subsets of Γ \Gamma Γ .
(c) There exists γ ∈ Γ \gamma\in\Gamma γ ∈ Γ such that γ ≫ 0 \gamma\gg0 γ ≫ 0 .
(d) If γ ≫ 0 \gamma\gg0 γ ≫ 0 and δ ≫ 0 \delta\gg0 δ ≫ 0 then glb ( γ , δ ) ≫ 0 \operatorname{glb}(\gamma,\delta) \gg0 glb ( γ , δ ) ≫ 0 .
If ( γ n ) n = 0 ∞ (\gamma_n)_{n=0}^{\infty} ( γ n ) n = 0 ∞ is a sequence in Γ \Gamma Γ , we say that γ n → ∞ \gamma_n \to\infty γ n → ∞ in Γ \Gamma Γ as n → ∞ n \to\infty n → ∞ if for all δ ∈ Γ \delta\in\Gamma δ ∈ Γ there exists N ≥ 0 N \geq0 N ≥ 0 such that γ n ⪰ δ \gamma_n \succeq\delta γ n ⪰ δ for all n ≥ N n \geq N n ≥ N . Then γ ≫ 0 \gamma\gg0 γ ≫ 0 if and only if n γ → ∞ n\gamma\to\infty nγ → ∞ as n → ∞ n \to\infty n → ∞ .
Example 5.5. (a) The trivial example is Γ = { 0 , ∞ } \Gamma=\{0,\infty\} Γ = { 0 , ∞ } with 0 + 0 = 0 0+0=0 0 + 0 = 0 , 0 + ∞ = ∞ 0+\infty=\infty 0 + ∞ = ∞ , ∞ + ∞ = ∞ \infty+\infty=\infty ∞ + ∞ = ∞ and 0 ≺ ∞ 0\prec\infty 0 ≺ ∞ .
(b) We can take Γ = R , Z , Q , [ 0 , ∞ ) \Gamma=\mathbb{R},\mathbb{Z},\mathbb{Q},[0,\infty) Γ = R , Z , Q , [ 0 , ∞ ) or N \mathbb{N} N with the usual total order ≤ \leq ≤ . Then lub ( γ , δ ) = max ( γ , δ ) \operatorname{lub}(\gamma,\delta)=\max(\gamma,\delta) lub ( γ , δ ) = max ( γ , δ ) , and glb ( γ , δ ) = min ( γ , δ ) \operatorname{glb}(\gamma,\delta)=\min(\gamma,\delta) glb ( γ , δ ) = min ( γ , δ ) , and γ ≫ 0 \gamma\gg0 γ ≫ 0 if and only if γ > 0 \gamma>0 γ > 0 .
(c) Let ( Γ 1 , ⪯ 1 ) (\Gamma_1,\preceq_1) ( Γ 1 , ⪯ 1 ) and ( Γ 2 , ⪯ 2 ) (\Gamma_2,\preceq_2) ( Γ 2 , ⪯ 2 ) be po-monoids. Define a partial order ⪯ 12 \preceq_{12} ⪯ 12 on the product monoid Γ 1 × Γ 2 \Gamma_1\times\Gamma_2 Γ 1 × Γ 2 by ( γ 1 , γ 2 ) ⪯ 12 ( δ 1 , δ 2 ) (\gamma_1,\gamma_2)\preceq_{12}(\delta_1,\delta_2) ( γ 1 , γ 2 ) ⪯ 12 ( δ 1 , δ 2 ) if γ 1 ⪯ 1 δ 1 \gamma_1\preceq_1\delta_1 γ 1 ⪯ 1 δ 1 and γ 2 ⪯ 2 δ 2 \gamma_2\preceq_2\delta_2 γ 2 ⪯ 2 δ 2 . It is easy to show that ( Γ 1 × Γ 2 , ⪯ 12 ) (\Gamma_1\times\Gamma_2,\preceq_{12}) ( Γ 1 × Γ 2 , ⪯ 12 ) is a po-monoid.
Definition 5.6. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid. A Γ \Gamma Γ -filtered ring R R R is a commutative ring R R R equipped with a decreasing filtration ( F γ R ) γ ∈ Γ (F^\gamma R)_{\gamma\in\Gamma} ( F γ R ) γ ∈ Γ satisfying:
(i) F γ R F^\gamma R F γ R is an additive subgroup of R R R , with F δ R ⊆ F γ R F^\delta R\subseteq F^\gamma R F δ R ⊆ F γ R if γ ⪯ δ \gamma\preceq\delta γ ⪯ δ in Γ \Gamma Γ .
(ii) 1 ∈ F 0 R 1\in F^0R 1 ∈ F 0 R .
(iii) F γ R ⋅ F δ R ⊆ F γ + δ R F^\gamma R\cdot F^\delta R\subseteq F^{\gamma+\delta}R F γ R ⋅ F δ R ⊆ F γ + δ R for all γ , δ ∈ Γ \gamma,\delta\in\Gamma γ , δ ∈ Γ .
(iv) ⋃ γ ∈ Γ F γ R = R \bigcup_{\gamma\in\Gamma}F^\gamma R=R ⋃ γ ∈ Γ F γ R = R and ⋂ γ ∈ Γ F γ R = { 0 } \bigcap_{\gamma\in\Gamma}F^\gamma R=\{0\} ⋂ γ ∈ Γ F γ R = { 0 } .
If γ ∈ Γ \gamma\in\Gamma γ ∈ Γ , we write F ≻ γ R = ⋃ δ ∈ Γ : δ ≻ 0 F γ + δ R F^{\succ\gamma}R = \bigcup_{\delta\in\Gamma:\delta\succ0} F^{\gamma+\delta}R F ≻ γ R = ⋃ δ ∈ Γ : δ ≻ 0 F γ + δ R . Observe that F 0 R F^0R F 0 R is a subring of R R R , and F ≻ 0 R F^{\succ0}R F ≻ 0 R is an ideal in F 0 R F^0R F 0 R , so the quotient F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R is a commutative ring. For any γ ∈ Γ \gamma\in\Gamma γ ∈ Γ , F γ R / F ≻ γ R F^\gamma R/F^{\succ\gamma}R F γ R / F ≻ γ R is a module over F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R .
A Γ \Gamma Γ -filtered module, or just filtered module, A A A over R R R is an R R R -module A A A equipped with a decreasing filtration ( F γ A ) γ ∈ Γ (F^\gamma A)_{\gamma\in\Gamma} ( F γ A ) γ ∈ Γ satisfying:
(v) F γ A F^\gamma A F γ A is an additive subgroup of A A A , with F δ A ⊆ F γ A F^\delta A \subseteq F^\gamma A F δ A ⊆ F γ A if γ ⪯ δ \gamma\preceq\delta γ ⪯ δ in Γ \Gamma Γ .
(vi) F γ R ⋅ F δ A ⊆ F γ + δ A F^\gamma R \cdot F^\delta A \subseteq F^{\gamma+\delta}A F γ R ⋅ F δ A ⊆ F γ + δ A for all γ , δ ∈ Γ \gamma,\delta\in\Gamma γ , δ ∈ Γ .
(vii) ⋃ γ ∈ Γ F γ A = A \bigcup_{\gamma\in\Gamma} F^\gamma A = A ⋃ γ ∈ Γ F γ A = A and ⋂ γ ∈ Γ F γ A = { 0 } \bigcap_{\gamma\in\Gamma} F^\gamma A = \{0\} ⋂ γ ∈ Γ F γ A = { 0 } .
If γ ∈ Γ \gamma\in\Gamma γ ∈ Γ , we write F ≻ γ A = ⋃ δ ∈ Γ : δ ≻ 0 F γ + δ A F^{\succ\gamma}A = \bigcup_{\delta\in\Gamma:\delta\succ0} F^{\gamma+\delta}A F ≻ γ A = ⋃ δ ∈ Γ : δ ≻ 0 F γ + δ A . Then F ≻ γ A ⊆ F γ A F^{\succ\gamma}A \subseteq F^\gamma A F ≻ γ A ⊆ F γ A , so we can take the quotient F γ A / F ≻ γ A F^\gamma A/F^{\succ\gamma}A F γ A / F ≻ γ A . It is a module over F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R .
Let ( a n ) n = 0 ∞ (a_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ be a sequence in A A A . We say that ( a n ) n = 0 ∞ (a_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ is Cauchy if for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ there exists N ≥ 0 N \geq0 N ≥ 0 such that a m − a n ∈ F γ A a_m-a_n \in F^\gamma A a m − a n ∈ F γ A for all m , n ≥ N m,n \geq N m , n ≥ N . We say that ( a n ) n = 0 ∞ (a_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ is convergent , with limit a ∈ A a \in A a ∈ A if for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ there exists N ≥ 0 N \geq0 N ≥ 0 such that a n − a ∈ F γ A a_n-a \in F^\gamma A a n − a ∈ F γ A for all n ≥ N n \geq N n ≥ N . We write this as a n → a a_n \to a a n → a as n → ∞ n \to\infty n → ∞ . The condition ⋂ γ ∈ Γ F γ A = { 0 } \bigcap_{\gamma\in\Gamma} F^\gamma A = \{0\} ⋂ γ ∈ Γ F γ A = { 0 } implies that limits are unique if they exist.
We can interpret convergence in topological terms, by making A A A into a topological space, where the topology on A A A is generated by open sets { a + F γ A : a ∈ A , γ ∈ Γ } \{a+F^\gamma A:a\in A,\ \gamma\in\Gamma\} { a + F γ A : a ∈ A , γ ∈ Γ } . The condition ⋂ γ ∈ Γ F γ A = { 0 } \bigcap_{\gamma\in\Gamma} F^\gamma A = \{0\} ⋂ γ ∈ Γ F γ A = { 0 } implies that A A A is Hausdorff.
We say that the filtered R R R -module A A A is complete if every Cauchy sequence ( a n ) n = 0 ∞ (a_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ is convergent. We call a filtered ring R R R complete if it is complete as a module over itself. Every filtered R R R -module A A A (or filtered ring R R R ) has a natural completion A ‾ \overline{A} A (or R ‾ \overline{R} R ), where points of A ‾ \overline{A} A are ∼ \sim ∼ -equivalence classes [ ( a n ) n = 0 ∞ ] [(a_n)_{n=0}^{\infty}] [( a n ) n = 0 ∞ ] of Cauchy sequences ( a n ) n = 0 ∞ (a_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ in A A A , with equivalence relation ( a n ) n = 0 ∞ ∼ ( a n ′ ) n = 0 ∞ (a_n)_{n=0}^{\infty}\sim(a'_n)_{n=0}^{\infty} ( a n ) n = 0 ∞ ∼ ( a n ′ ) n = 0 ∞ if a n − a n ′ → 0 a_n-a'_n\to0 a n − a n ′ → 0 as n → ∞ n\to\infty n → ∞ .
We can also define convergence and limits of infinite sums in A A A . A sum ∑ i ∈ I α i \sum_{i\in I}\alpha_i ∑ i ∈ I α i with α i ∈ A \alpha_i\in A α i ∈ A is called Cauchy if for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ there are only finitely many i ∈ I i\in I i ∈ I with α i ∉ F γ A \alpha_i\notin F^\gamma A α i ∈ / F γ A . A sum ∑ i ∈ I α i \sum_{i\in I}\alpha_i ∑ i ∈ I α i in A A A is called convergent , with limit α ∈ A \alpha\in A α ∈ A , if for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ we have
α − ∑ i ∈ I : α i ∉ F γ A α i ∈ F γ A . \alpha-\sum_{i\in I:\alpha_i\notin F^\gamma A}\alpha_i\in F^\gamma A. α − i ∈ I : α i ∈ / F γ A ∑ α i ∈ F γ A .
Limits are unique if they exist. If A A A is complete then every Cauchy sum is convergent.
Suppose ( γ n ) n = 0 ∞ (\gamma_n)_{n=0}^{\infty} ( γ n ) n = 0 ∞ is a sequence in Γ \Gamma Γ with γ n → ∞ \gamma_n\to\infty γ n → ∞ as n → ∞ n\to\infty n → ∞ , as in Definition 5.4, and ∑ n = 0 ∞ α n \sum_{n=0}^{\infty}\alpha_n ∑ n = 0 ∞ α n is an infinite sum in A A A with α n ∈ F γ n A \alpha_n\in F^{\gamma_n}A α n ∈ F γ n A for all n ≥ 0 n\geq0 n ≥ 0 . Then ∑ n = 0 ∞ α n \sum_{n=0}^{\infty}\alpha_n ∑ n = 0 ∞ α n is Cauchy, and convergent if A A A is complete.
If A , B A,B A , B are filtered modules over R R R then A ⊗ R B A\otimes_R B A ⊗ R B is a filtered module, such that F γ ( A ⊗ B ) F^\gamma(A\otimes B) F γ ( A ⊗ B ) is the subgroup of elements of A ⊗ R B A\otimes_R B A ⊗ R B spanned by elements r a ⊗ b r\,a\otimes b r a ⊗ b for r ∈ F γ 1 R r\in F^{\gamma_1}R r ∈ F γ 1 R , a ∈ F γ 2 A a\in F^{\gamma_2}A a ∈ F γ 2 A and b ∈ F γ 3 B b\in F^{\gamma_3}B b ∈ F γ 3 B with γ 1 + γ 2 + γ 3 ⪰ γ \gamma_1+\gamma_2+\gamma_3\succeq\gamma γ 1 + γ 2 + γ 3 ⪰ γ . If A , B A,B A , B are complete then A ⊗ R B A\otimes_R B A ⊗ R B need not be complete, but it has a unique completion, written A ⊗ ^ R B A\widehat{\otimes}_R B A ⊗ R B .
Remark 5.7. Definitions 5.4 and 5.6 are quite similar to parts of De Deken–Lowen [18, §2.1–§2.5]. Our definition is more general, in that they consider filtrations over ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) in which ⪯ \preceq ⪯ is a total order, not a partial order, and require that 0 ∈ Γ 0 \in\Gamma 0 ∈ Γ is minimal in ⪯ \preceq ⪯ . So [18] is a special case of our theory.
Example 5.8. Take the po-monoid Γ = { 0 , ∞ } \Gamma=\{0,\infty\} Γ = { 0 , ∞ } as in Example 5.5(a). Then any commutative ring R R R has a unique Γ \Gamma Γ -filtration satisfying Definition 5.6(i)–(iv), with F 0 R = R F^{0}R=R F 0 R = R and F ∞ R = { 0 } F^{\infty}R=\{0\} F ∞ R = { 0 } . Similarly, every R R R -module A A A has a unique Γ \Gamma Γ -filtration satisfying Definition 5.6(v)–(vii), with F 0 A = A F^{0}A=A F 0 A = A and F ∞ A = { 0 } F^{\infty}A=\{0\} F ∞ A = { 0 } . We have F ≻ 0 R = F ≻ 0 A = { 0 } F^{\succ0}R=F^{\succ0}A=\{0\} F ≻ 0 R = F ≻ 0 A = { 0 } . Then R R R , A A A are complete. An infinite sum ∑ i ∈ I a i \sum_{i\in I}a_i ∑ i ∈ I a i in A A A is convergent if and only if there are only finitely many i ∈ I i\in I i ∈ I with a i ≠ 0 a_i\ne0 a i = 0 .
The next example is based on Fukaya et al. [33, §1.7].
Example 5.9. Let K \mathbb{K} K be a commutative ring. The universal Novikov ring Λ K n o v \Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K nov over K \mathbb{K} K is the formal power series ring
Λ K n o v = { ∑ i = 1 N a i q λ i : N ∈ { 0 , 1 , 2 , … , ∞ } , a i ∈ K ∖ { 0 } , λ i ∈ R , \Lambda^{\mathrm{nov}}_{\mathbb{K}}=\left\{\sum_{i=1}^{N}a_iq^{\lambda_i}:N\in\{0,1,2,\ldots,\infty\},\ a_i\in\mathbb{K}\setminus\{0\},\ \lambda_i\in\mathbb{R},\right. Λ K nov = { i = 1 ∑ N a i q λ i : N ∈ { 0 , 1 , 2 , … , ∞ } , a i ∈ K ∖ { 0 } , λ i ∈ R , λ 1 < λ 2 < ⋯ , λ i → ∞ as i → ∞ if N = ∞ } . \left.\lambda_1<\lambda_2<\cdots,\ \lambda_i\to\infty\text{ as }i\to\infty\text{ if }N=\infty\right\}. λ 1 < λ 2 < ⋯ , λ i → ∞ as i → ∞ if N = ∞ } . It is a K \mathbb{K} K -algebra under addition and multiplication of formal power series, with unit 1 q 0 1q^0 1 q 0 . It has a decreasing R \mathbb{R} R -filtration ( F γ Λ K n o v ) γ ∈ R (F^\gamma\Lambda^{\mathrm{nov}}_{\mathbb{K}})_{\gamma\in\mathbb{R}} ( F γ Λ K nov ) γ ∈ R given by
F γ Λ K n o v = { ∑ i = 1 N a i q λ i : N ∈ { 0 , 1 , 2 , … , ∞ } , a i ∈ K ∖ { 0 } , λ i ∈ R , F^\gamma\Lambda^{\mathrm{nov}}_{\mathbb{K}}=\{\sum_{i=1}^{N}a_iq^{\lambda_i}:N\in\{0,1,2,\ldots,\infty\},\ a_i\in\mathbb{K}\setminus\{0\},\ \lambda_i\in\mathbb{R}, F γ Λ K nov = { i = 1 ∑ N a i q λ i : N ∈ { 0 , 1 , 2 , … , ∞ } , a i ∈ K ∖ { 0 } , λ i ∈ R , γ ≤ λ 1 < λ 2 < ⋯ , λ i → ∞ as i → ∞ if N = ∞ } . \gamma\leq\lambda_1<\lambda_2<\cdots,\ \lambda_i\to\infty\text{ as }i\to\infty\text{ if }N=\infty\}. γ ≤ λ 1 < λ 2 < ⋯ , λ i → ∞ as i → ∞ if N = ∞ } . Then Λ K n o v \Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K nov is a complete filtered ring over ( R , ≤ ) (\mathbb{R},\leq) ( R , ≤ ) . If K \mathbb{K} K is a field then Λ K n o v \Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K nov is a field. Also Λ K 0 , n o v : = F 0 Λ K n o v \Lambda^{0,\mathrm{nov}}_{\mathbb{K}}:=F^0\Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K 0 , nov := F 0 Λ K nov is a complete filtered ring over ( [ 0 , ∞ ) , ≤ ) ([0,\infty),\leq) ([ 0 , ∞ ) , ≤ ) .
Filtered A ∞ A_\infty A ∞ -algebras over both Λ K n o v \Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K nov and Λ K 0 , n o v \Lambda^{0,\mathrm{nov}}_{\mathbb{K}} Λ K 0 , nov are used in the Lagrangian Floer cohomology theory of Fukaya–Oh–Ohta–Ono [28–31, 33]. The operations m k m_k m k in the filtered A ∞ A_\infty A ∞ -algebras are defined using sums of counts of J J J -holomorphic curves Σ \Sigma Σ weighted by q area ( Σ ) q^{\operatorname{area}(\Sigma)} q area ( Σ ) in Λ K \Lambda_{\mathbb{K}} Λ K . Although there may be infinitely many possible homology classes for Σ \Sigma Σ , for any γ ∈ R \gamma\in\mathbb{R} γ ∈ R there are only finitely many homology classes for Σ \Sigma Σ with area ( Σ ) < γ \operatorname{area}(\Sigma)<\gamma area ( Σ ) < γ . So the sum over Σ \Sigma Σ is convergent in a complete Λ K n o v \Lambda^{\mathrm{nov}}_{\mathbb{K}} Λ K nov -module.
Filtered A ∞ A_\infty A ∞ -algebras
Next we define filtered A ∞ A_\infty A ∞ -algebras .
Definition 5.10. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring, as in Definitions 5.4 and 5.6. A Γ \Gamma Γ -filtered A ∞ A_\infty A ∞ -algebra, or just filtered A ∞ A_\infty A ∞ -algebra, ( A , m ) (A,\mathfrak{m}) ( A , m ) over R R R consists of:
(a) A Z \mathbb{Z} Z -graded Γ \Gamma Γ -filtered R R R -module A = ⨁ d ∈ Z A d A=\bigoplus_{d\in\mathbb{Z}}A^d A = ⨁ d ∈ Z A d , where A d A^d A d is a complete filtered R R R -module for each d ∈ Z d\in\mathbb{Z} d ∈ Z . (We do not require A A A to be complete.) We say that A A A is degree-wise complete , meaning that A d A^d A d is complete for each degree d ∈ Z d\in\mathbb{Z} d ∈ Z .
(b) Graded R R R -linear maps m k : A ⊗ k → A m_k : A^{\otimes k} \to A m k : A ⊗ k → A of degree 2 − k 2-k 2 − k for k ≥ 0 k \ge0 k ≥ 0 . When k = 0 k=0 k = 0 we take m 0 m_0 m 0 to be an R R R -linear map m 0 : R → A 2 m_0 : R \to A^2 m 0 : R → A 2 . Sometimes we identify m 0 m_0 m 0 with m 0 ( 1 ) ∈ A 2 m_0(1) \in A^2 m 0 ( 1 ) ∈ A 2 . Write m = ( m k ) k ≥ 0 \mathbf{m}=(m_k)_{k\ge0} m = ( m k ) k ≥ 0 .
We require m k : A ⊗ k → A m_k : A^{\otimes k} \to A m k : A ⊗ k → A to be filtration-preserving for all k ≥ 1 k \ge1 k ≥ 1 , that is, m k ( F γ ( A ⊗ k ) ) ⊆ F γ A m_k(F^\gamma(A^{\otimes k})) \subseteq F^\gamma A m k ( F γ ( A ⊗ k )) ⊆ F γ A for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ . Equivalently, if a i ∈ F γ i A a_i \in F^{\gamma_i}A a i ∈ F γ i A for i = 1 , … , k i=1,\ldots,k i = 1 , … , k , γ i ∈ Γ \gamma_i \in\Gamma γ i ∈ Γ , then m k ( a 1 ⊗ ⋯ ⊗ a k ) ∈ F γ 1 + ⋯ + γ k A m_k(a_1 \otimes\cdots\otimes a_k) \in F^{\gamma_1+\cdots+\gamma_k}A m k ( a 1 ⊗ ⋯ ⊗ a k ) ∈ F γ 1 + ⋯ + γ k A .
We require m 0 : R → A 2 m_0 : R \to A^2 m 0 : R → A 2 to satisfy m 0 ( 1 ) ∈ F ≻ 0 A 2 m_0(1) \in F^{\succ0}A^2 m 0 ( 1 ) ∈ F ≻ 0 A 2 . That is, m 0 ( 1 ) ∈ F γ A 2 m_0(1) \in F^\gamma A^2 m 0 ( 1 ) ∈ F γ A 2 for some γ ∈ Γ \gamma\in\Gamma γ ∈ Γ with γ ≽ 0 \gamma\succcurlyeq0 γ ≽ 0 , so γ ≻ 0 \gamma\succ0 γ ≻ 0 and n γ → 0 n\gamma\to0 nγ → 0 in Γ \Gamma Γ as n → ∞ n \to\infty n → ∞ .
As in (140) , these must satisfy the equation in maps A ⊗ n → A A^{\otimes n} \to A A ⊗ n → A for all n ≥ 0 n \ge0 n ≥ 0
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (142) \sum_{\substack{i,j,k\ge0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}\circ\left(\operatorname{id}_A^{\otimes i}\otimes m_j\otimes\operatorname{id}_A^{\otimes k}\right)=0. \tag*{(142)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = 0. (142)
Note that a filtered A ∞ A_\infty A ∞ -algebra ( A , m ) (A,\mathbf{m}) ( A , m ) is a curved A ∞ A_\infty A ∞ -algebra in the sense of Definition 5.1 , with extra structure.
A filtered A ∞ A_\infty A ∞ -algebra ( A , m ) (A,\mathbf{m}) ( A , m ) is called strict if m 0 = 0 m_0=0 m 0 = 0 . Then ( A , m ) (A,\mathbf{m}) ( A , m ) is an ordinary A ∞ A_\infty A ∞ -algebra, forgetting the filtration structure.
A unit for ( A , m ) (A,\mathbf{m}) ( A , m ) is an element 1 ∈ F 0 A 0 \mathbb{1}\in F^0A^0 1 ∈ F 0 A 0 satisfying (65) . Then we call ( A , m ) (A,\mathbf{m}) ( A , m ) unital . Units are unique if they exist.
We will not define cohomologically unital filtered A ∞ A_\infty A ∞ -algebras.
Example 5.11. If Γ = { 0 , ∞ } \Gamma=\{0,\infty\} Γ = { 0 , ∞ } as in Example 5.5 (a), then as in Example 5.8 , Γ \Gamma Γ -filtered rings R R R and modules A A A are equivalent to ordinary rings R R R and modules A A A , all R R R -linear maps are filtration-preserving, and F ≻ 0 A = 0 F^{\succ0}A=0 F ≻ 0 A = 0 . Thus in Definition 5.10 we have m 0 = 0 m_0=0 m 0 = 0 , and Γ \Gamma Γ -filtered A ∞ A_\infty A ∞ -algebras are equivalent to A ∞ A_\infty A ∞ -algebras.
Remark 5.12. (a) When Γ = ( R , ≤ ) \Gamma=(\mathbb{R},\le) Γ = ( R , ≤ ) and R = Λ K n o v R=\Lambda_K^{\mathrm{nov}} R = Λ K nov in Example 5.9 , or Γ = ( [ 0 , ∞ ) , ≤ ) \Gamma=([0,\infty),\le) Γ = ([ 0 , ∞ ) , ≤ ) and R = Λ K , 0 n o v R=\Lambda_{K,0}^{\mathrm{nov}} R = Λ K , 0 nov in Example 5.9 , and A A A is the completion of a free Λ K n o v \Lambda_K^{\mathrm{nov}} Λ K nov - or Λ K , 0 n o v \Lambda_{K,0}^{\mathrm{nov}} Λ K , 0 nov -module, then Definition 5.10 reduces to filtered A ∞ A_\infty A ∞ -algebras as defined in Fukaya–Oh–Ohta–Ono [33, §3.2]. Fukaya et al. associate filtered A ∞ A_\infty A ∞ -algebras of this kind to suitable Lagrangians L L L in a symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) , as part of their theory of Lagrangian Floer cohomology.
(b) Let R R R be a pro-Artinian local ring with maximal ideal i ⊂ R \mathfrak{i}\subset R i ⊂ R . Make R R R into a filtered ring over ( N , ≤ ) (\mathbb{N},\le) ( N , ≤ ) by setting F n R = i n F^nR=\mathfrak{i}^n F n R = i n for n ∈ N n\in\mathbb{N} n ∈ N . Then N \mathbb{N} N -filtered A ∞ A_\infty A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m ) over R R R in Definition 5.10 , with filtration F n A = i n ⋅ A F^nA=\mathfrak{i}^n\cdot A F n A = i n ⋅ A , are the curved A ∞ A_\infty A ∞ -algebras over R R R studied by Positselski [78].
We can generalize much of § 3.3 –§ 3.4 to the filtered case.
Definition 5.13. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and ( A , m ) , ( B , n ) (A,\mathbf{m}),(B,\mathbf{n}) ( A , m ) , ( B , n ) be filtered A ∞ A_\infty A ∞ -algebras over R R R . A filtered A ∞ A_\infty A ∞ -morphism f : ( A , m ) → ( B , n ) f:(A,\mathbf{m})\to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) is f = ( f k ) k ≥ 0 f=(f_k)_{k\ge0} f = ( f k ) k ≥ 0 , where f k : A ⊗ k → B f_k:A^{\otimes k}\to B f k : A ⊗ k → B for k = 0 , 1 , 2 , … k=0,1,2,\ldots k = 0 , 1 , 2 , … are graded R R R -linear maps of degree 1 − k 1-k 1 − k , satisfying:
(i) f k f_k f k is filtration-preserving for k ≥ 1 k\ge1 k ≥ 1 , that is, f k ( F γ ( A ⊗ k ) ) ⊆ F γ B f_k(F^\gamma(A^{\otimes k}))\subseteq F^\gamma B f k ( F γ ( A ⊗ k )) ⊆ F γ B for all $\gamma\in\Gamma. (ii) f 0 : R → B 1 f_0:R\to B^1 f 0 : R → B 1 has f 0 ( 1 ) ∈ F ≻ 0 B 1 f_0(1)\in F^{\succ0}B^1 f 0 ( 1 ) ∈ F ≻ 0 B 1 . That is, f 0 ( 1 ) ∈ F γ B 1 f_0(1)\in F^\gamma B^1 f 0 ( 1 ) ∈ F γ B 1 for some γ ∈ Γ \gamma\in\Gamma γ ∈ Γ with γ ≻ 0 \gamma\succ0 γ ≻ 0 , so γ ≻ 0 \gamma\succ0 γ ≻ 0 and n γ → 0 n\gamma\to0 nγ → 0 in Γ \Gamma Γ as n → ∞ n\to\infty n → ∞ .
(iii) Generalizing (66) , in maps A ⊗ n → B A^{\otimes n}\to B A ⊗ n → B for all n ≥ 0 n\geq0 n ≥ 0 we have
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k f i + k + 1 ( i d A ⊗ i ⊗ m j ⊗ i d A ⊗ k ) = ∑ k ≥ 0 , n = n 1 + ⋯ + n k n a ≥ 0 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (143) \begin{aligned}
\sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}f_{i+k+1}\left(\mathrm{id}_A^{\otimes i}\otimes m_j\otimes\mathrm{id}_A^{\otimes k}\right)
={}&\sum_{\substack{k\geq0,\ n=n_1+\cdots+n_k\\n_a\geq0}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}n_k\left(f_{n_1}\otimes f_{n_2}\otimes\cdots\otimes f_{n_k}\right). \tag*{(143)}
\end{aligned} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k f i + k + 1 ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) = k ≥ 0 , n = n 1 + ⋯ + n k n a ≥ 0 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (143)
Note that the second line of (143) is an infinite sum, as for example for any k ≥ 1 k\geq1 k ≥ 1 we can take n 1 = n n_1=n n 1 = n and n a = 0 n_a=0 n a = 0 for a = 2 , 3 , … , k a=2,3,\ldots,k a = 2 , 3 , … , k . For a term in this sum with k > n k>n k > n , we see that n a = 0 n_a=0 n a = 0 for at least k − n k-n k − n values of a = 1 , … , k a=1,\ldots,k a = 1 , … , k , so f n 1 ⊗ ⋯ ⊗ f n k f_{n_1}\otimes\cdots\otimes f_{n_k} f n 1 ⊗ ⋯ ⊗ f n k contains at least n − k n-k n − k factors f 0 ( 1 ) f_0(1) f 0 ( 1 ) .
Here is how to interpret (143) in terms of convergent sums. Let a i ∈ F γ i A d i a_i\in F^{\gamma_i}A^{d_i} a i ∈ F γ i A d i , γ i ∈ Γ \gamma_i\in\Gamma γ i ∈ Γ , d i ∈ Z d_i\in\mathbb{Z} d i ∈ Z for i = 1 , … , n i=1,\ldots,n i = 1 , … , n . Evaluating (143) on a 1 ⊗ ⋯ ⊗ a n a_1\otimes\cdots\otimes a_n a 1 ⊗ ⋯ ⊗ a n using Convention 3.1 gives an equation in B d B^d B d , for d = 2 − n + ∑ i = 1 n d i d=2-n+\sum_{i=1}^{n}d_i d = 2 − n + ∑ i = 1 n d i
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k + j + ∑ l = 1 i d l f i + k + 1 ( a 1 ⊗ ⋯ ⊗ a i ⊗ m j ( a i + 1 ⊗ ⋯ ⊗ a i + j ) ⊗ a i + j + 1 ⊗ ⋯ ⊗ a n ) = ∑ k ≥ 0 , n = n 1 + ⋯ + n k n a ≥ 0 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i + ∑ a = 1 n 1 + ⋯ + n i − 1 d a ) n k ( f n 1 ( a 1 ⊗ ⋯ ⊗ a n 1 ) ⊗ f n 2 ( a n 1 + 1 ⊗ ⋯ ⊗ a n 1 + n 2 ) ⊗ ⋯ ⊗ f n k ( a n 1 + ⋯ + n k − 1 + 1 ⊗ ⋯ ⊗ a n ) ) . (144) \begin{aligned}
\sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk+j+\sum_{l=1}^{i}d_l}f_{i+k+1}\bigl(a_1\otimes\cdots\otimes a_i\otimes{}&\\
m_j(a_{i+1}\otimes\cdots\otimes a_{i+j})\otimes a_{i+j+1}\otimes\cdots\otimes a_n\bigr)
={}&\\
\sum_{\substack{k\geq0,\ n=n_1+\cdots+n_k\\n_a\geq0}}(-1)^{\sum_{i=1}^{k}(n_i+1)\left(k-i+\sum_{a=1}^{n_1+\cdots+n_{i-1}}d_a\right)}n_k\bigl(f_{n_1}(a_1\otimes\cdots\otimes a_{n_1}){}&\\
\otimes f_{n_2}(a_{n_1+1}\otimes\cdots\otimes a_{n_1+n_2})\otimes\cdots\\
&\otimes f_{n_k}(a_{n_1+\cdots+n_{k-1}+1}\otimes\cdots\otimes a_n)\bigr). \tag*{(144)}
\end{aligned} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k + j + ∑ l = 1 i d l f i + k + 1 ( a 1 ⊗ ⋯ ⊗ a i ⊗ m j ( a i + 1 ⊗ ⋯ ⊗ a i + j ) ⊗ a i + j + 1 ⊗ ⋯ ⊗ a n ) = k ≥ 0 , n = n 1 + ⋯ + n k n a ≥ 0 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i + ∑ a = 1 n 1 + ⋯ + n i − 1 d a ) n k ( f n 1 ( a 1 ⊗ ⋯ ⊗ a n 1 ) ⊗ f n 2 ( a n 1 + 1 ⊗ ⋯ ⊗ a n 1 + n 2 ) ⊗ ⋯ ⊗ f n k ( a n 1 + ⋯ + n k − 1 + 1 ⊗ ⋯ ⊗ a n ) ) . (144)
We will show that the second line of (144) is a convergent infinite sum in the complete R R R -module B d B^d B d in the sense of Definition 5.6, and so is well defined. Consider the term in the second line of (144) from n = n 1 + ⋯ + n k n=n_1+\cdots+n_k n = n 1 + ⋯ + n k with k > n k>n k > n . This includes at least k − n k-n k − n terms f 0 ( 1 ) ∈ F γ B 1 f_0(1)\in F^\gamma B^1 f 0 ( 1 ) ∈ F γ B 1 . Thus by a i ∈ F γ i A d i a_i\in F^{\gamma_i}A^{d_i} a i ∈ F γ i A d i and the filtration-preserving properties of n k n_k n k and f n i f_{n_i} f n i for n i ≥ 1 n_i\geq1 n i ≥ 1 , we see see that
± n k ( f n 1 ( a 1 ⊗ ⋯ ⊗ a n 1 ) ⊗ ⋯ ⊗ f n k ( a n 1 + ⋯ + n k − 1 + 1 ⊗ ⋯ ⊗ a n ) ) ∈ F ( k − n ) γ + γ 1 + ⋯ + γ n B d . (145) \pm n_k\bigl(f_{n_1}(a_1\otimes\cdots\otimes a_{n_1})\otimes\cdots\otimes f_{n_k}(a_{n_1+\cdots+n_{k-1}+1}\otimes\cdots\otimes a_n)\bigr)
\in F^{(k-n)\gamma+\gamma_1+\cdots+\gamma_n}B^d. \tag*{(145)} ± n k ( f n 1 ( a 1 ⊗ ⋯ ⊗ a n 1 ) ⊗ ⋯ ⊗ f n k ( a n 1 + ⋯ + n k − 1 + 1 ⊗ ⋯ ⊗ a n ) ) ∈ F ( k − n ) γ + γ 1 + ⋯ + γ n B d . (145)
Since k γ → ∞ k\gamma\to\infty k γ → ∞ in Γ \Gamma Γ as k → ∞ k\to\infty k → ∞ , we see that ( k − n ) γ + γ 1 + ⋯ + γ n → ∞ (k-n)\gamma+\gamma_1+\cdots+\gamma_n\to\infty ( k − n ) γ + γ 1 + ⋯ + γ n → ∞ in Γ \Gamma Γ as k → ∞ k\to\infty k → ∞ . Hence for any δ ∈ Γ \delta\in\Gamma δ ∈ Γ there exists K ≥ 0 K\geq0 K ≥ 0 such that ( k − n ) γ + γ 1 + ⋯ + γ n ⪰ δ (k-n)\gamma+\gamma_1+\cdots+\gamma_n\succeq\delta ( k − n ) γ + γ 1 + ⋯ + γ n ⪰ δ for all k ≥ K k\geq K k ≥ K . Thus (5.6) lies in F δ B d F^\delta B^d F δ B d if k ≥ K k\geq K k ≥ K . As there are only finitely many terms in the sum with k < K k<K k < K , only finitely many terms in the sum in the second line of (5.5) do not lie in F δ B d F^\delta B^d F δ B d for each δ ∈ Γ \delta\in\Gamma δ ∈ Γ , so the sum is convergent.
We have shown that (5.4) makes sense as a convergent sum when evaluated on a 1 ⊗ ⋯ ⊗ a n a_1\otimes\cdots\otimes a_n a 1 ⊗ ⋯ ⊗ a n in A n A^n A n for a i ∈ A d i a_i\in A^{d_i} a i ∈ A d i , i = 1 , … , n i=1,\ldots,n i = 1 , … , n . Extending by linearity, equation (5.4) makes sense as an equation in R R R -linear maps A ⊗ n → B A^{\otimes n}\to B A ⊗ n → B .
If ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) are strict, that is, m 0 = n 0 = 0 m_0=n_0=0 m 0 = n 0 = 0 , we call a filtered A ∞ A_\infty A ∞ -morphism f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) strict if f 0 = 0 f_0=0 f 0 = 0 . Then f f f is an ordinary A ∞ A_\infty A ∞ -morphism of A ∞ A_\infty A ∞ -algebras.
The identity morphism id A : ( A , m ) → ( A , m ) \operatorname{id}_A : (A,\mathbf{m}) \to(A,\mathbf{m}) id A : ( A , m ) → ( A , m ) is id A = ( id A , k ) k ≥ 0 \operatorname{id}_A = (\operatorname{id}_{A,k})_{k\geq0} id A = ( id A , k ) k ≥ 0 with id A , 1 = id A \operatorname{id}_{A,1}=\operatorname{id}_A id A , 1 = id A and id A , k = 0 \operatorname{id}_{A,k}=0 id A , k = 0 for k ≥ 1 k\geq1 k ≥ 1 .
If ( A , m ) (A,\mathbf{m}) ( A , m ) , ( B , n ) (B,\mathbf{n}) ( B , n ) , ( C , o ) (C,\mathbf{o}) ( C , o ) are Γ \Gamma Γ -filtered A ∞ A_{\infty} A ∞ -algebras over R R R and f : ( A , m ) → ( B , n ) \mathfrak{f} : (A,\mathbf{m}) \to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) , g : ( B , n ) → ( C , o ) \mathfrak{g} : (B,\mathbf{n}) \to(C,\mathbf{o}) g : ( B , n ) → ( C , o ) are filtered A ∞ A_{\infty} A ∞ -morphisms, generalizing (69) , the composition g ∘ f : ( A , m ) → ( C , o ) \mathfrak{g}\circ\mathfrak{f} : (A,\mathbf{m}) \to(C,\mathbf{o}) g ∘ f : ( A , m ) → ( C , o ) is given by
( g ∘ f ) n = ∑ n = n 1 + ⋯ + n k n a ≥ 0 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) g k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) . (\mathfrak{g}\circ\mathfrak{f})_n=\sum_{\substack{n=n_1+\cdots+n_k\\ n_a\geq0}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\mathfrak{g}_k(\mathfrak{f}_{n_1}\otimes\mathfrak{f}_{n_2}\otimes\cdots\otimes\mathfrak{f}_{n_k}). ( g ∘ f ) n = n = n 1 + ⋯ + n k n a ≥ 0 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) g k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n k ) .
This is an infinite sum, but it is convergent in the same sense as (143) , as if k > n k>n k > n then g k ( f n 1 ⊗ ⋯ ⊗ f n k ) \mathfrak{g}_k(\mathfrak{f}_{n_1}\otimes\cdots\otimes\mathfrak{f}_{n_k}) g k ( f n 1 ⊗ ⋯ ⊗ f n k ) contains at least k − n k-n k − n factors of f 0 ( 1 ) \mathfrak{f}_0(1) f 0 ( 1 ) . Composition is associative. Write A ∞ f i Alg R A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_R A ∞ fi Alg R for the category with objects filtered A ∞ A_{\infty} A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m ) over R R R and morphisms filtered A ∞ A_{\infty} A ∞ -morphisms f : ( A , m ) → ( B , n ) \mathfrak{f} : (A,\mathbf{m}) \to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) .
If ( A , m ) (A,\mathbf{m}) ( A , m ) , ( B , n ) (B,\mathbf{n}) ( B , n ) are unital filtered A ∞ A_{\infty} A ∞ -algebras, as in Definition 5.10, we call a filtered A ∞ A_{\infty} A ∞ -morphism f : ( A , m ) → ( B , n ) \mathfrak{f} : (A,\mathbf{m}) \to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) unital if (70) holds. Unital filtered A ∞ A_{\infty} A ∞ -algebras over R R R and unital filtered A ∞ A_{\infty} A ∞ -morphisms form a non-full subcategory A ∞ f i , u n Alg R ⊂ A ∞ f i Alg R A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_R\subset A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_R A ∞ fi , un Alg R ⊂ A ∞ fi Alg R .
Define subcategories A ∞ f i Alg R , ∞ ⊂ A ∞ f i Alg R , 1 ⊂ A ∞ f i Alg R , 0 = A ∞ f i Alg R A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,1}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,0}=A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_R A ∞ fi Alg R , ∞ ⊂ A ∞ fi Alg R , 1 ⊂ A ∞ fi Alg R , 0 = A ∞ fi Alg R , such that A ∞ f i Alg R , 1 A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,1} A ∞ fi Alg R , 1 has objects strict filtered A ∞ A_{\infty} A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m ) with m 0 = 0 m_0=0 m 0 = 0 and morphisms all filtered A ∞ A_{\infty} A ∞ -morphisms f : ( A , m ) → ( B , n ) \mathfrak{f} : (A,\mathbf{m})\to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) , including those with f 0 ≠ 0 \mathfrak{f}_0\neq0 f 0 = 0 , and A ∞ f i Alg R , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty} A ∞ fi Alg R , ∞ has objects strict filtered A ∞ A_{\infty} A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m ) , with m 0 = 0 m_0=0 m 0 = 0 , and morphisms strict filtered A ∞ A_{\infty} A ∞ -morphisms f \mathfrak{f} f , with f 0 = 0 \mathfrak{f}_0=0 f 0 = 0 .
Objects of A ∞ f i Alg R , 1 A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,1} A ∞ fi Alg R , 1 , A ∞ f i Alg R , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty} A ∞ fi Alg R , ∞ are ordinary A ∞ A_{\infty} A ∞ -algebras (with extra filtration data), and morphisms of A ∞ f i Alg R , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty} A ∞ fi Alg R , ∞ are A ∞ A_{\infty} A ∞ -morphisms (which preserve filtrations), but morphisms of A ∞ f i Alg R , 1 A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,1} A ∞ fi Alg R , 1 are not A ∞ A_{\infty} A ∞ -morphisms. Thus there is a forgetful functor A ∞ f i Alg R , ∞ → A ∞ Alg R A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty}\to A_{\infty}\operatorname{Alg}_R A ∞ fi Alg R , ∞ → A ∞ Alg R which forgets filtrations. This gives a diagram of categories and functors
A ∞ f i Alg R , ∞ → A ∞ f i Alg R , 1 → A ∞ f i Alg R , 0 = A ∞ f i Alg R ↓ A ∞ Alg R (146) \begin{CD}
A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,\infty} @>>> A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,1} @>>> A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,0}=A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_R \\
@VVV \\
A_{\infty}\operatorname{Alg}_R \tag*{(146)}
\end{CD} A ∞ fi Alg R , ∞ ↓ ⏐ A ∞ Alg R A ∞ fi Alg R , 1 A ∞ fi Alg R , 0 = A ∞ fi Alg R (146)
We define A ∞ f i , u n Alg R , ∞ ⊂ A ∞ f i , u n Alg R , 1 ⊂ A ∞ f i , u n Alg R , 0 = A ∞ f i , u n Alg R A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_{R,\infty}\subset A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_{R,1}\subset A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_{R,0}=A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_R A ∞ fi , un Alg R , ∞ ⊂ A ∞ fi , un Alg R , 1 ⊂ A ∞ fi , un Alg R , 0 = A ∞ fi , un Alg R in the same way. For the categories A ∞ f i Alg R , K A^{\mathrm{fi}}_{\infty}\operatorname{Alg}_{R,K} A ∞ fi Alg R , K , A ∞ f i , u n Alg R , K A^{\mathrm{fi,un}}_{\infty}\operatorname{Alg}_{R,K} A ∞ fi , un Alg R , K for K = 0 , 1 , ∞ K=0,1,\infty K = 0 , 1 , ∞ we call K K K the type . It is the analogue of the type K K K of ∞ \infty ∞ -categories A ∞ u n Alg R , K ∞ , … A^{\mathrm{un}}_{\infty}\operatorname{Alg}_{R,K}^{\infty},\ldots A ∞ un Alg R , K ∞ , … in Chapters 6–8.
The next definition follows Fukaya et al. [33, Def. 4.2.35 & Prop. 4.2.40].
Definition 5.14. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, ( A , m ) (A,\mathbf{m}) ( A , m ) , ( B , n ) (B,\mathbf{n}) ( B , n ) be filtered A ∞ A_{\infty} A ∞ -algebras over R R R , and f , g : ( A , m ) → ( B , n ) \mathfrak{f},\mathfrak{g} : (A,\mathbf{m})\to(B,\mathbf{n}) f , g : ( A , m ) → ( B , n ) be filtered A ∞ A_{\infty} A ∞ -morphisms. A filtered homotopy H : f ⇒ g \mathfrak{H} : \mathfrak{f}\Rightarrow\mathfrak{g} H : f ⇒ g from f \mathfrak{f} f to g \mathfrak{g} g is H = ( H k ) k ≥ 0 \mathfrak{H}=(\mathfrak{H}_k)_{k\geq0} H = ( H k ) k ≥ 0 , where H k : A ⊗ k → B \mathfrak{H}_k:A^{\otimes k}\to B H k : A ⊗ k → B for k = 0 , 1 , 2 , … k=0,1,2,\ldots k = 0 , 1 , 2 , … are graded R R R -linear maps of degree − k -k − k , satisfying:
(i) H k \mathfrak{H}_k H k is filtration-preserving for k ≥ 1 k\geq1 k ≥ 1 , that is, H k ( F γ ( A ⊗ k ) ) ⊆ F γ B \mathfrak{H}_k(F^\gamma(A^{\otimes k}))\subseteq F^\gamma B H k ( F γ ( A ⊗ k )) ⊆ F γ B for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
(ii) H 0 : R → B 0 \mathcal{H}_0:R\to B^0 H 0 : R → B 0 has H 0 ( 1 ) ∈ F γ ≻ 0 B 0 \mathcal{H}_0(1)\in F^{\gamma\succ0}B^0 H 0 ( 1 ) ∈ F γ ≻ 0 B 0 . That is, H 0 ( 1 ) ∈ F γ B 0 \mathcal{H}_0(1)\in F^\gamma B^0 H 0 ( 1 ) ∈ F γ B 0 for some γ ∈ Γ \gamma\in\Gamma γ ∈ Γ with γ ≻ 0 \gamma\succ0 γ ≻ 0 , so γ ≻ 0 \gamma\succ0 γ ≻ 0 and n γ → 0 n\gamma\to0 nγ → 0 in Γ \Gamma Γ as n → ∞ n\to\infty n → ∞ .
(iii) Generalizing (71) , in maps A ⊗ n → B A^{\otimes n}\to B A ⊗ n → B for all n ≥ 0 n\geq0 n ≥ 0 we have
f n − g n = ∑ n = n 1 + ⋯ + n k , n i ≥ 0 1 ≤ l ≤ k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n l − 1 ⊗ H n l ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ) + ∑ i , j , k ≥ 0 , i + j + k = n ( − 1 ) i + j k H i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) . f_n-g_n=
\sum_{\substack{n=n_1+\cdots+n_k,\ n_i\geq0\\1\leq l\leq k}}
(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}
n_k\left(f_{n_1}\otimes f_{n_2}\otimes\cdots\otimes f_{n_{l-1}}\otimes\mathcal{H}_{n_l}\otimes g_{n_{l+1}}\otimes\cdots\otimes g_{n_k}\right)
+
\sum_{\substack{i,j,k\geq0,\ i+j+k=n}}
(-1)^{i+jk}\mathcal{H}_{i+k+1}\circ\left(\operatorname{id}_{A}^{\otimes i}\otimes m_j\otimes\operatorname{id}_{A}^{\otimes k}\right). f n − g n = n = n 1 + ⋯ + n k , n i ≥ 0 1 ≤ l ≤ k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) n k ( f n 1 ⊗ f n 2 ⊗ ⋯ ⊗ f n l − 1 ⊗ H n l ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ) + i , j , k ≥ 0 , i + j + k = n ∑ ( − 1 ) i + j k H i + k + 1 ∘ ( id A ⊗ i ⊗ m j ⊗ id A ⊗ k ) .
The second line is an infinite sum, which converges as for (143) .
If ( A , m ) , ( B , n ) , f , g (A,m),(B,n),f,g ( A , m ) , ( B , n ) , f , g are unital, we call H \mathcal{H} H unital if (72) holds.
Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be filtered A ∞ A_\infty A ∞ -algebras over R R R , and f : ( A , m ) → ( B , n ) f:(A,m)\to(B,n) f : ( A , m ) → ( B , n ) a filtered A ∞ A_\infty A ∞ -morphism. A filtered homotopy inverse for f f f is a filtered A ∞ A_\infty A ∞ -morphism g : ( B , n ) → ( A , m ) g:(B,n)\to(A,m) g : ( B , n ) → ( A , m ) such that g ∘ f : ( A , m ) → ( A , m ) g\circ f:(A,m)\to(A,m) g ∘ f : ( A , m ) → ( A , m ) is filtered homotopic to id A : ( A , m ) → ( A , m ) \operatorname{id}_A:(A,m)\to(A,m) id A : ( A , m ) → ( A , m ) , and f ∘ g : ( B , n ) → ( B , n ) f\circ g:(B,n)\to(B,n) f ∘ g : ( B , n ) → ( B , n ) is filtered homotopic to id B : ( B , n ) → ( B , n ) \operatorname{id}_B:(B,n)\to(B,n) id B : ( B , n ) → ( B , n ) . If f f f has a filtered homotopy inverse, we call f f f a filtered homotopy equivalence , and we call ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) filtered homotopy equivalent .
Here is the analogue of Theorem 3.11(a),(b), whose proof generalizes to the filtered case. (See Seidel [84, §1(h)] for a proof of (a) which extends easily.)
Theorem 5.15. Let ( A , m ) , ( B , n ) (A,m),(B,n) ( A , m ) , ( B , n ) be filtered A ∞ A_\infty A ∞ -algebras over R R R . Then
(a) Filtered homotopy is an equivalence relation on filtered A ∞ A_\infty A ∞ -morphisms f , g : ( A , m ) → ( B , n ) f,g:(A,m)\to(B,n) f , g : ( A , m ) → ( B , n ) .
(b) Filtered homotopy equivalence is an equivalence relation on filtered A ∞ A_\infty A ∞ -algebras.
Remark 5.16. Let ( A , m ) (A,m) ( A , m ) be a Γ \Gamma Γ -filtered A ∞ A_\infty A ∞ -algebra. Then F 0 A / F ≻ 0 A F^0A/F^{\succ0}A F 0 A / F ≻ 0 A is a module over the commutative ring F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R , and m k m_k m k for k ≥ 0 k\geq0 k ≥ 0 maps ( F 0 A ) ⊗ k → F 0 A (F^0A)^{\otimes k}\to F^0A ( F 0 A ) ⊗ k → F 0 A , and this descends to an F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R -linear map
m ˙ k : ( F 0 A / F ≻ 0 A ) ⊗ k ⟶ F 0 A / F ≻ 0 A , \dot{m}_k:(F^0A/F^{\succ0}A)^{\otimes k}\longrightarrow F^0A/F^{\succ0}A, m ˙ k : ( F 0 A / F ≻ 0 A ) ⊗ k ⟶ F 0 A / F ≻ 0 A ,
with m ˙ 0 = 0 \dot{m}_0=0 m ˙ 0 = 0 . (142) holds for the m ˙ k \dot{m}_k m ˙ k , so as m ˙ 0 = 0 \dot{m}_0=0 m ˙ 0 = 0 we see that ( F 0 A / F ≻ 0 A , m ˙ = ( m ˙ k ) k ≥ 1 ) (F^0A/F^{\succ0}A,\dot{m}=(\dot{m}_k)_{k\geq1}) ( F 0 A / F ≻ 0 A , m ˙ = ( m ˙ k ) k ≥ 1 ) is an ordinary A ∞ A_\infty A ∞ -algebra over F 0 R / F ≻ 0 R F^0R/F^{\succ0}R F 0 R / F ≻ 0 R .
Similarly, if we choose γ ≻ 0 \gamma\succ0 γ ≻ 0 with m 0 ( 1 ) ∈ F γ A 2 m_0(1)\in F^\gamma A^2 m 0 ( 1 ) ∈ F γ A 2 then F 0 A / F γ A F^0A/F^\gamma A F 0 A / F γ A becomes an ordinary A ∞ A_\infty A ∞ -algebra ( F 0 A / F γ A , m ¨ = ( m ¨ k ) k ≥ 1 ) (F^0A/F^\gamma A,\ddot{m}=(\ddot{m}_k)_{k\geq1}) ( F 0 A / F γ A , m ¨ = ( m ¨ k ) k ≥ 1 ) over F 0 R / F γ R F^0R/F^\gamma R F 0 R / F γ R .
Filtered A ∞ A_\infty A ∞ -morphisms f f f induce A ∞ A_\infty A ∞ -morphisms f ˙ \dot{f} f ˙ for the ( F 0 A / F ≻ 0 A , m ˙ ) (F^0A/F^{\succ0}A,\dot{m}) ( F 0 A / F ≻ 0 A , m ˙ ) , and (if we can choose γ \gamma γ ‘small enough’ in Γ \Gamma Γ ) f ¨ \ddot{f} f ¨ for the ( F 0 A / F γ A , m ¨ ) (F^0A/F^\gamma A,\ddot{m}) ( F 0 A / F γ A , m ¨ ) .
We could use this to make definitions for Γ \Gamma Γ -filtered A ∞ A_\infty A ∞ -algebras ( A , m ) (A,m) ( A , m ) , and then try and prove theorems. For example, we could define ( A , m ) (A,m) ( A , m ) to be cohomologically unital if ( F 0 A / F ≻ 0 A , m ˙ ) (F^0A/F^{\succ0}A,\dot{m}) ( F 0 A / F ≻ 0 A , m ˙ ) or ( F 0 A / F γ A , m ¨ ) (F^0A/F^\gamma A,\ddot{m}) ( F 0 A / F γ A , m ¨ ) is, in the sense of Definition 3.6, or we could define f f f to be a weak homotopy equivalence if f ˙ \dot{f} f ˙ or f ¨ \ddot{f} f ¨ is, in the sense of Definition 3.8. This approach is taken by Fukaya et al. [28–33] with their notion of filtered A ∞ A_\infty A ∞ -algebra. For example, in [33, Th. 4.2.45] they prove an analogue of of Theorem 3.11(c) for ‘gapped filtered A ∞ A_\infty A ∞ -algebras’.
We will not do this, as it is not helpful for the main results that we want to prove, but it may be useful in further developments of the theory.
Filtered A ∞ A_\infty A ∞ -categories
Next we extend much of §4.1–§4.2 to the filtered case. The next definition combines Definitions 4.1 and 5.10. It generalizes definitions in Fukaya [29, §1], [31, §2] and De Deken–Lowen [18, §2].
Definition 5.17. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring, as in Definitions 5.4 and 5.6. A Γ \Gamma Γ -filtered A ∞ A_\infty A ∞ -category, or just filtered A ∞ A_\infty A ∞ -category, A \mathcal{A} A over R R R consists of:
(b) For all objects x , y ∈ Obj A x,y \in\operatorname{Obj}\mathcal{A} x , y ∈ Obj A , a Z \mathbb{Z} Z -graded Γ \Gamma Γ -filtered R R R -module
Hom ( x , y ) = ⨁ d ∈ Z Hom d ( x , y ) , \operatorname{Hom}(x,y)=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}(x,y), Hom ( x , y ) = d ∈ Z ⨁ Hom d ( x , y ) ,
where Hom d ( x , y ) \operatorname{Hom}^{d}(x,y) Hom d ( x , y ) is a complete filtered R R R -module for each d ∈ Z d\in\mathbb{Z} d ∈ Z . (We do not require Hom ( x , y ) \operatorname{Hom}(x,y) Hom ( x , y ) to be complete.) We say that Hom ( x , y ) \operatorname{Hom}(x,y) Hom ( x , y ) is degree-wise complete . We also write Hom A ( x , y ) \operatorname{Hom}_{\mathcal{A}}(x,y) Hom A ( x , y ) when we want to identify the filtered A ∞ A_\infty A ∞ -category A \mathcal{A} A .
(c) For all k ≥ 0 k\geq0 k ≥ 0 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k ) m_k^{x_0,\ldots,x_k}:\operatorname{Hom}(x_0,x_1)\otimes_R\operatorname{Hom}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}(x_0,x_k) m k x 0 , … , x k : Hom ( x 0 , x 1 ) ⊗ R Hom ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom ( x k − 1 , x k ) ⟶ Hom ( x 0 , x k )
of degree 2 − k 2-k 2 − k . When k = 0 k=0 k = 0 , this is an R R R -linear map m 0 x 0 : R ⟶ Hom 2 ( x 0 , x 0 ) m_0^{x_0}:R\longrightarrow\operatorname{Hom}^{2}(x_0,x_0) m 0 x 0 : R ⟶ Hom 2 ( x 0 , x 0 ) . We write m k = ( m k x 0 , … , x k ) x 0 , … , x k ∈ Obj A m_k=(m_k^{x_0,\ldots,x_k})_{x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A}} m k = ( m k x 0 , … , x k ) x 0 , … , x k ∈ Obj A and m = ( m k ) k ≥ 0 m=(m_k)_{k\geq0} m = ( m k ) k ≥ 0 .
We require m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k to be filtration-preserving for all k ≥ 1 k\geq1 k ≥ 1 and x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , that is,
m k x 0 , … , x k ( F γ ( Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom ( x k − 1 , x k ) ) ) ⊆ F γ Hom ( x 0 , x k ) m_k^{x_0,\ldots,x_k}\left(F^\gamma\left(\operatorname{Hom}(x_0,x_1)\otimes\cdots\otimes\operatorname{Hom}(x_{k-1},x_k)\right)\right)\subseteq F^\gamma\operatorname{Hom}(x_0,x_k) m k x 0 , … , x k ( F γ ( Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom ( x k − 1 , x k ) ) ) ⊆ F γ Hom ( x 0 , x k )
for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
We require that m 0 x 0 ( 1 ) ∈ F ≻ 0 Hom 2 ( x 0 , x 0 ) m_0^{x_0}(1)\in F^{\succ0}\operatorname{Hom}^{2}(x_0,x_0) m 0 x 0 ( 1 ) ∈ F ≻ 0 Hom 2 ( x 0 , x 0 ) for all x 0 x_0 x 0 in Obj A \operatorname{Obj}\mathcal{A} Obj A . That is, m 0 x 0 ( 1 ) ∈ F γ Hom 2 ( x 0 , x 0 ) m_0^{x_0}(1)\in F^\gamma\operatorname{Hom}^{2}(x_0,x_0) m 0 x 0 ( 1 ) ∈ F γ Hom 2 ( x 0 , x 0 ) for some γ ∈ Γ \gamma\in\Gamma γ ∈ Γ with γ ≻ 0 \gamma\succ0 γ ≻ 0 , so γ ≻ 0 \gamma\succ0 γ ≻ 0 and n γ → 0 n\gamma\to0 nγ → 0 in Γ \Gamma Γ as n → ∞ n\to\infty n → ∞ . This γ \gamma γ may depend on x 0 x_0 x 0 .
Generalizing (4.1), we require that for all n ≥ 0 n\geq0 n ≥ 0 and objects x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (147) \sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}\circ\left(\operatorname{id}_{\operatorname{Hom}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{i-1},x_i)}\otimes m_j^{x_i,\ldots,x_{i+j}}\otimes\operatorname{id}_{\operatorname{Hom}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}(x_{n-1},x_n)}\right)=0. \tag*{(147)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom ( x i − 1 , x i ) ⊗ m j x i , … , x i + j ⊗ id Hom ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom ( x n − 1 , x n ) ) = 0. (147)
We say that A \mathcal{A} A is strict if m 0 x 0 = 0 m_0^{x_0}=0 m 0 x 0 = 0 for all x 0 ∈ Obj A x_0\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A . Then A \mathcal{A} A is an ordinary A ∞ A_\infty A ∞ -category, forgetting the filtration structure.
We call A \mathcal{A} A unital if for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A there exists 1 x ∈ F 0 Hom A 0 ( x , x ) 1_x\in F^0\operatorname{Hom}_{\mathcal{A}}^0(x,x) 1 x ∈ F 0 Hom A 0 ( x , x ) , called the identity for x x x , satisfying (4.5)–(4.6). Identities 1 x 1_x 1 x are unique if they exist.
The next definition combines Definitions 4.9 and 5.13.
Definition 5.18. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and A , B \mathcal{A},\mathcal{B} A , B be filtered A ∞ A_\infty A ∞ -categories over R R R . A filtered A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B consists of:
(a) A map F O b j : Obj A → Obj B \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{A}\to\operatorname{Obj}\mathcal{B} F Obj : Obj A → Obj B . When this will not cause confusion, for x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A we write F ( x ) ∈ Obj B \mathcal{F}(x)\in\operatorname{Obj}\mathcal{B} F ( x ) ∈ Obj B rather than F O b j ( x ) \mathcal{F}_{\mathrm{Obj}}(x) F Obj ( x ) .
(b) For all k ≥ 0 k\geq0 k ≥ 0 and objects x 0 , x 1 , … , x k x_0,x_1,\ldots,x_k x 0 , x 1 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , a graded R R R -linear map
F k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F O b j ( x 0 ) , F O b j ( x k ) ) , \mathcal{F}^{x_0,\ldots,x_k}_k:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{F}_{\mathrm{Obj}}(x_k)), F k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , F Obj ( x k )) , of degree 1 − k 1-k 1 − k . When k = 0 k=0 k = 0 , this is an R R R -linear map F 0 x 0 : R → Hom B 1 ( F ( x 0 ) , F ( x 0 ) ) \mathcal{F}^{x_0}_0:R\to\operatorname{Hom}_{\mathcal{B}}^1(\mathcal{F}(x_0),\mathcal{F}(x_0)) F 0 x 0 : R → Hom B 1 ( F ( x 0 ) , F ( x 0 )) .
We require F k x 0 , … , x k \mathcal{F}^{x_0,\ldots,x_k}_k F k x 0 , … , x k to be filtration-preserving for all k ≥ 1 k\geq1 k ≥ 1 and x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , that is,
F k x 0 , … , x k ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ) ) ⊆ F γ Hom B ( F ( x 0 ) , F ( x k ) ) \mathcal{F}^{x_0,\ldots,x_k}_k\left(\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes\cdots\otimes\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\right)\subseteq F^\gamma\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_k)) F k x 0 , … , x k ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ) ) ⊆ F γ Hom B ( F ( x 0 ) , F ( x k )) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
We require that F 0 x 0 ( 1 ) ∈ F ≻ 0 Hom B 1 ( F ( x 0 ) , F ( x 0 ) ) \mathcal{F}^{x_0}_0(1)\in F^{\succ0}\operatorname{Hom}_{\mathcal{B}}^1(\mathcal{F}(x_0),\mathcal{F}(x_0)) F 0 x 0 ( 1 ) ∈ F ≻ 0 Hom B 1 ( F ( x 0 ) , F ( x 0 )) for all x 0 x_0 x 0 in Obj A \operatorname{Obj}\mathcal{A} Obj A .
Generalizing (4.8) and (5.4), the F k x 0 , … , x k \mathcal{F}^{x_0,\ldots,x_k}_k F k x 0 , … , x k must satisfy for all n ≥ 0 n\geq0 n ≥ 0 and x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n in Obj A \operatorname{Obj}\mathcal{A} Obj A
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) \sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}\mathcal{F}^{x_0,\ldots,x_i,x_{i+j},\ldots,x_n}_{i+k+1}\circ\left(\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_i)}\otimes m^{\mathcal{A};x_i,\ldots,x_{i+j}}_j\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{n-1},x_n)}\right) i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k F i + k + 1 x 0 , … , x i , x i + j , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) = ∑ n = n 1 + ⋯ + n k n a ≥ 0 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F O b j ( x 0 ) , F O b j ( x n 1 ) , … , F O b j ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) . (148) =\sum_{\substack{n=n_1+\cdots+n_k\\n_a\geq0}}(-1)^{\sum_{i=1}^k(n_i+1)(k-i)}m^{\mathcal{B};\mathcal{F}_{\mathrm{Obj}}(x_0),\mathcal{F}_{\mathrm{Obj}}(x_{n_1}),\ldots,\mathcal{F}_{\mathrm{Obj}}(x_n)}_k\left(\mathcal{F}^{x_0,\ldots,x_{n_1}}_{n_1}\otimes\mathcal{F}^{x_{n_1},\ldots,x_{n_1+n_2}}_{n_2}\otimes\cdots\otimes\mathcal{F}^{x_{n-n_k},\ldots,x_n}_{n_k}\right). \tag*{(148)} = n = n 1 + ⋯ + n k n a ≥ 0 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F Obj ( x 0 ) , F Obj ( x n 1 ) , … , F Obj ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) . (148) The second line of (5.9) is an infinite sum, which converges in a similar way to (5.4), since F 0 x j ( 1 ) ∈ F ≻ 0 Hom B 1 ( F ( x j ) , F ( x j ) ) \mathcal{F}^{x_j}_0(1)\in F^{\succ0}\operatorname{Hom}_{\mathcal{B}}^1(\mathcal{F}(x_j),\mathcal{F}(x_j)) F 0 x j ( 1 ) ∈ F ≻ 0 Hom B 1 ( F ( x j ) , F ( x j )) .
If A , B \mathcal{A},\mathcal{B} A , B are strict, we call a filtered A ∞ A_\infty A ∞ -functor F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B strict if F 0 x 0 = 0 \mathcal{F}^{x_0}_0=0 F 0 x 0 = 0 for all x 0 ∈ Obj A x_0\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A . Then F \mathcal{F} F is an ordinary A ∞ A_\infty A ∞ -functor of A ∞ A_\infty A ∞ -categories.
The identity functor id A : A → A \operatorname{id}^{\mathcal{A}}:\mathcal{A}\to\mathcal{A} id A : A → A has id O b j A = id Obj A \operatorname{id}^{\mathcal{A}}_{\mathrm{Obj}}=\operatorname{id}_{\operatorname{Obj}\mathcal{A}} id Obj A = id Obj A , and id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) \operatorname{id}^{\mathcal{A},x}_1=\operatorname{id}:\operatorname{Hom}_{\mathcal{A}}(x,x)\to\operatorname{Hom}_{\mathcal{A}}(x,x) id 1 A , x = id : Hom A ( x , x ) → Hom A ( x , x ) , and id k A = 0 \operatorname{id}^{\mathcal{A}}_k=0 id k A = 0 for k ≠ 1 k\neq1 k = 1 .
If F : A → B \mathcal{F}:\mathcal{A}\to\mathcal{B} F : A → B , G : B → C \mathcal{G}:\mathcal{B}\to\mathcal{C} G : B → C are filtered A ∞ A_\infty A ∞ -functors, generalizing (4.10) the composition G ∘ F : A → C \mathcal{G}\circ\mathcal{F}:\mathcal{A}\to\mathcal{C} G ∘ F : A → C is given by ( G ∘ F ) O b j = G O b j ∘ F O b j (\mathcal{G}\circ\mathcal{F})_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}}\circ\mathcal{F}_{\mathrm{Obj}} ( G ∘ F ) Obj = G Obj ∘ F Obj and
( G ∘ F ) n x 0 , … , x n = ∑ n = n 1 + ⋯ + n k n a ≥ 0 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) , (149) (\mathcal{G}\circ\mathcal{F})^{x_0,\ldots,x_n}_n=
\sum_{\substack{n=n_1+\cdots+n_k\\n_a\geq0}}(-1)^{\sum_{i=1}^k(n_i+1)(k-i)}\mathcal{G}^{\mathcal{F}(x_0),\mathcal{F}(x_{n_1}),\mathcal{F}(x_{n_1+n_2}),\ldots,\mathcal{F}(x_n)}_k\left(\mathcal{F}^{x_0,\ldots,x_{n_1}}_{n_1}\otimes\mathcal{F}^{x_{n_1},\ldots,x_{n_1+n_2}}_{n_2}\otimes\cdots\otimes\mathcal{F}^{x_{n-n_k},\ldots,x_n}_{n_k}\right), \tag*{(149)} ( G ∘ F ) n x 0 , … , x n = n = n 1 + ⋯ + n k n a ≥ 0 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) G k F ( x 0 ) , F ( x n 1 ) , F ( x n 1 + n 2 ) , … , F ( x n ) ( F n 1 x 0 , … , x n 1 ⊗ F n 2 x n 1 , … , x n 1 + n 2 ⊗ ⋯ ⊗ F n k x n − n k , … , x n ) , (149) @page 107/200 starts_mid=0 ends_mid=1
for all n ≥ 0 n \ge0 n ≥ 0 and x 0 , … , x n ∈ Obj A x_0,\ldots,x_n \in\operatorname{Obj}\mathcal{A} x 0 , … , x n ∈ Obj A . Again, the second line of (5.10) is an infinite sum, which converges as for (5.9). Composition is associative.
Write A f i ∞ C a t R \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_R Afi ∞ Cat R for the category with objects filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A over R R R and morphisms filtered A ∞ A_{\infty} A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B .
If A , B \mathcal{A},\mathcal{B} A , B are unital filtered A ∞ A_{\infty} A ∞ -categories over R R R , as in Definition 4.5, we call a filtered A ∞ A_{\infty} A ∞ -functor F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B unital if (4.11)–(4.12) hold. Unital filtered A ∞ A_{\infty} A ∞ -categories over R R R and unital filtered A ∞ A_{\infty} A ∞ -functors form a non-full subcategory A f i ∞ C a t R u n ⊂ A f i ∞ C a t R \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_R \subset\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_R Afi ∞ Cat R un ⊂ Afi ∞ Cat R .
Define subcategories A f i ∞ C a t R , ∞ ⊂ A f i ∞ C a t R , 1 ⊂ A f i ∞ C a t R , 0 = A f i ∞ C a t R \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty} \subset\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,1} \subset\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,0} = \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_R Afi ∞ Cat R , ∞ ⊂ Afi ∞ Cat R , 1 ⊂ Afi ∞ Cat R , 0 = Afi ∞ Cat R , such that A f i ∞ C a t R , 1 \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,1} Afi ∞ Cat R , 1 has objects strict filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A over R R R , with m 0 x = 0 m_0^x=0 m 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A and morphisms all filtered A ∞ A_{\infty} A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B , including those with F ~ 0 x ≠ 0 \widetilde{\mathcal{F}}^x_0\ne0 F 0 x = 0 , and A f i ∞ C a t R , ∞ \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty} Afi ∞ Cat R , ∞ has objects strict filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A , with m 0 x = 0 m_0^x=0 m 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A , and morphisms strict filtered A ∞ A_{\infty} A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B , with F ~ 0 x = 0 \widetilde{\mathcal{F}}^x_0=0 F 0 x = 0 for all x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A .
Objects of A f i ∞ C a t R , 1 \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,1} Afi ∞ Cat R , 1 , A f i ∞ C a t R , ∞ \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty} Afi ∞ Cat R , ∞ are ordinary A ∞ A_{\infty} A ∞ -categories (with extra filtration data), and morphisms of A f i ∞ C a t R , ∞ \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty} Afi ∞ Cat R , ∞ are A ∞ A_{\infty} A ∞ -functors (which preserve filtrations), but morphisms of A f i ∞ C a t R , 1 \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,1} Afi ∞ Cat R , 1 are not A ∞ A_{\infty} A ∞ -functors. Thus there is a forgetful functor Π A f i ∞ A ∞ : A f i ∞ C a t R , ∞ → A ∞ C a t R \Pi_{\mathop{\mathrm{Afi}}\nolimits_{\infty}}^{A_{\infty}}:\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty}\to A_{\infty}\mathrm{Cat}_R Π Afi ∞ A ∞ : Afi ∞ Cat R , ∞ → A ∞ Cat R which forgets filtrations. As in (5.7), this gives a diagram of categories and functors
A f i ∞ C a t R , ∞ → A f i ∞ C a t R , 1 → A f i ∞ C a t R , 0 = A f i ∞ C a t R Π A f i ∞ A ∞ ↓ A ∞ C a t R (150) \begin{CD}
\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,\infty} @>>> \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,1} @>>> \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,0}=\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_R\\
@V{\Pi_{\mathop{\mathrm{Afi}}\nolimits_{\infty}}^{A_{\infty}}}VV @. @.\\
A_{\infty}\mathrm{Cat}_R @. @. @. \tag*{(150)}
\end{CD} Afi ∞ Cat R , ∞ Π Afi ∞ A ∞ ↓ ⏐ A ∞ Cat R Afi ∞ Cat R , 1 Afi ∞ Cat R , 0 = Afi ∞ Cat R (150)
We define A f i ∞ C a t R , ∞ u n ⊂ A f i ∞ C a t R , 1 u n ⊂ A f i ∞ C a t R , 0 u n = A f i ∞ C a t R u n \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R,\infty} \subset\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R,1} \subset\mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R,0} = \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_R Afi ∞ Cat R , ∞ un ⊂ Afi ∞ Cat R , 1 un ⊂ Afi ∞ Cat R , 0 un = Afi ∞ Cat R un in the same way. For the categories A f i ∞ C a t R , K \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}_{R,K} Afi ∞ Cat R , K , A f i ∞ C a t R , K u n \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R,K} Afi ∞ Cat R , K un for K = 0 , 1 , ∞ K=0,1,\infty K = 0 , 1 , ∞ we call K K K the type . (The choice of labels K = 0 , 1 , ∞ K=0,1,\infty K = 0 , 1 , ∞ correspond to the subscript K K K in the ∞ \infty ∞ -categories A f i ∞ C a t R , K , L ∞ \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\infty}_{R,K,L} Afi ∞ Cat R , K , L ∞ , A f i ∞ C a t R , K , L u n , ∞ \mathop{\mathrm{Afi}}\nolimits_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} Afi ∞ Cat R , K , L un , ∞ in Chapter 8, see equation (8.1).)
The A ∞ A_{\infty} A ∞ -category of filtered A ∞ A_{\infty} A ∞ -functors
When A , B \mathcal{A},\mathcal{B} A , B are A ∞ A_{\infty} A ∞ -categories over R R R , section 4.3 defined the A ∞ A_{\infty} A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) of A ∞ A_{\infty} A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B . There is an analogue of this for filtered A ∞ A_{\infty} A ∞ -categories, which is discussed by De Deken–Lowen [18, §3.4–§3.5] and Fukaya [31, Def. 2.18, Th. 2.19 & Prop. 2.20] for their notions of filtered A ∞ A_{\infty} A ∞ -categories.
Definition 5.19. (a) Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and A , B \mathcal{A},\mathcal{B} A , B be filtered A ∞ A_{\infty} A ∞ -categories over R R R . We will define a strict filtered A ∞ A_{\infty} A ∞ -category Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) over R R R . Objects of Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) are filtered A ∞ A_{\infty} A ∞ -functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B , as in §5.4. For objects F ~ , G ~ \widetilde{\mathcal{F}},\widetilde{\mathcal{G}} F , G , we define a Γ \Gamma Γ -filtered graded R R R -module
Hom Fun A ∞ f i ( A , B ) ( F ~ , G ~ ) = Hom A ∞ f i ( F ~ , G ~ ) = ⨁ l ∈ Z Hom A ∞ f i l ( F ~ , G ~ ) \operatorname{Hom}_{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}})=\operatorname{Hom}_{A_{\infty}^{\mathrm{fi}}}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}})=\bigoplus_{l\in\mathbb{Z}}\operatorname{Hom}^{l}_{A_{\infty}^{\mathrm{fi}}}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}}) Hom Fun A ∞ fi ( A , B ) ( F , G ) = Hom A ∞ fi ( F , G ) = l ∈ Z ⨁ Hom A ∞ fi l ( F , G )
as follows. As in (4.13), an element T \mathcal{T} T of Hom A ∞ l ( F ~ , G ~ ) \operatorname{Hom}_{\mathcal{A}_{\infty}}^{l}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}}) Hom A ∞ l ( F , G ) consists of, for all k ≥ 0 k \ge0 k ≥ 0 and objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in A \mathcal{A} A , a graded R R R -linear map of degree l − k l-k l − k
T k = T k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F ~ O b j ( x 0 ) , G ~ O b j ( x k ) ) , \mathcal{T}_k=\mathcal{T}_{k}^{x_0,\ldots,x_k}:\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\widetilde{\mathcal{F}}_{\mathrm{Obj}}(x_0),\widetilde{\mathcal{G}}_{\mathrm{Obj}}(x_k)), T k = T k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , G Obj ( x k )) ,
such that there exists δ ∈ Γ \delta\in\Gamma δ ∈ Γ such that for all k ≥ 0 k\ge0 k ≥ 0 , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and γ ∈ Γ \gamma\in\Gamma γ ∈ Γ we have
T k x 0 , … , x k ( F γ ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ) ) ) ⊆ F γ + δ Hom B ( F ~ ( x 0 ) , G ~ ( x k ) ) . (151) \mathcal{T}_{k}^{x_0,\ldots,x_k}\left(F^\gamma\left(\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)\otimes\cdots\otimes\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_k)\right)\right)\subseteq F^{\gamma+\delta}\operatorname{Hom}_{\mathcal{B}}(\widetilde{\mathcal{F}}(x_0),\widetilde{\mathcal{G}}(x_k)). \tag*{(151)} T k x 0 , … , x k ( F γ ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ) ) ) ⊆ F γ + δ Hom B ( F ( x 0 ) , G ( x k )) . (151)
We define the Γ \Gamma Γ -filtration on Hom A ∞ l ( F ~ , G ~ ) \operatorname{Hom}_{\mathcal{A}_{\infty}}^{l}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}}) Hom A ∞ l ( F , G ) by T ∈ F δ Hom A ∞ l ( F ~ , G ~ ) \mathcal{T}\in F^\delta\operatorname{Hom}_{\mathcal{A}_{\infty}}^{l}(\widetilde{\mathcal{F}},\widetilde{\mathcal{G}}) T ∈ F δ Hom A ∞ l ( F , G ) for δ ∈ Γ \delta\in\Gamma δ ∈ Γ if (5.12) holds for all k ≥ 0 k\ge0 k ≥ 0 , x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A and γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
The A ∞ A_\infty A ∞ -operations n p F ~ 0 , … , F ~ p \mathfrak{n}_{p}^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p} n p F 0 , … , F p for p ≥ 0 p\ge0 p ≥ 0 are defined by n 0 F ~ 0 = 0 \mathfrak{n}_0^{\widetilde{\mathcal{F}}_0}=0 n 0 F 0 = 0 , and n p F ~ 0 , … , F ~ p \mathfrak{n}_{p}^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p} n p F 0 , … , F p for p ≥ 1 p\ge1 p ≥ 1 are as in (4.14)–(4.17), except that we include terms m 0 A ; x 0 m_{0}^{\mathcal{A};x_0} m 0 A ; x 0 , m 0 B ; y 0 m_{0}^{\mathcal{B};y_0} m 0 B ; y 0 , F ~ 0 x 0 \widetilde{\mathcal{F}}_{0}^{x_0} F 0 x 0 , so in (4.15) and (4.17) we omit the conditions l i > l i − 1 l_i>l_{i-1} l i > l i − 1 , l j > l j − 1 l_j>l_{j-1} l j > l j − 1 . Then the second line of (4.15), and (4.17), become infinite sums, which converge in the usual way. As in De Deken–Lowen [18, §3.4.2], this makes Fun A ∞ ( A , B ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) into a strict filtered A ∞ A_\infty A ∞ -category over R R R , and thus an A ∞ A_\infty A ∞ -category over R R R .
(b) Let A 1 , A 2 , B \mathcal{A}_1,\mathcal{A}_2,\mathcal{B} A 1 , A 2 , B be filtered A ∞ A_\infty A ∞ -categories over R R R , and E : A 1 → A 2 \mathcal{E}:\mathcal{A}_1\to\mathcal{A}_2 E : A 1 → A 2 be a filtered A ∞ A_\infty A ∞ -functor. Generalizing Definition 4.13(a), we define a strict filtered A ∞ A_\infty A ∞ -functor E ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) \mathcal{E}^*:\operatorname{Fun}_{A_{\infty}}(\mathcal{A}_2,\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}(\mathcal{A}_1,\mathcal{B}) E ∗ : Fun A ∞ ( A 2 , B ) → Fun A ∞ ( A 1 , B ) to act on objects by E O b j ∗ : F ~ ↦ F ~ ∘ E \mathcal{E}^*_{\mathrm{Obj}}:\widetilde{\mathcal{F}}\mapsto\widetilde{\mathcal{F}}\circ\mathcal{E} E Obj ∗ : F ↦ F ∘ E for F ~ : A 2 → B \widetilde{\mathcal{F}}:\mathcal{A}_2\to\mathcal{B} F : A 2 → B a filtered A ∞ A_\infty A ∞ -functor, and if p ≥ 0 p\ge0 p ≥ 0 and F ~ 0 , … , F ~ p \widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p F 0 , … , F p are objects in Fun A ∞ ( A 2 , B ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A}_2,\mathcal{B}) Fun A ∞ ( A 2 , B ) , we define
( E ∗ ) p F ~ 0 , … , F ~ p : Hom A ∞ ( F ~ 0 , F ~ 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F ~ p − 1 , F ~ p ) ⟶ Hom A ∞ ( F ~ 0 ∘ E , F ~ p ∘ E ) (\mathcal{E}^*)_{p}^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p}:\operatorname{Hom}_{\mathcal{A}_{\infty}}(\widetilde{\mathcal{F}}_0,\widetilde{\mathcal{F}}_1)\otimes\cdots\otimes\operatorname{Hom}_{\mathcal{A}_{\infty}}(\widetilde{\mathcal{F}}_{p-1},\widetilde{\mathcal{F}}_p)\longrightarrow\operatorname{Hom}_{\mathcal{A}_{\infty}}(\widetilde{\mathcal{F}}_0\circ\mathcal{E},\widetilde{\mathcal{F}}_p\circ\mathcal{E}) ( E ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( F 0 ∘ E , F p ∘ E )
by ( E ∗ ) p F ~ 0 , … , F ~ p = 0 (\mathcal{E}^*)_{p}^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p}=0 ( E ∗ ) p F 0 , … , F p = 0 if p ≠ 1 p\ne1 p = 1 , and ( E ∗ ) 1 F ~ 0 , F ~ 1 (\mathcal{E}^*)_{1}^{\widetilde{\mathcal{F}}_0,\widetilde{\mathcal{F}}_1} ( E ∗ ) 1 F 0 , F 1 is as in (4.20), except that we include terms E 0 x 0 \mathcal{E}_{0}^{x_0} E 0 x 0 , so we sum over n a ≥ 0 n_a\ge0 n a ≥ 0 , and do not require k ≤ n k\le n k ≤ n . Then (4.20) becomes an infinite sum, which converges in the usual way. As in De Deken–Lowen [18, Prop. 3.49], this defines a strict filtered A ∞ A_\infty A ∞ -functor.
(c) Let A , B 1 , B 2 \mathcal{A},\mathcal{B}_1,\mathcal{B}_2 A , B 1 , B 2 be filtered A ∞ A_\infty A ∞ -categories over R R R , and G : B 1 → B 2 \mathcal{G}:\mathcal{B}_1\to\mathcal{B}_2 G : B 1 → B 2 be a filtered A ∞ A_\infty A ∞ -functor. Generalizing Definition 4.13(b), we define a strict filtered A ∞ A_\infty A ∞ -functor G ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) \mathcal{G}_*:\operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}_1)\to\operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}_2) G ∗ : Fun A ∞ ( A , B 1 ) → Fun A ∞ ( A , B 2 ) to act on objects by ( G ∗ ) O b j : F ~ ↦ G ∘ F ~ (\mathcal{G}_*)_{\mathrm{Obj}}:\widetilde{\mathcal{F}}\mapsto\mathcal{G}\circ\widetilde{\mathcal{F}} ( G ∗ ) Obj : F ↦ G ∘ F for F ~ : A → B 1 \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B}_1 F : A → B 1 a filtered A ∞ A_\infty A ∞ -functor, and if p ≥ 0 p\ge0 p ≥ 0 and F ~ 0 , … , F ~ p \widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p F 0 , … , F p are objects in Fun A ∞ ( A , B 1 ) \operatorname{Fun}_{A_{\infty}}(\mathcal{A},\mathcal{B}_1) Fun A ∞ ( A , B 1 ) , we define
( G ∗ ) p F ~ 0 , … , F ~ p : Hom A ∞ ( F ~ 0 , F ~ 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F ~ p − 1 , F ~ p ) ⟶ Hom A ∞ ( G ∘ F ~ 0 , G ∘ F ~ p ) (\mathcal{G}_*)_{p}^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p}:\operatorname{Hom}_{\mathcal{A}_{\infty}}(\widetilde{\mathcal{F}}_0,\widetilde{\mathcal{F}}_1)\otimes\cdots\otimes\operatorname{Hom}_{\mathcal{A}_{\infty}}(\widetilde{\mathcal{F}}_{p-1},\widetilde{\mathcal{F}}_p)\longrightarrow\operatorname{Hom}_{\mathcal{A}_{\infty}}(\mathcal{G}\circ\widetilde{\mathcal{F}}_0,\mathcal{G}\circ\widetilde{\mathcal{F}}_p) ( G ∗ ) p F 0 , … , F p : Hom A ∞ ( F 0 , F 1 ) ⊗ ⋯ ⊗ Hom A ∞ ( F p − 1 , F p ) ⟶ Hom A ∞ ( G ∘ F 0 , G ∘ F p )
by ( G ∗ ) 0 F ~ 0 = 0 (\mathcal{G}_*)_0^{\widetilde{\mathcal{F}}_0}=0 ( G ∗ ) 0 F 0 = 0 , and ( G ∗ ) p F ~ 0 , … , F ~ p (\mathcal{G}_*)_p^{\widetilde{\mathcal{F}}_0,\ldots,\widetilde{\mathcal{F}}_p} ( G ∗ ) p F 0 , … , F p for p ≥ 1 p\ge1 p ≥ 1 are as in (4.21), except that we include terms F ~ 0 x j \widetilde{\mathcal{F}}_{0}^{x_j} F 0 x j and G 0 y 0 \mathcal{G}_{0}^{y_0} G 0 y 0 , so we sum over n j ≥ 0 n_j\ge0 n j ≥ 0 , and do not require k ≤ n k\le n k ≤ n .
Then (4.21) becomes an infinite sum, which converges in the usual way. As in De Deken–Lowen [18, Prop. 3.50], this defines a strict filtered A ∞ A_{\infty} A ∞ -functor.
(d) Generalizing Definition 4.13(c), combining (a)–(c) gives a functor
Fun A ∞ f i : ( A ∞ Cat R f i ) o p × A ∞ Cat R f i ⟶ A ∞ Cat R , ∞ f i . \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}} : (A_{\infty}\operatorname{Cat}_{R}^{\mathrm{fi}})^{\mathrm{op}} \times A_{\infty}\operatorname{Cat}_{R}^{\mathrm{fi}} \longrightarrow A_{\infty}\operatorname{Cat}_{R,\infty}^{\mathrm{fi}}. Fun A ∞ fi : ( A ∞ Cat R fi ) op × A ∞ Cat R fi ⟶ A ∞ Cat R , ∞ fi .
acting on objects ( A , B ) ↦ Fun A ∞ f i ( A , B ) (\mathcal{A},\mathcal{B}) \mapsto\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) ( A , B ) ↦ Fun A ∞ fi ( A , B ) and on morphisms by ( E o p , G ) ↦ E ∗ ∘ G ∗ (\mathcal{E}^{\mathrm{op}},\mathcal{G}) \mapsto\mathcal{E}^{*}\circ\mathcal{G}_{*} ( E op , G ) ↦ E ∗ ∘ G ∗ . Here E ∗ \mathcal{E}^{*} E ∗ and G ∗ \mathcal{G}_{*} G ∗ commute, so their order does not matter. Note that the target A ∞ Cat R , ∞ f i A_{\infty}\operatorname{Cat}_{R,\infty}^{\mathrm{fi}} A ∞ Cat R , ∞ fi is as in Definition 5.18, with strict objects and morphisms, and has a forgetful functor A ∞ Cat R , ∞ f i ⟶ A ∞ Cat R , ∞ A_{\infty}\operatorname{Cat}_{R,\infty}^{\mathrm{fi}}\longrightarrow A_{\infty}\operatorname{Cat}_{R,\infty} A ∞ Cat R , ∞ fi ⟶ A ∞ Cat R , ∞ forgetting filtrations.
(e) The analogue of (a)–(d) works for unital filtered A ∞ A_{\infty} A ∞ -categories and unital filtered A ∞ A_{\infty} A ∞ -functors, giving a functor
Fun A ∞ f i , u n : ( A ∞ Cat R f i , u n ) o p × A ∞ Cat R f i , u n ⟶ A ∞ Cat R , ∞ f i , u n . \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}},\mathrm{un}} : (A_{\infty}\operatorname{Cat}_{R}^{\mathrm{fi},\mathrm{un}})^{\mathrm{op}} \times A_{\infty}\operatorname{Cat}_{R}^{\mathrm{fi},\mathrm{un}} \longrightarrow A_{\infty}\operatorname{Cat}_{R,\infty}^{\mathrm{fi},\mathrm{un}}. Fun A ∞ fi , un : ( A ∞ Cat R fi , un ) op × A ∞ Cat R fi , un ⟶ A ∞ Cat R , ∞ fi , un .
(f) Define Γ 0 = { γ ∈ Γ : γ ⪰ 0 } \Gamma_{0}=\{\gamma\in\Gamma:\gamma\succeq0\} Γ 0 = { γ ∈ Γ : γ ⪰ 0 } , so that ( Γ 0 , ⪯ ) (\Gamma_{0},\preceq) ( Γ 0 , ⪯ ) is a po-submonoid of ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) , and F 0 R F^{0}R F 0 R is a complete ( Γ 0 , ⪯ ) (\Gamma_{0},\preceq) ( Γ 0 , ⪯ ) -filtered ring. Define F 0 Fun A ∞ f i ( A , B ) F^{0}\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ fi ( A , B ) to be the strict filtered A ∞ A_{\infty} A ∞ -category over F 0 R F^{0}R F 0 R which has the same objects F , G \mathcal{F},\mathcal{G} F , G as Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) , and has
Hom F 0 Fun A ∞ f i ( A , B ) ( F , G ) = F 0 Hom Fun A ∞ f i ( A , B ) ( F , G ) , \operatorname{Hom}_{F^{0}\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\mathcal{F},\mathcal{G})=F^{0}\operatorname{Hom}_{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\mathcal{F},\mathcal{G}), Hom F 0 Fun A ∞ fi ( A , B ) ( F , G ) = F 0 Hom Fun A ∞ fi ( A , B ) ( F , G ) ,
as a Γ 0 \Gamma_{0} Γ 0 -filtered F 0 R F^{0}R F 0 R -module. The A ∞ A_{\infty} A ∞ -operations n F 0 , … , F p \mathfrak{n}_{\mathcal{F}^{0},\ldots,\mathcal{F}^{p}} n F 0 , … , F p for p ≥ 0 p\geq0 p ≥ 0 are the restrictions of those for Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) in (a) to F 0 Hom Fun A ∞ f i ( A , B ) ( F 0 , F 1 ) ⊗ ⋯ ⊗ F 0 Hom Fun A ∞ f i ( A , B ) ( F p − 1 , F p ) ⊂ Hom Fun A ∞ f i ( A , B ) ( F 0 , F p ) F^{0}\operatorname{Hom}_{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\mathcal{F}^{0},\mathcal{F}^{1})\otimes\cdots\otimes F^{0}\operatorname{Hom}_{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\mathcal{F}^{p-1},\mathcal{F}^{p})\subset\operatorname{Hom}_{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B})}(\mathcal{F}^{0},\mathcal{F}^{p}) F 0 Hom Fun A ∞ fi ( A , B ) ( F 0 , F 1 ) ⊗ ⋯ ⊗ F 0 Hom Fun A ∞ fi ( A , B ) ( F p − 1 , F p ) ⊂ Hom Fun A ∞ fi ( A , B ) ( F 0 , F p ) , and the A ∞ A_{\infty} A ∞ -equations also hold by restriction.
For A , B \mathcal{A},\mathcal{B} A , B unital we define a strict filtered A ∞ A_{\infty} A ∞ -category F 0 Fun A ∞ f i , u n ( A , B ) F^{0}\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}},\mathrm{un}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ fi , un ( A , B ) over F 0 R F^{0}R F 0 R by applying F 0 ( ⋯ ) F^{0}(\cdots) F 0 ( ⋯ ) to Fun A ∞ f i , u n ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}},\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi , un ( A , B ) in the same way.
As we explain in §5.6, the point of this is that morphisms in Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) are a kind of A ∞ A_{\infty} A ∞ -natural transformation; if we want these to preserve filtrations, we should restrict to morphisms in F 0 Fun A ∞ f i ( A , B ) F^{0}\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ fi ( A , B ) .
Remark 5.20. We emphasize the surprising fact in Definition 5.19 that although A , B \mathcal{A},\mathcal{B} A , B are general filtered A ∞ A_{\infty} A ∞ -categories (with m 0 A ≠ 0 \mathfrak{m}_{0}^{\mathcal{A}}\ne0 m 0 A = 0 ), Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) is a strict filtered A ∞ A_{\infty} A ∞ -category (with n 0 F = 0 \mathfrak{n}_{0}^{\mathcal{F}}=0 n 0 F = 0 ), so in particular Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) is an A ∞ A_{\infty} A ∞ -category. Similarly, although E , G \mathcal{E},\mathcal{G} E , G are general filtered A ∞ A_{\infty} A ∞ -functors, E ∗ \mathcal{E}^{*} E ∗ and G ∗ \mathcal{G}_{*} G ∗ are strict filtered A ∞ A_{\infty} A ∞ -functors, so in particular they are A ∞ A_{\infty} A ∞ -functors.
This is known to De Deken–Lowen [18, §3.4–§3.5]. But Fukaya [31, Th. 2.19] says that Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) is strict if B \mathcal{B} B is strict, which is an unnecessary condition.
The reason why Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) is strict is that, in a similar equation to (4.14)–(4.17), the appropriate definition of n 0 F 0 \mathfrak{n}_{0}^{\mathcal{F}^{0}} n 0 F 0 in Fun A ∞ f i ( A , B ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi ( A , B ) is the right hand side of (5.9) for F 0 \mathcal{F}^{0} F 0 minus the left hand side, and so is zero.
This means we can use cohomology categories such as H 0 ( Fun A ∞ f i , u n ( A , B ) ) H^{0}(\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}},\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ fi , un ( A , B )) , even though cohomology categories H 0 ( A ) H^{0}(\mathcal{A}) H 0 ( A ) for general filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A are not defined.
Here is the analogue of Definitions 4.21 and 5.14, which is equivalent to Fukaya [31, Def. 13.5] as in [33, Prop. 4.2.20].
Definition 5.21. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, A , B \mathcal{A},\mathcal{B} A , B be filtered A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}\colon\mathcal{A}\to\mathcal{B} F , G : A → B be filtered A ∞ A_{\infty} A ∞ -functors which act the same way on objects, F O b j = G O b j \mathcal{F}_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}} F Obj = G Obj . A filtered A ∞ A_{\infty} A ∞ -homotopy H : F ⇒ G \mathcal{H}\colon\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G consists of for all k ≥ 0 k\geq0 k ≥ 0 and objects x 0 , x 1 , … , x k x_{0},x_{1},\ldots,x_{k} x 0 , x 1 , … , x k in A \mathcal{A} A , a graded R R R -linear map of degree − k -k − k
H k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F O b j ( x 0 ) , F O b j ( x k ) ) . \mathcal{H}^{x_{0},\ldots,x_{k}}_{k}\colon\operatorname{Hom}_{\mathcal{A}}(x_{0},x_{1})\otimes_{R}\operatorname{Hom}_{\mathcal{A}}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_{k})\longrightarrow\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}_{\mathrm{Obj}}(x_{0}),\mathcal{F}_{\mathrm{Obj}}(x_{k})). H k x 0 , … , x k : Hom A ( x 0 , x 1 ) ⊗ R Hom A ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A ( x k − 1 , x k ) ⟶ Hom B ( F Obj ( x 0 ) , F Obj ( x k )) .
We require H k x 0 , … , x k \mathcal{H}^{x_{0},\ldots,x_{k}}_{k} H k x 0 , … , x k to be filtration-preserving for all k ≥ 1 k\geq1 k ≥ 1 and x 0 , … , x k x_{0},\ldots,x_{k} x 0 , … , x k in Obj A \operatorname{Obj}\mathcal{A} Obj A , that is H k x 0 , … , x k ( F γ ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ) ) ) ⊆ F γ Hom B ( F ( x 0 ) , F ( x k ) ) \mathcal{H}^{x_{0},\ldots,x_{k}}_{k}(F^{\gamma}(\operatorname{Hom}_{\mathcal{A}}(x_{0},x_{1})\otimes\cdots\otimes\operatorname{Hom}_{\mathcal{A}}(x_{k-1},x_{k})))\subseteq F^{\gamma}\operatorname{Hom}_{\mathcal{B}}(\mathcal{F}(x_{0}),\mathcal{F}(x_{k})) H k x 0 , … , x k ( F γ ( Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ Hom A ( x k − 1 , x k ))) ⊆ F γ Hom B ( F ( x 0 ) , F ( x k )) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
We require that H 0 x 0 ( 1 ) ∈ F γ > 0 Hom B 0 ( F ( x 0 ) , F ( x 0 ) ) \mathcal{H}^{x_{0}}_{0}(1)\in F^{\gamma>0}\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x_{0}),\mathcal{F}(x_{0})) H 0 x 0 ( 1 ) ∈ F γ > 0 Hom B 0 ( F ( x 0 ) , F ( x 0 )) for all x 0 x_{0} x 0 in Obj A \operatorname{Obj}\mathcal{A} Obj A .
Generalizing (4.27), the H k x 0 , … , x k \mathcal{H}^{x_{0},\ldots,x_{k}}_{k} H k x 0 , … , x k must satisfy for all n ≥ 0 n\geq0 n ≥ 0 and objects x 0 , x 1 , … , x n x_{0},x_{1},\ldots,x_{n} x 0 , x 1 , … , x n in A \mathcal{A} A
F n x 0 , … , x n − G n x 0 , … , x n = ∑ n = n 1 + ⋯ + n k 1 ≤ l ≤ k , n i ≥ 0 ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F ( x p 0 ) , F ( x p 1 ) , … , F ( x p k ) ∘ [ H n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ F n l − 1 x p l − 2 , … , x p l − 1 ⊗ H n l x p l − 1 , … , x p l ⊗ G n l + 1 x p l , … , x p l + 1 ⊗ ⋯ ⊗ G n k x p k − 1 , … , x p k ] \mathcal{F}^{x_{0},\ldots,x_{n}}_{n}-\mathcal{G}^{x_{0},\ldots,x_{n}}_{n}=
\sum_{\substack{n=n_{1}+\cdots+n_{k}\\1\leq l\leq k,\ n_{i}\geq0}}
(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_{i}+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_{i}+1)(k-i)}
\mathfrak{m}^{\mathcal{B};\mathcal{F}(x_{p_{0}}),\mathcal{F}(x_{p_{1}}),\ldots,\mathcal{F}(x_{p_{k}})}_{k}\circ
\left[
\mathcal{H}^{x_{p_{0}},\ldots,x_{p_{1}}}_{n_{1}}\otimes\cdots\otimes\mathcal{F}^{x_{p_{l-2}},\ldots,x_{p_{l-1}}}_{n_{l-1}}\otimes\mathcal{H}^{x_{p_{l-1}},\ldots,x_{p_{l}}}_{n_{l}}\otimes\mathcal{G}^{x_{p_{l}},\ldots,x_{p_{l+1}}}_{n_{l+1}}\otimes\cdots\otimes\mathcal{G}^{x_{p_{k-1}},\ldots,x_{p_{k}}}_{n_{k}}
\right] F n x 0 , … , x n − G n x 0 , … , x n = n = n 1 + ⋯ + n k 1 ≤ l ≤ k , n i ≥ 0 ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k B ; F ( x p 0 ) , F ( x p 1 ) , … , F ( x p k ) ∘ [ H n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ F n l − 1 x p l − 2 , … , x p l − 1 ⊗ H n l x p l − 1 , … , x p l ⊗ G n l + 1 x p l , … , x p l + 1 ⊗ ⋯ ⊗ G n k x p k − 1 , … , x p k ]
Set p j = ∑ i = 1 j n i p_{j}=\sum_{i=1}^{j}n_{i} p j = ∑ i = 1 j n i , 0 ≤ j ≤ k 0\leq j\leq k 0 ≤ j ≤ k ,
+ ∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k H i + k + 1 x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , x i + 1 , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) . (152) +\sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}\mathcal{H}^{x_{0},x_{1},\ldots,x_{i},x_{i+j},x_{i+j+1},\ldots,x_{n}}_{i+k+1}\circ
\left(
\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{0},x_{1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i-1},x_{i})}\otimes\mathfrak{m}^{\mathcal{A};x_{i},x_{i+1},\ldots,x_{i+j}}_{j}
\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_{n-1},x_{n})}
\right). \tag*{(152)} + i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k H i + k + 1 x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id Hom A ( x 0 , x 1 ) ⊗ ⋯ ⊗ id Hom A ( x i − 1 , x i ) ⊗ m j A ; x i , x i + 1 , … , x i + j ⊗ id Hom A ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id Hom A ( x n − 1 , x n ) ) . (152)
If A , B , F , G \mathcal{A},\mathcal{B},\mathcal{F},\mathcal{G} A , B , F , G are strict, we call H \mathcal{H} H strict if H 0 x 0 = 0 \mathcal{H}^{x_{0}}_{0}=0 H 0 x 0 = 0 for all x 0 ∈ Obj A x_{0}\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A . Then H \mathcal{H} H is an ordinary A ∞ A_{\infty} A ∞ -homotopy of A ∞ A_{\infty} A ∞ -functors.
If F , G \mathcal{F},\mathcal{G} F , G are unital filtered A ∞ A_{\infty} A ∞ -functors of unital filtered A ∞ A_{\infty} A ∞ -categories A , B \mathcal{A},\mathcal{B} A , B , we call a filtered A ∞ A_{\infty} A ∞ -homotopy H : F ⇒ G \mathcal{H}\colon\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G unital if (4.28) holds for all i , k , a j , x l i,k,a_{j},x_{l} i , k , a j , x l .
Here is the analogue of Definition 4.22. Part (a) follows Fukaya [31, Lem.-Def. 2.17].
Definition 5.22. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, A , B \mathcal{A},\mathcal{B} A , B be filtered A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}\colon\mathcal{A}\to\mathcal{B} F , G : A → B be filtered A ∞ A_{\infty} A ∞ -functors. Use the notation of Definition 5.19.
(a) A filtered A ∞ A_{\infty} A ∞ -natural transformation T : F ⇒ G \mathcal{T}\colon\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is an element T ∈ F 0 Hom A ∞ 0 ( F , G ) \mathcal{T}\in F^{0}\operatorname{Hom}^{0}_{A_{\infty}}(\mathcal{F},\mathcal{G}) T ∈ F 0 Hom A ∞ 0 ( F , G ) with $\mathfrak{n}^{\mathcal{F},\mathcal{G}}_{1}(\mathcal{T})=0. Then [ T ] : F → G [\mathcal{T}]\colon\mathcal{F}\to\mathcal{G} [ T ] : F → G is a morphism in the (non-unital) cohomology category H 0 ( F 0 Fun A ∞ ( A , B ) ) H^0(F^0\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B})) H 0 ( F 0 Fun A ∞ ( A , B )) of the (non-unital) A ∞ A_\infty A ∞ -category F 0 Fun A ∞ ( A , B ) F^0\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ ( A , B ) over F 0 R F^0R F 0 R from Definition 5.19(f).
(b) A weakly filtered A ∞ A_\infty A ∞ -natural transformation T : F ⇒ G \mathcal{T}\colon\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is an element T ∈ Hom A ∞ 0 ( F , G ) \mathcal{T}\in\operatorname{Hom}_{A_\infty}^0(\mathcal{F},\mathcal{G}) T ∈ Hom A ∞ 0 ( F , G ) with n 1 F , G ( T ) = 0 n_1^{\mathcal{F},\mathcal{G}}(\mathcal{T})=0 n 1 F , G ( T ) = 0 .
Note that this implies that T ∈ F δ Hom A ∞ 0 ( F , G ) \mathcal{T}\in F^\delta\operatorname{Hom}_{A_\infty}^0(\mathcal{F},\mathcal{G}) T ∈ F δ Hom A ∞ 0 ( F , G ) for some δ ∈ Γ \delta\in\Gamma δ ∈ Γ , and by replacing δ \delta δ by glb ( 0 , δ ) \operatorname{glb}(0,\delta) glb ( 0 , δ ) we can suppose that δ ⪯ 0 \delta\preceq0 δ ⪯ 0 . Then each T k x 0 , … , x k \mathcal{T}^{x_0,\ldots,x_k}_k T k x 0 , … , x k in T \mathcal{T} T maps F γ ( ⋯ ) → F γ + δ ( ⋯ ) F^\gamma(\cdots)\to F^{\gamma+\delta}(\cdots) F γ ( ⋯ ) → F γ + δ ( ⋯ ) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ , as in (5.12).
Then [ T ] : F → G [\mathcal{T}]\colon\mathcal{F}\to\mathcal{G} [ T ] : F → G is a morphism in the (non-unital) cohomology category H 0 ( Fun A ∞ ( A , B ) ) H^0(\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ ( A , B )) of the (non-unital) A ∞ A_\infty A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) .
Now suppose A , B , F , G \mathcal{A},\mathcal{B},\mathcal{F},\mathcal{G} A , B , F , G are unital, so that F , G \mathcal{F},\mathcal{G} F , G are objects in the unital A ∞ A_\infty A ∞ -subcategory Fun A ∞ u n ( A , B ) ⊂ Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\subset\operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) ⊂ Fun A ∞ ( A , B ) .
(c) A unital filtered A ∞ A_\infty A ∞ -natural transformation T : F ⇒ G \mathcal{T}\colon\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is an element T ∈ F 0 Hom A ∞ u n , 0 ( F , G ) \mathcal{T}\in F^0\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\mathcal{F},\mathcal{G}) T ∈ F 0 Hom A ∞ un , 0 ( F , G ) with n 1 F , G ( T ) = 0 n_1^{\mathcal{F},\mathcal{G}}(\mathcal{T})=0 n 1 F , G ( T ) = 0 .
Then [ T ] : F → G [\mathcal{T}]\colon\mathcal{F}\to\mathcal{G} [ T ] : F → G is a morphism in the unital cohomology category H 0 ( F 0 Fun A ∞ u n ( A , B ) ) H^0(F^0\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( F 0 Fun A ∞ un ( A , B )) of the unital A ∞ A_\infty A ∞ -category F 0 Fun A ∞ u n ( A , B ) F^0\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ un ( A , B ) .
We call T \mathcal{T} T a filtered A ∞ A_\infty A ∞ -natural isomorphism if it is a homotopy equivalence in the unital A ∞ A_\infty A ∞ -category F 0 Fun A ∞ u n ( A , B ) F^0\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ un ( A , B ) over F 0 R F^0R F 0 R , in the sense of Definition 4.8, that is, [ T ] [\mathcal{T}] [ T ] is an isomorphism in H 0 ( F 0 Fun A ∞ u n ( A , B ) ) H^0(F^0\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( F 0 Fun A ∞ un ( A , B )) .
(d) A unital weakly filtered A ∞ A_\infty A ∞ -natural transformation T : F ⇒ G \mathcal{T}\colon\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is an element T ∈ Hom A ∞ u n , 0 ( F , G ) \mathcal{T}\in\operatorname{Hom}_{A_\infty}^{\mathrm{un},0}(\mathcal{F},\mathcal{G}) T ∈ Hom A ∞ un , 0 ( F , G ) with n 1 F , G ( T ) = 0 n_1^{\mathcal{F},\mathcal{G}}(\mathcal{T})=0 n 1 F , G ( T ) = 0 .
Then [ T ] : F → G [\mathcal{T}]\colon\mathcal{F}\to\mathcal{G} [ T ] : F → G is a morphism in the unital cohomology category H 0 ( Fun A ∞ u n ( A , B ) ) H^0(\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ un ( A , B )) of the unital A ∞ A_\infty A ∞ -category Fun A ∞ u n ( A , B ) \operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) .
We call T \mathcal{T} T a weakly filtered A ∞ A_\infty A ∞ -natural isomorphism if it is a homotopy equivalence in the unital A ∞ A_\infty A ∞ -category Fun A ∞ u n ( A , B ) \operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) over R R R , that is, [ T ] [\mathcal{T}] [ T ] is an isomorphism in H 0 ( Fun A ∞ u n ( A , B ) ) H^0(\operatorname{Fun}_{A_\infty}^{\mathrm{un}}(\mathcal{A},\mathcal{B})) H 0 ( Fun A ∞ un ( A , B )) .
(Weakly) filtered A ∞ A_\infty A ∞ -natural isomorphism is an equivalence relation on unital filtered A ∞ A_\infty A ∞ -functors F : A → B \mathcal{F}\colon\mathcal{A}\to\mathcal{B} F : A → B .
Remark 5.23. (a) In our definitions of filtered A ∞ A_\infty A ∞ -categories A \mathcal{A} A and A ∞ A_\infty A ∞ -functors F \mathcal{F} F , we required the A ∞ A_\infty A ∞ -maps m k x 0 , … , x k m_k^{x_0,\ldots,x_k} m k x 0 , … , x k , F k x 0 , … , x k \mathcal{F}^{x_0,\ldots,x_k}_k F k x 0 , … , x k to preserve filtrations, and to increase them when k = 0 k=0 k = 0 . This was necessary to avoid divergent sums. For the filtered versions of A ∞ A_\infty A ∞ -natural transformations and A ∞ A_\infty A ∞ -natural isomorphisms, we can relax these conditions without causing divergent sums. In Definition 5.22(a), requiring that T ∈ F 0 Hom A ∞ 0 ( F , G ) ⊂ Hom A ∞ 0 ( F , G ) \mathcal{T}\in F^0\operatorname{Hom}_{A_\infty}^0(\mathcal{F},\mathcal{G})\subset\operatorname{Hom}_{A_\infty}^0(\mathcal{F},\mathcal{G}) T ∈ F 0 Hom A ∞ 0 ( F , G ) ⊂ Hom A ∞ 0 ( F , G ) means that the maps T k x 0 , … , x k \mathcal{T}^{x_0,\ldots,x_k}_k T k x 0 , … , x k preserve filtrations. But in Definition 5.22(b), we must have T ∈ F ϵ Hom A ∞ 0 ( F , G ) \mathcal{T}\in F^\epsilon\operatorname{Hom}_{A_\infty}^0(\mathcal{F},\mathcal{G}) T ∈ F ϵ Hom A ∞ 0 ( F , G ) for some ϵ ∈ Γ \epsilon\in\Gamma ϵ ∈ Γ , and then T k x 0 , … , x k \mathcal{T}^{x_0,\ldots,x_k}_k T k x 0 , … , x k maps F γ ( ⋯ ) → F γ + ϵ ( ⋯ ) F^\gamma(\cdots)\to F^{\gamma+\epsilon}(\cdots) F γ ( ⋯ ) → F γ + ϵ ( ⋯ ) for all k , x 0 , … , x k k,x_0,\ldots,x_k k , x 0 , … , x k . So for weakly filtered A ∞ A_\infty A ∞ -natural transformations, the maps T k x 0 , … , x k \mathcal{T}^{x_0,\ldots,x_k}_k T k x 0 , … , x k need not preserve filtrations, but can decrease them by a bounded amount ϵ \epsilon ϵ .
We outline an application of weakly filtered A ∞ A_{\infty} A ∞ -natural isomorphisms in Symplectic Geometry, to justify the definition. In the ongoing project of Fukaya–Oh–Ohta–Ono, if ( X , ω ) (X,\omega) ( X , ω ) is a suitable symplectic manifold, they aim (though this may not yet be complete?) to define a filtered Fukaya category F f i ( X , ω ) \mathcal{F}^{\mathrm{fi}}(X,\omega) F fi ( X , ω ) , a filtered A ∞ A_{\infty} A ∞ -category over Λ n o v \Lambda_{\mathrm{nov}} Λ nov in Example 5.9, with objects suitable Lagrangians L L L in ( X , ω ) (X,\omega) ( X , ω ) . Although [33] is not written in terms of A ∞ A_{\infty} A ∞ -categories, the author expects that some of their results [33, Th. G(4), Th. 4.1.5, Def. 5.2.1, & Th. 5.3.65] may be interpreted as follows.
Let Φ : ( X , ω ) → ( X , ω ) \Phi:(X,\omega)\to(X,\omega) Φ : ( X , ω ) → ( X , ω ) be a symplectomorphism Hamiltonian isotopic to the identity. Then one should construct a filtered A ∞ A_{\infty} A ∞ -functor Φ ∗ : F f i ( X , ω ) → F f i ( X , ω ) \Phi_*:\mathcal{F}^{\mathrm{fi}}(X,\omega)\to\mathcal{F}^{\mathrm{fi}}(X,\omega) Φ ∗ : F fi ( X , ω ) → F fi ( X , ω ) acting on objects by L ↦ Φ ( L ) L\mapsto\Phi(L) L ↦ Φ ( L ) . We might expect there to exist a filtered A ∞ A_{\infty} A ∞ -natural isomorphism H : Id ⇒ Φ ∗ \mathcal{H}:\operatorname{Id}\Rightarrow\Phi_* H : Id ⇒ Φ ∗ . However, it seems clear from [33] that the maps H k x 0 , … , x k \mathcal{H}^{x_0,\ldots,x_k}_k H k x 0 , … , x k should not be filtration-preserving; instead, following ‘weakly filtered A ∞ A_{\infty} A ∞ -bimodule homomorphisms’ in [33, Def. 5.2.1], we should allow H k x 0 , … , x k \mathcal{H}^{x_0,\ldots,x_k}_k H k x 0 , … , x k to map F λ ( ⋯ ) ↦ F λ − ϵ ( ⋯ ) F^\lambda(\cdots)\mapsto F^{\lambda-\epsilon}(\cdots) F λ ( ⋯ ) ↦ F λ − ϵ ( ⋯ ) for some ϵ ≥ 0 \epsilon\geq0 ϵ ≥ 0 independent of k , x 0 , … , x k k,x_0,\ldots,x_k k , x 0 , … , x k . So H \mathcal{H} H should be a weakly filtered A ∞ A_{\infty} A ∞ -natural isomorphism.
(b) We do not define a ‘weak’ version of filtered A ∞ A_{\infty} A ∞ -homotopy in Definition 5.21. This is because, as in [31, Def. 13.5], filtered A ∞ A_{\infty} A ∞ -homotopies can be defined as examples of filtered A ∞ A_{\infty} A ∞ -functors, so they should satisfy the same filtration conditions as filtered A ∞ A_{\infty} A ∞ -functors.
(c) The analogues of Proposition 4.24, Theorem 4.45 , and Definition 4.28 do not make sense in the filtered case, as filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A do not have cohomology categories H 0 ( A ) H^0(\mathcal{A}) H 0 ( A ) .
Here is the analogue of Proposition 4.25.
Proposition 5.24. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, A , B \mathcal{A},\mathcal{B} A , B be unital filtered A ∞ A_{\infty} A ∞ -categories over R R R , and F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B be unital filtered A ∞ A_{\infty} A ∞ -functors which act the same way on objects, F O b j = G O b j \mathcal{F}_{\mathrm{Obj}}=\mathcal{G}_{\mathrm{Obj}} F Obj = G Obj , and H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G be a unital filtered A ∞ A_{\infty} A ∞ -homotopy. Define T ∈ Hom A ∞ f i 0 ( F , G ) \mathcal{T}\in\operatorname{Hom}^{0}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{F},\mathcal{G}) T ∈ Hom A ∞ fi 0 ( F , G ) by
T k x 0 , … , x k = { 1 B F ( x 0 ) + H 0 x 0 , k = 0 , H k x 0 , … , x k , k ≥ 1 , \mathcal{T}^{x_0,\ldots,x_k}_k=
\begin{cases}
\mathbf{1}_{\mathcal{B}}^{\mathcal{F}(x_0)}+\mathcal{H}^{x_0}_0, & k=0,\\
\mathcal{H}^{x_0,\ldots,x_k}_k, & k\geq1,
\end{cases} T k x 0 , … , x k = { 1 B F ( x 0 ) + H 0 x 0 , H k x 0 , … , x k , k = 0 , k ≥ 1 , for all k ≥ 0 k\geq0 k ≥ 0 and x 0 , … , x k ∈ Obj A x_0,\ldots,x_k\in\operatorname{Obj}\mathcal{A} x 0 , … , x k ∈ Obj A . Then T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G is a filtered A ∞ A_{\infty} A ∞ -natural isomorphism. This gives a 1-1 correspondence between unital filtered A ∞ A_{\infty} A ∞ -homotopies H : F ⇒ G \mathcal{H}:\mathcal{F}\Rightarrow\mathcal{G} H : F ⇒ G and filtered A ∞ A_{\infty} A ∞ -natural isomorphisms T : F ⇒ G \mathcal{T}:\mathcal{F}\Rightarrow\mathcal{G} T : F ⇒ G with T 0 x 0 ( 1 ) − 1 B F ( x 0 ) ∈ F γ 0 Hom B 0 ( F ( x 0 ) , F ( x 0 ) ) \mathcal{T}^{x_0}_0(1)-1_{\mathcal{B}}^{\mathcal{F}(x_0)}\in F^{\gamma_0}\operatorname{Hom}^{0}_{\mathcal{B}}(\mathcal{F}(x_0),\mathcal{F}(x_0)) T 0 x 0 ( 1 ) − 1 B F ( x 0 ) ∈ F γ 0 Hom B 0 ( F ( x 0 ) , F ( x 0 )) for all x 0 ∈ Obj A x_0\in\operatorname{Obj}\mathcal{A} x 0 ∈ Obj A .
Proof. Essentially the same proof as that of Proposition 4.25 works in the filtered case, but with the following modification. Equation (117) no longer holds, and should be replaced by
m 2 X ( x 0 ) , X ( x 0 ) , X ( x n ) ( X 0 B ( 1 ) ⊗ U n x 0 ′ , … , x n ′ + (terms in U k x 0 ′ , … , x k ′ for k < n ) \mathfrak{m}^{\mathfrak{X}(x_0),\mathfrak{X}(x_0),\mathfrak{X}(x_n)}_2\left(\mathfrak{X}^{\mathcal{B}}_0(1)\otimes\mathfrak{U}^{x'_0,\ldots,x'_n}_n + \text{(terms in } \mathfrak{U}^{x'_0,\ldots,x'_k}_k \text{ for } k<n)
\right. m 2 X ( x 0 ) , X ( x 0 ) , X ( x n ) ( X 0 B ( 1 ) ⊗ U n x 0 ′ , … , x n ′ + (terms in U k x 0 ′ , … , x k ′ for k < n )
+ (terms in U n ′ x 0 ′ , … , x n ′ ′ , n ′ ≥ n , and at least one X ~ x 0 ( 1 ) , Φ x 0 ( 1 ) , T x 0 ( 1 ) − 1 X ( x ) B ) ) = { 1 X ( x 0 ) B , n = 0 , 0 , n > 0. \left.
+ \text{(terms in } \mathfrak{U}^{x'_0,\ldots,x'_{n'}}_{n'},\ n'\geq n,\text{ and at least one } \widetilde{\mathfrak{X}}^{x_0}(1),\mathfrak{\Phi}^{x_0}(1),\mathfrak{T}^{x_0}(1)-1^{\mathcal{B}}_{\mathfrak{X}(x)}\text{)}
\right)
=
\begin{cases}
1^{\mathcal{B}}_{\mathfrak{X}(x_0)}, & n=0,\\
0, & n>0.
\end{cases} + (terms in U n ′ x 0 ′ , … , x n ′ ′ , n ′ ≥ n , and at least one X x 0 ( 1 ) , Φ x 0 ( 1 ) , T x 0 ( 1 ) − 1 X ( x ) B ) ) = { 1 X ( x 0 ) B , 0 , n = 0 , n > 0.
Since X ~ x 0 ( 1 ) , G x 0 ( 1 ) , T x 0 ( 1 ) − 1 X ( x ) B \widetilde{\mathfrak{X}}^{x_0}(1),\mathfrak{G}^{x_0}(1),\mathfrak{T}^{x_0}(1)-1^{\mathcal{B}}_{\mathfrak{X}(x)} X x 0 ( 1 ) , G x 0 ( 1 ) , T x 0 ( 1 ) − 1 X ( x ) B lie in F Γ > 0 Hom B ( X ( x ) , X ( x ) ) F^{\Gamma>0}\operatorname{Hom}_{\mathcal{B}}(\mathfrak{X}(x),\mathfrak{X}(x)) F Γ > 0 Hom B ( X ( x ) , X ( x )) , the terms on the second line increase the filtration of U n ′ x 0 ′ , … , x n ′ ′ \mathfrak{U}^{x'_0,\ldots,x'_{n'}}_{n'} U n ′ x 0 ′ , … , x n ′ ′ . We then show there is a unique U \mathfrak{U} U satisfying (4.30) by a double induction, in which we determine U n ′ x 0 ′ , … , x n ′ \mathfrak{U}^{x'_0,\ldots,x'_n}_{n'} U n ′ x 0 ′ , … , x n ′ modulo F γ ( ⋯ ) F^{\gamma}(\cdots) F γ ( ⋯ ) by induction firstly on n n n , and secondly on increasing γ \gamma γ .
Here is the analogue of Definition 4.27.
Definition 5.25. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, A , B \mathcal{A},\mathcal{B} A , B be unital filtered A ∞ A_\infty A ∞ -categories over R R R , and F ~ : A → B \widetilde{\mathfrak{F}}:\mathcal{A}\to\mathcal{B} F : A → B be a unital filtered A ∞ A_\infty A ∞ -functor. We call F ~ \widetilde{\mathfrak{F}} F a homotopy equivalence if there exists a unital filtered A ∞ A_\infty A ∞ -functor G : B → A \mathfrak{G}:\mathcal{B}\to\mathcal{A} G : B → A and filtered A ∞ A_\infty A ∞ -natural isomorphisms S : G ∘ F ~ ⇒ id A \mathfrak{S}:\mathfrak{G}\circ\widetilde{\mathfrak{F}}\Rightarrow\operatorname{id}_{\mathcal{A}} S : G ∘ F ⇒ id A and T : F ~ ∘ G ⇒ id B \mathfrak{T}:\widetilde{\mathfrak{F}}\circ\mathfrak{G}\Rightarrow\operatorname{id}_{\mathcal{B}} T : F ∘ G ⇒ id B . Then we also call A , B \mathcal{A},\mathcal{B} A , B homotopy equivalent .
We call F ~ \widetilde{\mathfrak{F}} F a weak homotopy equivalence , and A , B \mathcal{A},\mathcal{B} A , B weakly homotopy equivalent , if the same happens with S , T \mathfrak{S},\mathfrak{T} S , T weakly filtered A ∞ A_\infty A ∞ -natural isomorphisms.
Compositions of (weak) homotopy equivalences are (weak) homotopy equivalences. (Weak) homotopy equivalence is an equivalence relation on unital filtered A ∞ A_\infty A ∞ -categories.
Bounding cochains, and their applications
Bounding cochains are a method of modifying filtered A ∞ A_\infty A ∞ -algebras or filtered A ∞ A_\infty A ∞ -categories to make them strict. They were introduced by Fukaya et al. [33, §3.6] for A ∞ A_\infty A ∞ -algebras, and by Fukaya [29, Def. 1.18], [31, Def. 2.5] for A ∞ A_\infty A ∞ -categories, and are a vital part of their programme for constructing Lagrangian Floer cohomology and Fukaya categories. We first explain them following [29, 31, 33], and then in Proposition 5.29 we give the author’s perspective on them.
Definition 5.26. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and ( A , m ) (A,m) ( A , m ) be a filtered A ∞ A_\infty A ∞ -algebra over R R R . Let b ∈ F Γ > 0 A 1 b\in F^{\Gamma>0}A^1 b ∈ F Γ > 0 A 1 . Define graded R R R -linear maps m k b : A ⊗ k → A m^b_k:A^{\otimes k}\to A m k b : A ⊗ k → A for k = 0 , 1 , … k=0,1,\ldots k = 0 , 1 , … , of degree 2 − k 2-k 2 − k , by
m k b ( a 1 ⊗ ⋯ ⊗ a k ) = ∑ n 0 , … , n k ≥ 0 m k + n 0 + ⋯ + n k ( b ⊗ n 0 ⊗ a 1 ⊗ b ⊗ n 1 ⊗ a 2 ⊗ ⋯ ⊗ b ⊗ n k − 1 ⊗ a k ⊗ b ⊗ n k ) . (153) m^b_k(a_1\otimes\cdots\otimes a_k)=
\sum_{n_0,\ldots,n_k\geq0}
m_{k+n_0+\cdots+n_k}
\left(b^{\otimes n_0}\otimes a_1\otimes b^{\otimes n_1}\otimes a_2\otimes\cdots\otimes b^{\otimes n_{k-1}}\otimes a_k\otimes b^{\otimes n_k}\right). \tag*{(153)} m k b ( a 1 ⊗ ⋯ ⊗ a k ) = n 0 , … , n k ≥ 0 ∑ m k + n 0 + ⋯ + n k ( b ⊗ n 0 ⊗ a 1 ⊗ b ⊗ n 1 ⊗ a 2 ⊗ ⋯ ⊗ b ⊗ n k − 1 ⊗ a k ⊗ b ⊗ n k ) . (153)
This is an infinite sum, but converges as b ∈ F Γ > 0 A 1 b\in F^{\Gamma>0}A^1 b ∈ F Γ > 0 A 1 . It is filtration-preserving. Write m b = ( m k b ) k ≥ 0 m^b=(m^b_k)_{k\geq0} m b = ( m k b ) k ≥ 0 . Then as in [33, §3.6.1], ( A , m b = ( m k b ) k ≥ 0 ) (A,m^b=(m^b_k)_{k\geq0}) ( A , m b = ( m k b ) k ≥ 0 ) is another filtered A ∞ A_\infty A ∞ -algebra over R R R , with A ∞ A_\infty A ∞ -maps m k m_k m k ‘twisted by b b b ’.
If 1 1 1 is a unit for ( A , m ) (A,m) ( A , m ) then it is also a unit for ( A , m b ) (A,m^b) ( A , m b ) . There is a filtered A ∞ A_\infty A ∞ -isomorphism J b : ( A , m ) → ( A , m b ) \mathcal{J}^b:(A,m)\to(A,m^b) J b : ( A , m ) → ( A , m b ) given by J 0 b = − b \mathcal{J}^b_0=-b J 0 b = − b , J 1 b = id A \mathcal{J}^b_1=\operatorname{id}_A J 1 b = id A and J k b = 0 \mathcal{J}^b_k=0 J k b = 0 if k ≥ 2 k\ge2 k ≥ 2 . It has an inverse filtered A ∞ A_\infty A ∞ -isomorphism J b : ( A , m b ) → ( A , m ) \mathcal{J}^b:(A,m^b)\to(A,m) J b : ( A , m b ) → ( A , m ) given by J 0 b = b \mathcal{J}^b_0=b J 0 b = b , J 1 b = id A \mathcal{J}^b_1=\operatorname{id}_A J 1 b = id A and J k b = 0 \mathcal{J}^b_k=0 J k b = 0 if k ≥ 2 k\ge2 k ≥ 2 . We call b b b a bounding cochain for ( A , m ) (A,m) ( A , m ) if m 0 b = 0 m^b_0=0 m 0 b = 0 , i.e. if
∑ k ≥ 0 m k ( b , … , b ) = 0 in A 2 . (154) \sum_{k\ge0}m_k(b,\ldots,b)=0\quad\text{in }A^2. \tag*{(154)} k ≥ 0 ∑ m k ( b , … , b ) = 0 in A 2 . (154)
This is the Maurer–Cartan equation , or Batalin–Vilkovisky master equation . If b b b is a bounding cochain then ( A , m b ) (A,m^b) ( A , m b ) is a strict filtered A ∞ A_\infty A ∞ -algebra, and thus an A ∞ A_\infty A ∞ -algebra.
Definition 5.27. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and A \mathcal{A} A a filtered A ∞ A_\infty A ∞ -category over R R R . Suppose x x x is an object of A \mathcal{A} A . A bounding cochain b b b for x x x is an element b b b in F ≻ 0 Hom A 1 ( x , x ) F^{\succ0}\operatorname{Hom}_{\mathcal{A}}^1(x,x) F ≻ 0 Hom A 1 ( x , x ) satisfying, as in (154)
∑ k ≥ 0 m k x , … , x ( b , … , b ) = 0 in Hom A 2 ( x , x ) . (155) \sum_{k\ge0}m_k^{x,\ldots,x}(b,\ldots,b)=0\quad\text{in }\operatorname{Hom}_{\mathcal{A}}^2(x,x). \tag*{(155)} k ≥ 0 ∑ m k x , … , x ( b , … , b ) = 0 in Hom A 2 ( x , x ) . (155)
(b) We will define a strict filtered A ∞ A_\infty A ∞ -category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) over R R R called the associated strict category to A \mathcal{A} A . Objects of Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) are pairs ( x , b ) (x,b) ( x , b ) of an object x x x in A \mathcal{A} A and a bounding cochain b b b for x x x in A \mathcal{A} A . For objects ( x , b ) , ( x ′ , b ′ ) (x,b),(x',b') ( x , b ) , ( x ′ , b ′ ) we define Hom Str ( A ) ( ( x , b ) , ( x ′ , b ′ ) ) = Hom A ( x , x ′ ) \operatorname{Hom}_{\operatorname{Str}(\mathcal{A})}((x,b),(x',b'))=\operatorname{Hom}_{\mathcal{A}}(x,x') Hom Str ( A ) (( x , b ) , ( x ′ , b ′ )) = Hom A ( x , x ′ ) as a Γ \Gamma Γ -filtered Z \mathbb{Z} Z -graded R R R -module. If k ≥ 0 k\ge0 k ≥ 0 and ( x 0 , b 0 ) , … , ( x k , b k ) (x_0,b_0),\ldots,(x_k,b_k) ( x 0 , b 0 ) , … , ( x k , b k ) are objects in Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) , then generalizing (153) we define the A ∞ A_\infty A ∞ -operations n k ( x 0 , b 0 ) , … , ( x k , b k ) n_k^{(x_0,b_0),\ldots,(x_k,b_k)} n k ( x 0 , b 0 ) , … , ( x k , b k ) in Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) by
n k ( x 0 , b 0 ) , … , ( x k , b k ) ( a 1 ⊗ ⋯ ⊗ a k ) = ∑ n 0 , … , n k ≥ 0 m k + n 0 + ⋯ + n k ( n 0 + 1 ) x 0 , ( n 1 + 1 ) x 1 , … , ( n k + 1 ) x k ( b 0 ⊗ n 0 ⊗ a 1 ⊗ b 1 ⊗ n 1 ⊗ a 2 ⊗ ⋯ ⊗ a k ⊗ b k ⊗ n k ) . (156) n_k^{(x_0,b_0),\ldots,(x_k,b_k)}(a_1\otimes\cdots\otimes a_k)=
\sum_{n_0,\ldots,n_k\ge0}
m_{k+n_0+\cdots+n_k}^{(n_0+1)x_0,(n_1+1)x_1,\ldots,(n_k+1)x_k}
\left(b_0^{\otimes n_0}\otimes a_1\otimes b_1^{\otimes n_1}\otimes a_2\otimes\cdots\otimes a_k\otimes b_k^{\otimes n_k}\right). \tag*{(156)} n k ( x 0 , b 0 ) , … , ( x k , b k ) ( a 1 ⊗ ⋯ ⊗ a k ) = n 0 , … , n k ≥ 0 ∑ m k + n 0 + ⋯ + n k ( n 0 + 1 ) x 0 , ( n 1 + 1 ) x 1 , … , ( n k + 1 ) x k ( b 0 ⊗ n 0 ⊗ a 1 ⊗ b 1 ⊗ n 1 ⊗ a 2 ⊗ ⋯ ⊗ a k ⊗ b k ⊗ n k ) . (156)
Here by ( n 0 + 1 ) x 0 , ( n 1 + 1 ) x 1 , … (n_0+1)x_0,(n_1+1)x_1,\ldots ( n 0 + 1 ) x 0 , ( n 1 + 1 ) x 1 , … we mean the list with x 0 x_0 x 0 repeated n 0 + 1 n_0+1 n 0 + 1 times, followed by x 1 x_1 x 1 repeated n 1 + 1 n_1+1 n 1 + 1 times, and so on. Equation (156) is an infinite sum in Hom Str ( A ) ( ( x 0 , b 0 ) , ( x k , b k ) ) = Hom A ( x 0 , x k ) \operatorname{Hom}_{\operatorname{Str}(\mathcal{A})}((x_0,b_0),(x_k,b_k))=\operatorname{Hom}_{\mathcal{A}}(x_0,x_k) Hom Str ( A ) (( x 0 , b 0 ) , ( x k , b k )) = Hom A ( x 0 , x k ) , which converges as b i ∈ F ≻ 0 Hom A 1 ( x i , x i ) b_i\in F^{\succ0}\operatorname{Hom}_{\mathcal{A}}^1(x_i,x_i) b i ∈ F ≻ 0 Hom A 1 ( x i , x i ) . As in Fukaya [29, Def. 1.18], [31, Def. 2.5], this defines a filtered A ∞ A_\infty A ∞ -category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) over R R R .
(c) If A \mathcal{A} A is a unital filtered A ∞ A_\infty A ∞ -category over R R R , with units 1 x A ∈ Hom A 0 ( x , x ) 1_x^{\mathcal{A}}\in\operatorname{Hom}_{\mathcal{A}}^0(x,x) 1 x A ∈ Hom A 0 ( x , x ) , then Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) is also unital, with units 1 ( x , b ) Str ( A ) = 1 x A 1_{(x,b)}^{\operatorname{Str}(\mathcal{A})}=1_x^{\mathcal{A}} 1 ( x , b ) Str ( A ) = 1 x A .
Remark 5.28. Here is how bounding cochains and associated strict categories are used in Fukaya–Oh–Ohta–Ono’s programme. Suppose ( X , ω ) (X,\omega) ( X , ω ) is a symplectic manifold, satisfying some assumptions. They perform the following steps:
(i) For the po-monoid ( Γ , ⪯ ) = ( R , ≤ ) (\Gamma,\preceq)=(\mathbb{R},\le) ( Γ , ⪯ ) = ( R , ≤ ) and R \mathbb{R} R -filtered ring R = Λ n o v Q R=\Lambda_{\mathrm{nov}}^{\mathbb{Q}} R = Λ nov Q from Example 5.9, after making some additional choices such as an almost complex structure J J J on X X X , they construct a Λ n o v Q \Lambda_{\mathrm{nov}}^{\mathbb{Q}} Λ nov Q -filtered A ∞ A_\infty A ∞ -category F ( X , ω ) \mathcal{F}(X,\omega) F ( X , ω ) , whose objects are suitable Lagrangians L L L in ( X , ω ) (X,\omega) ( X , ω ) . This construction is very long and difficult, with many challenging analytic issues. Most of the work is done in [33], although this is written in terms of filtered A ∞ A_\infty A ∞ -algebras rather than filtered A ∞ A_\infty A ∞ -categories. (ii) They pass to the associated strict category Str ( F ( X , ω ) ) \operatorname{Str}(\mathcal{F}(X,\omega)) Str ( F ( X , ω )) , an A ∞ A_\infty A ∞ -category over Λ 0 n o v \Lambda^{\mathrm{nov}}_0 Λ 0 nov , whose objects are pairs ( L , b ) (L,b) ( L , b ) of a Lagrangian L L L in ( X , ω ) (X,\omega) ( X , ω ) and a bounding cochain b b b for the filtered A ∞ A_\infty A ∞ -algebra of L L L .
(iii) They apply the ‘twist construction’ Tw ∘ Str ( F ( X , ω ) ) \operatorname{Tw}\circ\operatorname{Str}(\mathcal{F}(X,\omega)) Tw ∘ Str ( F ( X , ω )) from §4.9 to get a pretriangulated A ∞ A_\infty A ∞ -category.
(iv) The derived Fukaya category is D b F ( X , ω ) = H 0 ( Tw ∘ Str ( F ( X , ω ) ) ) D^b\mathcal{F}(X,\omega)=H^0(\operatorname{Tw}\circ\operatorname{Str}(\mathcal{F}(X,\omega))) D b F ( X , ω ) = H 0 ( Tw ∘ Str ( F ( X , ω ))) . It is a triangulated category, of great interest in Symplectic Geometry. For example, Kontsevich’s Homological Mirror Symmetry Conjecture [53] explains Mirror Symmetry of Calabi–Yau manifolds X , X ˇ X,\check{X} X , X ˇ in terms of equivalences of triangulated categories D b F ( X , ω ) ≃ D b coh X ˇ D^b\mathcal{F}(X,\omega)\simeq D^b\operatorname{coh}\check{X} D b F ( X , ω ) ≃ D b coh X ˇ , D b F ( X ˇ , ω ˇ ) ≃ D b coh X D^b\mathcal{F}(\check{X},\check{\omega})\simeq D^b\operatorname{coh}X D b F ( X ˇ , ω ˇ ) ≃ D b coh X .
We will now interpret associated strict categories in terms of filtered A ∞ A_\infty A ∞ -functor categories in a way which seems very natural, but the author has not found it in the literature, even in Fukaya [31] which defines both ideas. Our point of view will make it easy to generalize associated strict categories to the ∞ \infty ∞ -categorical setting. Recall that Definition 4.15 defined the point A ∞ A_\infty A ∞ -category P \mathcal{P} P and Proposition 4.16 showed that Fun A ∞ ( P , A ) ≅ A \operatorname{Fun}_{A_\infty}(\mathcal{P},\mathcal{A})\cong\mathcal{A} Fun A ∞ ( P , A ) ≅ A for any A ∞ A_\infty A ∞ -category A \mathcal{A} A , and Definition 4.17 and Proposition 4.18 did the same for unital A ∞ A_\infty A ∞ -categories. In the filtered case, instead we find that Fun A ∞ f i ( P , A ) ≅ Str ( A ) \operatorname{Fun}_{A_\infty^{\mathrm{fi}}}(\mathcal{P},\mathcal{A})\cong\operatorname{Str}(\mathcal{A}) Fun A ∞ fi ( P , A ) ≅ Str ( A ) .
Proposition 5.29. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and A \mathcal{A} A be a filtered A ∞ A_\infty A ∞ -category over R R R . Let P \mathcal{P} P and P u n \mathcal{P}^{\mathrm{un}} P un be the non-unital and unital point A ∞ A_\infty A ∞ -categories over R R R as in Definitions 4.15 and 4.17, made into strict filtered A ∞ A_\infty A ∞ -categories in the obvious way.
(a) A filtered A ∞ A_\infty A ∞ -functor F : P → A \mathcal{F}:\mathcal{P}\to\mathcal{A} F : P → A is equivalent to a pair ( x , b ) (x,b) ( x , b ) where x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A and b b b is a bounding cochain for x x x in A \mathcal{A} A .
(b) The strict filtered A ∞ A_\infty A ∞ -category Fun A ∞ f i ( P , A ) \operatorname{Fun}_{A_\infty^{\mathrm{fi}}}(\mathcal{P},\mathcal{A}) Fun A ∞ fi ( P , A ) from Definition 5.19 may be identified with the associated strict category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) from Definition 5.27(b). Define a functor Str = Fun A ∞ f i ( P , − ) : A ∞ f i Cat R → A ∞ f i Cat R , ∞ \operatorname{Str}=\operatorname{Fun}_{A_\infty^{\mathrm{fi}}}(\mathcal{P},-):A_\infty^{\mathrm{fi}}\operatorname{Cat}_{R}\to A_\infty^{\mathrm{fi}}\operatorname{Cat}_{R,\infty} Str = Fun A ∞ fi ( P , − ) : A ∞ fi Cat R → A ∞ fi Cat R , ∞ . This makes associated strict A ∞ A_\infty A ∞ -categories functorial.
(c) If A \mathcal{A} A is unital then a unital filtered A ∞ A_\infty A ∞ -functor F : P u n → A \mathcal{F}:\mathcal{P}^{\mathrm{un}}\to\mathcal{A} F : P un → A is equivalent to a pair ( x , b ) (x,b) ( x , b ) where x ∈ Obj A x\in\operatorname{Obj}\mathcal{A} x ∈ Obj A and b b b is a bounding cochain for x x x in A \mathcal{A} A .
(d) If A \mathcal{A} A is unital then the unital strict filtered A ∞ A_\infty A ∞ -category Fun A ∞ f i u n ( P u n , A ) \operatorname{Fun}^{\mathrm{un}}_{A_\infty^{\mathrm{fi}}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) Fun A ∞ fi un ( P un , A ) from Definition 5.19 may be identified with the associated strict category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) from Definition 5.27(b). The restriction of the functor Str \operatorname{Str} Str in (b) to A ∞ f i Cat u n A_\infty^{\mathrm{fi}}\operatorname{Cat}^{\mathrm{un}} A ∞ fi Cat un is naturally isomorphic to Fun A ∞ f i u n ( P u n , − ) : A ∞ f i Cat R u n → A ∞ f i Cat R , ∞ u n \operatorname{Fun}^{\mathrm{un}}_{A_\infty^{\mathrm{fi}}}(\mathcal{P}^{\mathrm{un}},-):A_\infty^{\mathrm{fi}}\operatorname{Cat}^{\mathrm{un}}_{R}\to A_\infty^{\mathrm{fi}}\operatorname{Cat}^{\mathrm{un}}_{R,\infty} Fun A ∞ fi un ( P un , − ) : A ∞ fi Cat R un → A ∞ fi Cat R , ∞ un .
Proof. A filtered A ∞ A_\infty A ∞ -functor F : P → A \mathcal{F}:\mathcal{P}\to\mathcal{A} F : P → A has a map F O b j : Obj P = { ∗ } → Obj A \mathcal{F}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{P}=\{\ast\}\to\operatorname{Obj}\mathcal{A} F Obj : Obj P = { ∗ } → Obj A , and so determines an object x = F ( ∗ ) x=\mathcal{F}(\ast) x = F ( ∗ ) in A \mathcal{A} A . For the maps F ∗ , … , ∗ k + 1 \mathcal{F}^{k+1}_{\ast,\ldots,\ast} F ∗ , … , ∗ k + 1 for k ≥ 0 k\geq0 k ≥ 0 , we see that F ∗ , … , ∗ k + 1 = 0 \mathcal{F}^{k+1}_{\ast,\ldots,\ast}=0 F ∗ , … , ∗ k + 1 = 0 for k ≥ 1 k\geq1 k ≥ 1 as Hom P ( ∗ , ∗ ) = 0 \operatorname{Hom}_{\mathcal{P}}(\ast,\ast)=0 Hom P ( ∗ , ∗ ) = 0 , and F ∗ 1 : R → Hom A ( x , x ) \mathcal{F}^{1}_{\ast}:R\to\operatorname{Hom}_{\mathcal{A}}(x,x) F ∗ 1 : R → Hom A ( x , x ) is equivalent to a choice of F ∗ 1 ( 1 ) \mathcal{F}^{1}_{\ast}(1) F ∗ 1 ( 1 ) in Hom A ( x , x ) \operatorname{Hom}_{\mathcal{A}}(x,x) Hom A ( x , x ) . By definition b ∈ F 1 Hom A ( x , x ) b\in\mathcal{F}^{1}\operatorname{Hom}_{\mathcal{A}}(x,x) b ∈ F 1 Hom A ( x , x ) , and equation (5.9) for n = 0 n=0 n = 0 is equivalent to (5.16). Thus b b b is a bounding cochain, and (a) follows. Part (b) is then easy to check.
For the unital case, let F : P u n → A \mathcal{F}:\mathcal{P}^{\mathrm{un}}\to\mathcal{A} F : P un → A be a unital filtered A ∞ A_\infty A ∞ -functor, and set x = F ( ∗ ) x=\mathcal{F}(\ast) x = F ( ∗ ) in Obj A \operatorname{Obj}\mathcal{A} Obj A . Since Hom P u n ( ∗ , ∗ ) = R ⋅ 1 ∗ \operatorname{Hom}_{\mathcal{P}^{\mathrm{un}}}(\ast,\ast)=R\cdot1_{\ast} Hom P un ( ∗ , ∗ ) = R ⋅ 1 ∗ , equations (4.11)–(4.12) imply that F 1 , ∗ \mathcal{F}^{1,*} F 1 , ∗ maps λ 1 P u n ↦ λ 1 A \lambda1_{\mathcal{P}^{un}} \mapsto\lambda1_{\mathcal{A}} λ 1 P u n ↦ λ 1 A for λ ∈ R \lambda\in R λ ∈ R and F k , ∗ … , ∗ = 0 \mathcal{F}^{k,*\ldots,*}=0 F k , ∗ … , ∗ = 0 for k ≥ 2 k\geq2 k ≥ 2 . Again we set b = F 0 , ∗ ( 1 ) b=\mathcal{F}^{0,*}(1) b = F 0 , ∗ ( 1 ) in F ▹ 0 Hom A 1 ( x , x ) F^{\triangleright0}\operatorname{Hom}^{1}_{\mathcal{A}}(x,x) F ▹ 0 Hom A 1 ( x , x ) . Then (5.9) for n = 0 n=0 n = 0 is equivalent to (5.16), so b b b is a bounding cochain. We can show that (5.9) for n ≥ 1 n\geq1 n ≥ 1 holds automatically, and impose no extra conditions on ( x , b ) (x,b) ( x , b ) . Part (c) follows, and (d) is then easy to check. □
Explicit definition of the twist functor
Let R R R be a commutative ring. As in Definition 4.62, the twist functor
Tw : A ∞ Cat R ⟶ A ∞ △ Cat R , (157) \operatorname{Tw}: A_{\infty}\operatorname{Cat}_{R}\longrightarrow A_{\infty}^{\triangle}\operatorname{Cat}_{R}, \tag*{(157)} Tw : A ∞ Cat R ⟶ A ∞ △ Cat R , (157)
is defined in Seidel [84, §3], Keller [51, §7.6], Lefèvre-Hasegawa [57, §7.2], and Bespalov–Lyubashenko–Manzyuk [4, §12]. We will describe a construction of Tw \operatorname{Tw} Tw which passes through Z \mathbb{Z} Z -filtered A ∞ A_{\infty} A ∞ -categories, and uses the functor A ↦ Fun A ∞ f i ( P , A ) \mathcal{A}\mapsto\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{P},\mathcal{A}) A ↦ Fun A ∞ fi ( P , A ) defining associated strict categories in Proposition 5.29(a),(b) as an intermediate step. This will be easy to extend to ∞ \infty ∞ -categories in Chapter 9.
Definition 5.30. Let R R R be a commutative ring. Make R R R into a complete filtered ring over the po-monoid ( Z , ≤ ) (\mathbb{Z},\leq) ( Z , ≤ ) by setting F λ R = R F^{\lambda}R=R F λ R = R if λ ≤ 0 \lambda\leq0 λ ≤ 0 in Z \mathbb{Z} Z and F λ R = 0 F^{\lambda}R=0 F λ R = 0 if λ > 0 \lambda>0 λ > 0 in Z \mathbb{Z} Z . We will define Tw \operatorname{Tw} Tw in (5.18) as the composition of three functors
A ∞ Cat R → Σ A ∞ f i Cat R , ∞ Tw ↓ ↓ Fun A ∞ f i ( P , − ) A ∞ Cat R ← forget filtrations A ∞ f i Cat R , ∞ (158) \begin{CD}
A_{\infty}\operatorname{Cat}_{R} @>{\Sigma}>> A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}\\
@V{\operatorname{Tw}}VV @VV{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{P},-)}V\\
A_{\infty}\operatorname{Cat}_{R} @<<{\text{forget filtrations}}< A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty} \tag*{(158)}
\end{CD} A ∞ Cat R Tw ↓ ⏐ A ∞ Cat R Σ forget filtrations A ∞ fi Cat R , ∞ ↓ ⏐ Fun A ∞ fi ( P , − ) A ∞ fi Cat R , ∞ (158)
The restriction of Tw \operatorname{Tw} Tw to A ∞ Cat R u n A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un in (4.43) may equivalently be defined as the composition
A ∞ Cat R u n → Σ A ∞ f i Cat R , ∞ u n Tw ↓ ↓ Fun A ∞ f i u n ( P u n , − ) A ∞ Cat R u n ← forget filtrations A ∞ f i Cat R , ∞ u n (159) \begin{CD}
A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} @>{\Sigma}>> A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}^{\mathrm{un}}\\
@V{\operatorname{Tw}}VV @VV{\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},-)}V\\
A_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} @<<{\text{forget filtrations}}< A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}^{\mathrm{un}} \tag*{(159)}
\end{CD} A ∞ Cat R un Tw ↓ ⏐ A ∞ Cat R un Σ forget filtrations A ∞ fi Cat R , ∞ un ↓ ⏐ Fun A ∞ fi un ( P un , − ) A ∞ fi Cat R , ∞ un (159)
where P u n \mathcal{P}^{\mathrm{un}} P un is as in Definition 4.17 .
Here for each A \mathcal{A} A in A ∞ Cat R A_{\infty}\operatorname{Cat}_{R} A ∞ Cat R we define Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) in A ∞ f i Cat R A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R} A ∞ fi Cat R as follows:
(a) Objects of Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) are triples ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) ((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) where l ≥ 0 l\geq0 l ≥ 0 , and x 1 , … , x l ∈ Obj A x_{1},\ldots,x_{l}\in\operatorname{Obj}\mathcal{A} x 1 , … , x l ∈ Obj A , and s i , f i ∈ Z s_{i},f_{i}\in\mathbb{Z} s i , f i ∈ Z for i = 1 , … , l i=1,\ldots,l i = 1 , … , l . We call s i s_{i} s i the shift of x i x_{i} x i and f i f_{i} f i the filtration weight of x i x_{i} x i . As a shorthand we write ( x , s , f ) = ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) (x,s,f)=((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) ( x , s , f ) = (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) .
(b) Let ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) ((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) and ( ( x 1 ′ , … , x l ′ ′ ) , ( s 1 ′ , … , s l ′ ′ ) , ( f 1 ′ , … , f l ′ ′ ) ) ((x'_{1},\ldots,x'_{l'}),(s'_{1},\ldots,s'_{l'}),(f'_{1},\ldots,f'_{l'})) (( x 1 ′ , … , x l ′ ′ ) , ( s 1 ′ , … , s l ′ ′ ) , ( f 1 ′ , … , f l ′ ′ )) be objects of Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) . Define a graded R R R -module
Hom Σ ( A ) ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) = ⨁ d ∈ Z Hom Σ ( A ) d ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) by \operatorname{Hom}_{\Sigma(\mathcal{A})}((x,s,f),(x',s',f'))=\bigoplus_{d\in\mathbb{Z}}\operatorname{Hom}^{d}_{\Sigma(\mathcal{A})}((x,s,f),(x',s',f'))\quad\text{by} Hom Σ ( A ) (( x , s , f ) , ( x ′ , s ′ , f ′ )) = d ∈ Z ⨁ Hom Σ ( A ) d (( x , s , f ) , ( x ′ , s ′ , f ′ )) by
Hom Σ ( A ) d ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) = ⨁ i = 1 , … , l j = 1 , … , l ′ Hom A d + s i − s j ′ ( x i , x j ′ ) . \operatorname{Hom}^{d}_{\Sigma(\mathcal{A})}((x,s,f),(x',s',f'))=\bigoplus_{\substack{i=1,\ldots,l\\j=1,\ldots,l'}}\operatorname{Hom}^{d+s_{i}-s'_{j}}_{\mathcal{A}}(x_{i},x'_{j}). Hom Σ ( A ) d (( x , s , f ) , ( x ′ , s ′ , f ′ )) = i = 1 , … , l j = 1 , … , l ′ ⨁ Hom A d + s i − s j ′ ( x i , x j ′ ) .
We write elements of Hom Σ ( A ) ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) \operatorname{Hom}_{\Sigma(\mathcal{A})}((x,s,\mathbf{f}),(x',s',\mathbf{f}')) Hom Σ ( A ) (( x , s , f ) , ( x ′ , s ′ , f ′ )) as ( T i j ) i = 1 , … , l j = 1 , … , l ′ (T_{ij})^{j=1,\ldots,l'}_{i=1,\ldots,l} ( T ij ) i = 1 , … , l j = 1 , … , l ′ in the obvious way. Define a Z \mathbb{Z} Z -filtration on Hom Σ ( A ) d ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) \operatorname{Hom}^{d}_{\Sigma(\mathcal{A})}((x,s,\mathbf{f}),(x',s',\mathbf{f}')) Hom Σ ( A ) d (( x , s , f ) , ( x ′ , s ′ , f ′ )) by, for λ ∈ Z \lambda\in\mathbb{Z} λ ∈ Z
F λ Hom Σ ( A ) d ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) = ⨁ i = 1 , … , l , j = 1 , … , l ′ f i − f j ′ ≥ λ Hom A d + s i − s j ′ ( x i , x j ′ ) . F^{\lambda}\operatorname{Hom}^{d}_{\Sigma(\mathcal{A})}((x,s,\mathbf{f}),(x',s',\mathbf{f}'))=\bigoplus_{\substack{i=1,\ldots,l,\ j=1,\ldots,l'\\ f_i-f_j'\geq\lambda}}\operatorname{Hom}^{d+s_i-s_j'}_{\mathcal{A}}(x_i,x_j'). F λ Hom Σ ( A ) d (( x , s , f ) , ( x ′ , s ′ , f ′ )) = i = 1 , … , l , j = 1 , … , l ′ f i − f j ′ ≥ λ ⨁ Hom A d + s i − s j ′ ( x i , x j ′ ) .
(c) Let ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) (x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) be objects of Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) . For k ≥ 0 k\geq0 k ≥ 0 , define a graded R R R -linear map of degree 2 − k 2-k 2 − k
n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : Hom Σ ( A ) ( ( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 ) ) ⊗ R ⋯ ⊗ R Hom Σ ( A ) ( ( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k ) ) ⟶ Hom Σ ( A ) ( ( x 0 , s 0 , f 0 ) , ( x k , s k , f k ) ) n_k^{(x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k)}:\operatorname{Hom}_{\Sigma(\mathcal{A})}((x^0,s^0,\mathbf{f}^0),(x^1,s^1,\mathbf{f}^1))\otimes_R\cdots\otimes_R\operatorname{Hom}_{\Sigma(\mathcal{A})}((x^{k-1},s^{k-1},\mathbf{f}^{k-1}),(x^k,s^k,\mathbf{f}^k))
\longrightarrow\operatorname{Hom}_{\Sigma(\mathcal{A})}((x^0,s^0,\mathbf{f}^0),(x^k,s^k,\mathbf{f}^k)) n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : Hom Σ ( A ) (( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 )) ⊗ R ⋯ ⊗ R Hom Σ ( A ) (( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k )) ⟶ Hom Σ ( A ) (( x 0 , s 0 , f 0 ) , ( x k , s k , f k ))
by n 0 ( x 0 , s 0 , f 0 ) = 0 n_0^{(x^0,s^0,\mathbf{f}^0)}=0 n 0 ( x 0 , s 0 , f 0 ) = 0 for k = 0 k=0 k = 0 , and for k ≥ 1 k\geq1 k ≥ 1 define, for j 0 = 1 , … , l 0 j^0=1,\ldots,l^0 j 0 = 1 , … , l 0 and j k = 1 , … , l k j^k=1,\ldots,l^k j k = 1 , … , l k
( n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 1 = 1 , … , l 1 i 0 = 1 , … , l 0 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k = 1 , … , l k i k − 1 = 1 , … , l k − 1 ) ) j 0 j k = ∑ j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 m k A ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) . \left(n_k^{(x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k)}\left((T^1_{i_0i_1})_{\substack{i^1=1,\ldots,l^1\\i^0=1,\ldots,l^0}}\otimes\cdots\otimes(T^k_{i_{k-1}i_k})_{\substack{i^k=1,\ldots,l^k\\i^{k-1}=1,\ldots,l^{k-1}}}\right)\right)_{j^0j^k}
=
\sum_{\substack{j^1=1,\ldots,l^1,\ldots,\\j^{k-1}=1,\ldots,l^{k-1}}}
m_k^{\mathcal{A};x^0_{j^0},x^1_{j^1},\ldots,x^{k-1}_{j^{k-1}},x^k_{j^k}}(T^1_{j^0j^1}\otimes\cdots\otimes T^k_{j^{k-1}j^k}). ( n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 1 = 1 , … , l 1 i 0 = 1 , … , l 0 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k = 1 , … , l k i k − 1 = 1 , … , l k − 1 ) ) j 0 j k = j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 ∑ m k A ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) .
It is now easy to show that this data makes Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) into a filtered A ∞ A_\infty A ∞ -category over R R R , which is strict as m 0 ( x , s , f ) = 0 m_0^{(x,s,\mathbf{f})}=0 m 0 ( x , s , f ) = 0 for all ( x , s , f ) (x,s,\mathbf{f}) ( x , s , f ) . In particular, equation (4.1) for the m k x 0 , … , x k m_k^{x^0,\ldots,x^k} m k x 0 , … , x k implies (5.8) for the n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) n_k^{(x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k)} n k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) .
Next let F : A → B \mathfrak{F}:\mathcal{A}\to\mathcal{B} F : A → B be a morphism in A ∞ Cat R A_\infty\operatorname{Cat}_R A ∞ Cat R . Define a filtered A ∞ A_\infty A ∞ -functor Σ ( F ) : Σ ( A ) → Σ ( B ) \Sigma(\mathfrak{F}):\Sigma(\mathcal{A})\to\Sigma(\mathcal{B}) Σ ( F ) : Σ ( A ) → Σ ( B ) by:
(d) Σ ( F ) \Sigma(\mathfrak{F}) Σ ( F ) acts on objects by ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ↦ ( ( F ( x 1 ) , … , F ( x l ) ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) (x_1,\ldots,x_l),(s_1,\ldots,s_l),(f_1,\ldots,f_l)\mapsto((\mathfrak{F}(x_1),\ldots,\mathfrak{F}(x_l)),(s_1,\ldots,s_l),(f_1,\ldots,f_l)) ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ↦ (( F ( x 1 ) , … , F ( x l )) , ( s 1 , … , s l ) , ( f 1 , … , f l )) .
(e) Let ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) (x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) be objects of Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) . For k ≥ 0 k\geq0 k ≥ 0 , define a graded R R R -linear map of degree 1 − k 1-k 1 − k
Σ ( F k ) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : Hom Σ ( A ) ( ( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 ) ) ⊗ R ⋯ ⊗ R Hom Σ ( A ) ( ( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k ) ) ⟶ Hom Σ ( B ) ( ( F ( x 0 ) , s 0 , f 0 ) , ( F ( x k ) , s k , f k ) ) \Sigma(\mathfrak{F}_k)^{(x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k)}:\operatorname{Hom}_{\Sigma(\mathcal{A})}((x^0,s^0,\mathbf{f}^0),(x^1,s^1,\mathbf{f}^1))\otimes_R\cdots\otimes_R\operatorname{Hom}_{\Sigma(\mathcal{A})}((x^{k-1},s^{k-1},\mathbf{f}^{k-1}),(x^k,s^k,\mathbf{f}^k))
\longrightarrow\operatorname{Hom}_{\Sigma(\mathcal{B})}((\mathfrak{F}(x^0),s^0,\mathbf{f}^0),(\mathfrak{F}(x^k),s^k,\mathbf{f}^k)) Σ ( F k ) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : Hom Σ ( A ) (( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 )) ⊗ R ⋯ ⊗ R Hom Σ ( A ) (( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k )) ⟶ Hom Σ ( B ) (( F ( x 0 ) , s 0 , f 0 ) , ( F ( x k ) , s k , f k ))
by Σ ( F 0 ) ( x 0 , s 0 , f 0 ) = 0 \Sigma(\mathfrak{F}_0)^{(x^0,s^0,\mathbf{f}^0)}=0 Σ ( F 0 ) ( x 0 , s 0 , f 0 ) = 0 for k = 0 k=0 k = 0 , and for k ≥ 1 k\geq1 k ≥ 1 define
( Σ ( F k ) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 1 = 1 , … , l 1 i 0 = 1 , … , l 0 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k = 1 , … , l k i k − 1 = 1 , … , l k − 1 ) ) j 0 j k = ∑ j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 F k A ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) , \left(\Sigma(\mathfrak{F}_k)^{(x^0,s^0,\mathbf{f}^0),\ldots,(x^k,s^k,\mathbf{f}^k)}\left((T^1_{i_0i_1})_{\substack{i^1=1,\ldots,l^1\\i^0=1,\ldots,l^0}}\otimes\cdots\otimes(T^k_{i_{k-1}i_k})_{\substack{i^k=1,\ldots,l^k\\i^{k-1}=1,\ldots,l^{k-1}}}\right)\right)_{j^0j^k}
=
\sum_{\substack{j^1=1,\ldots,l^1,\ldots,\\j^{k-1}=1,\ldots,l^{k-1}}}
\mathfrak{F}^{\mathcal{A};x^0_{j^0},x^1_{j^1},\ldots,x^{k-1}_{j^{k-1}},x^k_{j^k}}_k(T^1_{j^0j^1}\otimes\cdots\otimes T^k_{j^{k-1}j^k}), ( Σ ( F k ) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 1 = 1 , … , l 1 i 0 = 1 , … , l 0 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k = 1 , … , l k i k − 1 = 1 , … , l k − 1 ) ) j 0 j k = j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 ∑ F k A ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) ,
for j 0 = 1 , … , l 0 j^0=1,\ldots,l^0 j 0 = 1 , … , l 0 and j k = 1 , … , l k j^k=1,\ldots,l^k j k = 1 , … , l k . It is now easy to show that this data makes Σ ( F ) \Sigma(\mathcal{F}) Σ ( F ) into a filtered A ∞ A_{\infty} A ∞ -functor over R R R , which is strict as Σ ( F 0 ) ( x 0 , s 0 , f 0 ) = 0 \Sigma(\mathcal{F}_0)_{(x^0,s^0,f^0)}=0 Σ ( F 0 ) ( x 0 , s 0 , f 0 ) = 0 for all ( x 0 , s 0 , f 0 ) (x^0,s^0,f^0) ( x 0 , s 0 , f 0 ) . In particular, equation (94) for the F k x 0 , … , x k \mathcal{F}^{x_0,\ldots,x_k}_k F k x 0 , … , x k implies equation (148) for the Σ ( F ) k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) \Sigma(\mathcal{F})^{(x^0,s^0,f^0),\ldots,(x^k,s^k,f^k)}_k Σ ( F ) k ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) .
As F ↦ Σ ( F ) \mathcal{F}\mapsto\Sigma(\mathcal{F}) F ↦ Σ ( F ) is compatible with compositions and identities, this defines a functor Σ : A ∞ Cat R → A ∞ f i Cat R , ∞ \Sigma:A_{\infty}\operatorname{Cat}_R\to A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty} Σ : A ∞ Cat R → A ∞ fi Cat R , ∞ , where the target is A ∞ f i Cat R , ∞ ⊂ A ∞ f i Cat R A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,\infty}\subset A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_R A ∞ fi Cat R , ∞ ⊂ A ∞ fi Cat R as Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) and Σ ( F ) \Sigma(\mathcal{F}) Σ ( F ) are both strict.
The right hand functor Fun A ∞ f i ( P , − ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}(\mathcal{P},-) Fun A ∞ fi ( P , − ) in (158) is as in Proposition 5.29, where P \mathcal{P} P is the non-unital point A ∞ A_{\infty} A ∞ -category over R R R from Definition 4.15 , made into a Z \mathbb{Z} Z -filtered A ∞ A_{\infty} A ∞ -category over R R R in the unique way. Note that this functor maps A ↦ Str ( A ) \mathcal{A}\mapsto\operatorname{Str}(\mathcal{A}) A ↦ Str ( A ) , where Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) is the associated strict category from Definition 5.27(b). The bottom functor in (156) forgets filtration data, as in the vertical functor in (150) . This completes (156) , and the definition of Tw \operatorname{Tw} Tw .
Objects of Tw A \operatorname{Tw}\mathcal{A} Tw A , called twisted complexes , are pairs ( x , s , f ) , b (x,s,f),b ( x , s , f ) , b where ( x , s , f ) (x,s,f) ( x , s , f ) is an object in Σ ( A ) \Sigma(\mathcal{A}) Σ ( A ) and b b b is a bounding cochain for ( x , s , f ) (x,s,f) ( x , s , f ) . Explicitly, b = ( b i j ) j = 1 , … , l i = 1 , … , l b=(b_{ij})_{\substack{j=1,\ldots,l\\i=1,\ldots,l}} b = ( b ij ) j = 1 , … , l i = 1 , … , l with b i j ∈ Hom A 1 + s i − s j ( x i , x j ) b_{ij}\in\operatorname{Hom}_{\mathcal{A}}^{1+s_i-s_j}(x_i,x_j) b ij ∈ Hom A 1 + s i − s j ( x i , x j ) , where the condition that b ∈ F > 0 ( ( x , s , f ) , ( x , s , f ) ) b\in F^{>0}((x,s,f),(x,s,f)) b ∈ F > 0 (( x , s , f ) , ( x , s , f )) means that b i j = 0 b_{ij}=0 b ij = 0 if f i − f j ≤ 0 f_i-f_j\leq0 f i − f j ≤ 0 . This b b b must satisfy (155) , which has only finitely many nonzero terms.
The natural transformation inc : Id ⇒ Tw \operatorname{inc}:\operatorname{Id}\Rightarrow\operatorname{Tw} inc : Id ⇒ Tw in Definition 4.62 is defined as follows: for each A \mathcal{A} A in A ∞ Cat R A_{\infty}\operatorname{Cat}_R A ∞ Cat R , the A ∞ A_{\infty} A ∞ -functor inc A : A ↪ Tw A \operatorname{inc}_{\mathcal{A}}:\mathcal{A}\hookrightarrow\operatorname{Tw}\mathcal{A} inc A : A ↪ Tw A acts on objects by x ↦ ( ( x ) , ( 0 ) , ( 0 ) ) , ( 0 ) x\mapsto((x),(0),(0)),(0) x ↦ (( x ) , ( 0 ) , ( 0 )) , ( 0 ) , and ( inc A ) x 0 , … , x k (\operatorname{inc}_{\mathcal{A}})^{x_0,\ldots,x_k} ( inc A ) x 0 , … , x k is id Hom A ( x 0 , x 1 ) \operatorname{id}_{\operatorname{Hom}_{\mathcal{A}}(x_0,x_1)} id Hom A ( x 0 , x 1 ) if k = 1 k=1 k = 1 and 0 0 0 if k ≠ 1 k\neq1 k = 1 . This is a strict isomorphism with a full A ∞ A_{\infty} A ∞ -subcategory of Tw A \operatorname{Tw}\mathcal{A} Tw A .
∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras
We now define many different ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras over a commutative ring R R R , which fit into a diagram of inclusions of ∞ \infty ∞ -categories:
A ∞ Alg R u n , ∞ = A ∞ Alg R , 2 u n , ∞ ↑ A ∞ Alg R , 3 u n , ∞ = A ˙ ∞ Alg R , 3 u n , ∞ ⟶ A ∞ Alg R , 3 c u , ∞ ⟶ A ∞ Alg R , 3 ∞ ↑ ↑ ↑ A ∞ Alg R , 4 u n , ∞ = A ˙ ∞ Alg R , 4 u n , ∞ ⟶ A ∞ Alg R , 4 c u , ∞ ⟶ A ∞ Alg R , 4 ∞ ↑ ↑ ↑ A ∞ Alg R , 5 u n , ∞ = A ˙ ∞ Alg R , 5 u n , ∞ ⟶ A ∞ Alg R , 5 c u , ∞ ⟶ A ∞ Alg R , 5 ∞ ⋮ ⋮ ⋮ A ∞ Alg R , ∞ u n , ∞ = A ˙ ∞ Alg R , ∞ u n , ∞ ⟶ A ∞ Alg R , ∞ c u , ∞ ⟶ A ∞ Alg R , ∞ ∞ ↑ ↑ ↑ N ( A ∞ Alg R u n ) = N ( A ∞ Alg R u n ) ⟶ N ( A ∞ Alg R c u ) ⟶ N ( A ∞ Alg R ) . (160) \begin{aligned}
&A_\infty\operatorname{Alg}_{R}^{\mathrm{un},\infty} = A_\infty\operatorname{Alg}_{R,2}^{\mathrm{un},\infty} \\
&\qquad\uparrow\\
&A_\infty\operatorname{Alg}_{R,3}^{\mathrm{un},\infty} = \dot{A}_\infty\operatorname{Alg}_{R,3}^{\mathrm{un},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,3}^{\mathrm{cu},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,3}^{\infty} \\
&\qquad\uparrow\qquad\qquad\qquad\qquad\qquad\qquad\uparrow\qquad\qquad\qquad\qquad\uparrow\\
&A_\infty\operatorname{Alg}_{R,4}^{\mathrm{un},\infty} = \dot{A}_\infty\operatorname{Alg}_{R,4}^{\mathrm{un},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,4}^{\mathrm{cu},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,4}^{\infty} \\
&\qquad\uparrow\qquad\qquad\qquad\qquad\qquad\qquad\uparrow\qquad\qquad\qquad\qquad\uparrow\\
&A_\infty\operatorname{Alg}_{R,5}^{\mathrm{un},\infty} = \dot{A}_\infty\operatorname{Alg}_{R,5}^{\mathrm{un},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,5}^{\mathrm{cu},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,5}^{\infty} \\
&\qquad\vdots\qquad\qquad\qquad\qquad\qquad\qquad\vdots\qquad\qquad\qquad\qquad\vdots\\
&A_\infty\operatorname{Alg}_{R,\infty}^{\mathrm{un},\infty} = \dot{A}_\infty\operatorname{Alg}_{R,\infty}^{\mathrm{un},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,\infty}^{\mathrm{cu},\infty}
\longrightarrow A_\infty\operatorname{Alg}_{R,\infty}^{\infty} \\
&\qquad\uparrow\qquad\qquad\qquad\qquad\qquad\qquad\uparrow\qquad\qquad\qquad\qquad\uparrow\\
&\mathcal{N}(A_\infty\operatorname{Alg}_{R}^{\mathrm{un}})
= \mathcal{N}(A_\infty\operatorname{Alg}_{R}^{\mathrm{un}})
\longrightarrow\mathcal{N}(A_\infty\operatorname{Alg}_{R}^{\mathrm{cu}})
\longrightarrow\mathcal{N}(A_\infty\operatorname{Alg}_{R}). \tag*{(160)}
\end{aligned} A ∞ Alg R un , ∞ = A ∞ Alg R , 2 un , ∞ ↑ A ∞ Alg R , 3 un , ∞ = A ˙ ∞ Alg R , 3 un , ∞ ⟶ A ∞ Alg R , 3 cu , ∞ ⟶ A ∞ Alg R , 3 ∞ ↑ ↑ ↑ A ∞ Alg R , 4 un , ∞ = A ˙ ∞ Alg R , 4 un , ∞ ⟶ A ∞ Alg R , 4 cu , ∞ ⟶ A ∞ Alg R , 4 ∞ ↑ ↑ ↑ A ∞ Alg R , 5 un , ∞ = A ˙ ∞ Alg R , 5 un , ∞ ⟶ A ∞ Alg R , 5 cu , ∞ ⟶ A ∞ Alg R , 5 ∞ ⋮ ⋮ ⋮ A ∞ Alg R , ∞ un , ∞ = A ˙ ∞ Alg R , ∞ un , ∞ ⟶ A ∞ Alg R , ∞ cu , ∞ ⟶ A ∞ Alg R , ∞ ∞ ↑ ↑ ↑ N ( A ∞ Alg R un ) = N ( A ∞ Alg R un ) ⟶ N ( A ∞ Alg R cu ) ⟶ N ( A ∞ Alg R ) . (160)
We comment on these:
(i) The ∞ \infty ∞ -categories in the first and second columns, with superscripts ‘un’, have objects unital A ∞ A_\infty A ∞ -algebras and 1 1 1 -morphisms unital A ∞ A_\infty A ∞ -morphisms.
(ii) The ∞ \infty ∞ -categories in the third column, with superscript ‘cu’, have objects cohomologically unital A ∞ A_\infty A ∞ -algebras and 1 1 1 -morphisms cohomologically unital A ∞ A_\infty A ∞ -morphisms.
(iii) The ∞ \infty ∞ -categories in the fourth column have objects all A ∞ A_\infty A ∞ -algebras, and 1 1 1 -morphisms all A ∞ A_\infty A ∞ -morphisms.
(iv) The ∞ \infty ∞ -categories in the first and second columns, linked by ‘= = = ’, are strictly isomorphic, but with definitions written in a different way. See Theorem 6.31 for the isomorphism I R , K : A ∞ Alg R , K u n , ∞ → A ˙ ∞ Alg R , K u n , ∞ I_{R,K}: A_\infty\operatorname{Alg}_{R,K}^{\mathrm{un},\infty}\to\dot{A}_\infty\operatorname{Alg}_{R,K}^{\mathrm{un},\infty} I R , K : A ∞ Alg R , K un , ∞ → A ˙ ∞ Alg R , K un , ∞ .
(v) In each row we have subscripts ‘R , K R,K R , K ’ for K = 2 , 3 , 4 , … , ∞ K=2,3,4,\ldots,\infty K = 2 , 3 , 4 , … , ∞ , where we call K K K the type of the ∞ \infty ∞ -category.
(vi) N ( A ∞ Alg R u n ) , … , N ( A ∞ Alg R ) \mathcal{N}(A_\infty\operatorname{Alg}_{R}^{\mathrm{un}}),\ldots,\mathcal{N}(A_\infty\operatorname{Alg}_{R}) N ( A ∞ Alg R un ) , … , N ( A ∞ Alg R ) are the nerves of the ordinary categories A ∞ Alg R u n , … , A ∞ Alg R A_\infty\operatorname{Alg}_{R}^{\mathrm{un}},\ldots,A_\infty\operatorname{Alg}_{R} A ∞ Alg R un , … , A ∞ Alg R from Definition 3.8.
Here is what we mean by ‘type’. Recall that A ∞ A_\infty A ∞ -algebras, A ∞ A_\infty A ∞ -morphisms, and A ∞ A_\infty A ∞ -homotopies in §3, and A ∞ A_\infty A ∞ -categories and A ∞ A_\infty A ∞ -functors in §4.1–§4.2, involved families of maps m k , f k , h k m_k,f_k,\mathfrak{h}_k m k , f k , h k for k ≥ 1 k \ge1 k ≥ 1 . However, the A ∞ A_\infty A ∞ -category Fun A ∞ ( A , B ) \operatorname{Fun}_{A_\infty}(\mathcal{A},\mathcal{B}) Fun A ∞ ( A , B ) in §4.3, A ∞ A_\infty A ∞ -natural transformations in §4.4, and filtered A ∞ A_\infty A ∞ -algebras and A ∞ A_\infty A ∞ -categories in Chapter 5, involved families of maps T k , m k , f k \mathcal{T}_k,m_k,f_k T k , m k , f k for k ≥ 0 k \ge0 k ≥ 0 .
The N N N -simplices in the ∞ \infty ∞ -categories in (160) all involve families of maps m k I m^I_k m k I for ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I \subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } . For ∞ \infty ∞ -categories of type K K K , the rule is that the maps m k I m^I_k m k I are defined for k ≥ 1 k \ge1 k ≥ 1 if ∣ I ∣ ≤ K |I| \le K ∣ I ∣ ≤ K , and for k ≥ 0 k \ge0 k ≥ 0 if ∣ I ∣ > K |I| > K ∣ I ∣ > K , that is, we set m 0 I = 0 m^I_0=0 m 0 I = 0 if ∣ I ∣ ≤ K |I| \le K ∣ I ∣ ≤ K . In effect this means that N N N -morphisms in ∞ \infty ∞ -categories of type K K K include maps m k m_k m k for k ≥ 1 k \ge1 k ≥ 1 if N < K N < K N < K , and maps m k m_k m k for k ≥ 0 k \ge0 k ≥ 0 if N ≥ K N \ge K N ≥ K .
When N = 2 N=2 N = 2 , in the ∞ \infty ∞ -categories A ∞ Alg R u n , ∞ A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_{R} A ∞ Alg R un , ∞ of type 2 in the first row of (160) , the 2-morphisms correspond to (cohomologically) unital A ∞ A_\infty A ∞ -natural isomorphisms in Definition 4.22, regarding A ∞ A_\infty A ∞ -algebras as A ∞ A_\infty A ∞ -categories with one object. But in the ∞ \infty ∞ -categories of type K ≥ 3 K \ge3 K ≥ 3 in the subsequent rows, the 2-morphisms correspond to (unital) homotopies of A ∞ A_\infty A ∞ -algebras, as in §3.4. This suggests that ∞ \infty ∞ -categories of type K ≥ 3 K \ge3 K ≥ 3 may be better for studying A ∞ A_\infty A ∞ -algebras, but when we pass to A ∞ A_\infty A ∞ -categories in Chapter 7, type 2 should be preferred.
We give two different, but related, ways to define ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras. The first, in §6.3, gives the first column of (160) , and the second, in §6.4, gives the last three columns. The first method works for types K ≥ 2 K \ge2 K ≥ 2 , but only for unital A ∞ A_\infty A ∞ -algebras, as the degeneracy maps involve units. The second method only works for types K ≥ 3 K \ge3 K ≥ 3 , but for general A ∞ A_\infty A ∞ -algebras.
We postpone most discussion of the motivation for the definitions, and their relation to the literature, to Chapter 7. Here we just note that our definition of A ∞ Alg R ∞ A_\infty\operatorname{Alg}^{\infty}_{R} A ∞ Alg R ∞ was a byproduct of an ongoing project with Guillem Cazassus and Alexander Ritter, on a new construction of Fukaya categories of symplectic manifolds ( X , ω ) (X,\omega) ( X , ω ) , which we hope will eventually appear in [16].
Our definitions of A ∞ Alg R u n , ∞ , … , A ∞ Alg R , ∞ ∞ A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_{R},\ldots,A_\infty\operatorname{Alg}^{\infty}_{R,\infty} A ∞ Alg R un , ∞ , … , A ∞ Alg R , ∞ ∞ will be written using sums over a class of graphs Γ \Gamma Γ called rooted planar trees . The vertices and edges of Γ \Gamma Γ will be labelled by data including R R R -linear graded operators m k I m^I_k m k I . Section 6.1 defines graphical notation suitable for describing (filtered) A ∞ A_\infty A ∞ -algebras, A ∞ A_\infty A ∞ -functors, and homotopies, as in §3 and §5.2–§5.4. Then §6.2 extends the notation to simplices of morphisms. The ∞ \infty ∞ -categories in (160) are defined in §6.3–§6.4.
Graphical notation for compositions of operators
In working with A ∞ A_\infty A ∞ -algebras, we meet many formulae involving multiple compositions of the maps m k , f k , g k , h k m_k,f_k,g_k,\mathfrak{h}_k m k , f k , g k , h k of §3.2 and §3.4. We will explain a way to draw pictures of these.
Definition 6.1. We will define a class of finite directed graphs Γ \Gamma Γ called rooted planar trees . (For brevity we often shorten this to tree .)
(i) Γ \Gamma Γ has finitely many (and at least two) vertices v v v and finitely many (and at least one) directed edges e e e , which go between two vertices. An edge e e e starting at vertex v v v and ending at vertex w w w is drawn v ∙ → e ∙ w v \mathbin{\bullet}\xrightarrow{e}\mathbin{\bullet} w v ∙ e ∙ w .
(ii) We require that Γ \Gamma Γ is a tree , that is, Γ \Gamma Γ is connected and simply-connected.
(iii) There is one distinguished vertex r r r called the root vertex , drawn ‘▲’. (This is why we call Γ \Gamma Γ rooted .) There should be exactly one edge e e e ending at r r r , and no edges starting at r r r .
(iv) As Γ \Gamma Γ is a tree, there is a unique shortest path in Γ \Gamma Γ from any vertex v v v to the root vertex r r r . We require that all edges in Γ \Gamma Γ should be oriented in the direction of the shortest path to r r r .
This implies that every non-root vertex v v v has exactly one edge e e e starting at v v v , and finitely many edges ending at v v v .
(v) The vertices v v v in Γ \Gamma Γ which have no edges ending at v v v are divided into two classes, leaves , drawn as ‘◦’ and often labelled l i l_i l i , and buds , drawn as ‘■’ and often labelled b i b_i b i . The distinction between leaves and buds is part of the data of Γ \Gamma Γ .
Vertices which are not leaves, buds, or the root vertex, are called internal vertices , drawn as ‘•’, and often labelled v i v_i v i .
(vi) When we say that Γ \Gamma Γ is planar , we mean that Γ \Gamma Γ is drawn in the plane R 2 \mathbb{R}^2 R 2 , and we care about Γ \Gamma Γ up to continuous deformations in R 2 \mathbb{R}^2 R 2 .
Since Γ \Gamma Γ is a tree, being planar is equivalent to the data of a cyclic order of the edges starting and ending at each vertex v v v . That is, if Γ \Gamma Γ is drawn in the plane then set of the edges starting and ending at v v v has a natural cyclic order going anticlockwise around v v v .
(vii) Let v v v be an internal vertex, one outgoing edge starting at v v v and k ≥ 1 k \ge1 k ≥ 1 incoming edges ending at v v v . We sometimes write these edges as e v , 0 , e v , 1 , … , e v , k e_{v,0}, e_{v,1}, \ldots, e_{v,k} e v , 0 , e v , 1 , … , e v , k , where e v , 0 e_{v,0} e v , 0 is the outgoing edge, and e v , 1 , … , e v , k e_{v,1}, \ldots, e_{v,k} e v , 1 , … , e v , k are the incoming edges, and the edges are in the anticlockwise cyclic order e v , 0 , e v , 1 , … , e v , k e_{v,0}, e_{v,1}, \ldots, e_{v,k} e v , 0 , e v , 1 , … , e v , k around v v v . This gives a numbering 0 , 1 , … , k 0,1,\ldots,k 0 , 1 , … , k of the edges around v v v .
(viii) Let there be n ≥ 0 n \ge0 n ≥ 0 leaves in Γ \Gamma Γ . We sometimes write these as l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n , such that the root r r r and the leaves appear in the anticlockwise cyclic order r , l 1 , … , l n r,l_1,\ldots,l_n r , l 1 , … , l n , going round the outside of Γ \Gamma Γ in R 2 \mathbb{R}^2 R 2 .
We call Γ \Gamma Γ an n n n -leaved rooted planar tree if it has n n n leaves.
(ix) Let there be m ≥ 0 m \ge0 m ≥ 0 buds in Γ \Gamma Γ . We sometimes write these as b 1 , … , b m b_1,\ldots,b_m b 1 , … , b m , such that the root r r r and the buds appear in the anticlockwise cyclic order r , b 1 , … , b m r,b_1,\ldots,b_m r , b 1 , … , b m , going round the outside of Γ \Gamma Γ in R 2 \mathbb{R}^2 R 2 .
(x) We sometimes write the m + n m+n m + n leaves and buds together as u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n , such that the root r r r and the leaves and buds appear in the anticlockwise cyclic order r , u 1 , … , u m + n r,u_1,\ldots,u_{m+n} r , u 1 , … , u m + n , going round the outside of Γ \Gamma Γ in R 2 \mathbb{R}^2 R 2 . Then u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n consists of the leaves l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n and the buds b 1 , … , b m b_1,\ldots,b_m b 1 , … , b m . For example, if m = 2 m=2 m = 2 and n = 3 n=3 n = 3 we might have ( u 1 , … , u 5 ) = ( b 1 , l 1 , l 2 , b 2 , l 3 ) (u_1,\ldots,u_5)=(b_1,l_1,l_2,b_2,l_3) ( u 1 , … , u 5 ) = ( b 1 , l 1 , l 2 , b 2 , l 3 ) .
A rooted planar tree (or just tree ) Γ \Gamma Γ means the data above: a finite set of vertices V V V , a finite set of edges E E E , head and tail maps h , t : V → E h,t: V \to E h , t : V → E , a division of vertices with no incoming edges into leaves and buds, a distinguished root r ∈ V r \in V r ∈ V , and cyclic orders of edges at each internal vertex, satisfying the conditions above. We illustrate a 6-leaved rooted planar tree in Figure 6.1 .
Figure 6.1. A rooted planar tree
Definition 6.2. Suppose that R R R is a commutative ring, and A A A is a graded R R R -module, and m k : A ⊗ k → A m_k:A^{\otimes k}\to A m k : A ⊗ k → A are graded R R R -linear maps of some degree for k ≥ 0 k \ge0 k ≥ 0 (or possibly k ≥ 1 k \ge1 k ≥ 1 , but then we just define m 0 = 0 m_0=0 m 0 = 0 ), where when k = 0 k=0 k = 0 by convention we write A ⊗ 0 = R A^{\otimes0}=R A ⊗ 0 = R , so that m 0 : R → A m_0:R\to A m 0 : R → A and often we identify m 0 m_0 m 0 with m 0 ( 1 ) ∈ A m_0(1)\in A m 0 ( 1 ) ∈ A . For example, such maps arise in A ∞ A_\infty A ∞ -algebras in Definition 3.2 , and curved A ∞ A_\infty A ∞ -algebras in Definition 5.1 . We can use rooted planar trees Γ \Gamma Γ to represent multiple compositions of the m k m_k m k operators as follows:
(i) Label each internal vertex v v v with k ≥ 1 k\ge1 k ≥ 1 incoming edges by the operator m k m_k m k .
(ii) Label each bud b b b with k k k incoming edges by the element m 0 ( 1 ) ∈ A m_0(1)\in A m 0 ( 1 ) ∈ A .
(iii) To get an expression like m 3 ( a 1 , m 2 ( a 2 , a 3 ) , a 4 ) m_3(a_1,m_2(a_2,a_3),a_4) m 3 ( a 1 , m 2 ( a 2 , a 3 ) , a 4 ) as in (63) involving evaluation on particular elements a 1 , a 2 , … a_1,a_2,\ldots a 1 , a 2 , … in A A A , label the leaves l i l_i l i of Γ \Gamma Γ by elements of A A A . When the leaves are ordered l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n as in Definition 6.1 (viii), we usually label l i l_i l i by a i a_i a i for a 1 , … , a n a_1,\ldots,a_n a 1 , … , a n in A A A .
(iv) To get an expression like m 3 ( id A ⊗ m 2 ⊗ id A ) : A ⊗ 4 → A m_3(\operatorname{id}_A\otimes m_2\otimes\operatorname{id}_A):A^{\otimes4}\to A m 3 ( id A ⊗ m 2 ⊗ id A ) : A ⊗ 4 → A as in (62) , with no evaluation on particular elements of A A A , do not label the leaves l i l_i l i .
(v) The root r r r represents the output, the composition of the whole formula.
We then build up the multiple composition by starting at the leaves, which represent the domain ( a 1 , … , a n ) ∈ A n (a_1,\ldots,a_n)\in A^n ( a 1 , … , a n ) ∈ A n or A ⊗ n A^{\otimes n} A ⊗ n of the formula, and following the directed edges, applying the operator m k m_k m k at each internal vertex to its k k k inputs coming along the k k k incoming edges, using the numbering e v , 1 , … , e v , k e_{v,1},\ldots,e_{v,k} e v , 1 , … , e v , k of the incoming edges in Definition 6.1 (vii) to determine the order of the k k k inputs into m k m_k m k , and substituting m 0 ( 1 ) m_0(1) m 0 ( 1 ) as the output at each bud. Some examples of this are shown in Figure 6.2 .
Figure 6.2. Examples of multiple compositions from labelled trees
We sometimes write bold Γ \boldsymbol{\Gamma} Γ to mean a rooted planar tree Γ \Gamma Γ together with choices of suitable labels at vertices (and later, also edges) which determine a multiple composition in an A ∞ A_{\infty} A ∞ -algebra. Then we call Γ \Gamma Γ a labelled tree . Note that we also use Γ \Gamma Γ in §6.2 and §7.1 to mean Γ \Gamma Γ together with other kinds of labels.
We will write C ( Γ ) C(\Gamma) C ( Γ ) , or C ( Γ , a 1 , … , a n ) C(\Gamma,a_1,\ldots,a_n) C ( Γ , a 1 , … , a n ) , for these multiple compositions of operators m k m_k m k , and elements a 1 , … , a n a_1,\ldots,a_n a 1 , … , a n .
In cases when we only have operators m k m_k m k for k ≥ 1 k \ge1 k ≥ 1 , not k ≥ 0 k \ge0 k ≥ 0 , such as the definitions of A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories, we restrict to trees with no buds.
Definition 6.3. Let Γ \Gamma Γ be a rooted planar tree. Define a total order ≺ \prec ≺ on the internal vertices v v v of Γ \Gamma Γ such that if v , w v,w v , w are internal vertices then v ≺ w v \prec w v ≺ w if either:
There is a directed path w → e 1 ∙ → e 2 ∙ ⋯ → e k ∙ w \xrightarrow{e_1} \bullet\xrightarrow{e_2} \bullet\cdots\xrightarrow{e_k} \bullet w e 1 ∙ e 2 ∙ ⋯ e k ∙ in Γ \Gamma Γ for k > 0 k>0 k > 0 ; or
The shortest paths from v , w v,w v , w to the root r r r first intersect at an internal vertex v ^ \hat{v} v ^ with v ≠ v ^ ≠ w v\ne\hat{v}\ne w v = v ^ = w , and writing the directed paths in Γ \Gamma Γ from v , w v,w v , w to v ′ v' v ′ as v → e 1 ∙ → e 2 ∙ ⋯ → e k v ^ v \xrightarrow{e_1} \bullet\xrightarrow{e_2} \bullet\cdots\xrightarrow{e_k} \hat{v} v e 1 ∙ e 2 ∙ ⋯ e k v ^ and w → e 1 ′ ∙ → e 2 ′ ∙ ⋯ → e k ′ ′ v ^ w \xrightarrow{e'_1} \bullet\xrightarrow{e'_2} \bullet\cdots\xrightarrow{e'_{k'}} \hat{v} w e 1 ′ ∙ e 2 ′ ∙ ⋯ e k ′ ′ v ^ for k , k ′ > 0 k,k'>0 k , k ′ > 0 , then the incoming edge e k e_k e k at v ^ \hat{v} v ^ is smaller than e k ′ ′ e'_{k'} e k ′ ′ in the numbering of Definition 6.1(vii). That is, writing e k = e v ^ , j e_k=e_{\hat{v},j} e k = e v ^ , j and e k ′ ′ = e v ^ , j ′ e'_{k'}=e_{\hat{v},j'} e k ′ ′ = e v ^ , j ′ , then j < j ′ j<j' j < j ′ .
By definition v ⊀ v v\not\prec v v ≺ v , that is, ≺ \prec ≺ is a total order like < < < , not like ≤ \le ≤ .
The importance of ≺ \prec ≺ is that when we write out a multiple composition C ( Γ ) C(\Gamma) C ( Γ ) in a form such as m 2 ∘ ( ( m 3 ∘ ( m 1 ⊗ m 2 ⊗ m 5 ) ) ⊗ m 4 ) m_2\circ\bigl((m_3\circ(m_1\otimes m_2\otimes m_5))\otimes m_4\bigr) m 2 ∘ ( ( m 3 ∘ ( m 1 ⊗ m 2 ⊗ m 5 )) ⊗ m 4 ) , the operators m k = m k v m_k=m_k^v m k = m k v associated to the internal vertices v v v appear in the order ≺ \prec ≺ on vertices. This is important in determining signs in formulae, as in Convention 3.1.
We can also extend ≺ \prec ≺ to both internal vertices v v v and leaves l l l of Γ \Gamma Γ by the same definition, and then ≺ \prec ≺ determines the order of both operators m k = m k v m_k=m_k^v m k = m k v associated to internal vertices v v v and also elements a i ∈ A a_i\in A a i ∈ A associated to leaves l i l_i l i of Γ \Gamma Γ in expressions such as m 2 ( m 2 ( m 1 ( a 1 ) , m 2 ( a 2 , a 3 ) ) , m 3 ( a 4 , a 5 , a 6 ) ) m_2\left(m_2\left(m_1(a_1),m_2(a_2,a_3)\right),m_3(a_4,a_5,a_6)\right) m 2 ( m 2 ( m 1 ( a 1 ) , m 2 ( a 2 , a 3 ) ) , m 3 ( a 4 , a 5 , a 6 ) ) .
Definition 6.4. Let f : ( A , m ) → ( B , n ) \mathbf{f}:(A,\mathbf{m})\to(B,\mathbf{n}) f : ( A , m ) → ( B , n ) be a morphism of A ∞ A_\infty A ∞ -algebras over R R R . Then we can also use labelled trees Γ \Gamma Γ to describe multiple compositions of maps f i , m j , n k f_i,m_j,n_k f i , m j , n k , as in (66) . Internal vertices of Γ \Gamma Γ can be labelled with operators f i , m j f_i,m_j f i , m j or n k n_k n k . In this case, it can also be helpful to label the edges of Γ \Gamma Γ by the A ∞ A_\infty A ∞ -algebra A A A or B B B , with the rules:
(i) If an internal vertex v v v is labelled by m k m_k m k , all edges at v v v are labelled A A A .
(ii) If an internal vertex v v v is labelled by n k n_k n k , all edges at v v v are labelled B B B .
(iii) If an internal vertex v v v is labelled by f k f_k f k , all incoming edges at v v v are labelled A A A , and the outgoing edge is labelled B B B .
(iv) Leaves l i l_i l i may be labelled by an element of A A A or B B B depending on whether the outgoing edge is labelled by A A A or B B B .
These ensure that compositions make sense, that is, the inputs to f i , m j , n k f_i,m_j,n_k f i , m j , n k lie in A A A or B B B as appropriate.
Given another morphism g : ( B , n ) → ( C , o ) \mathbf{g}:(B,\mathbf{n})\to(C,\mathbf{o}) g : ( B , n ) → ( C , o ) , we can also use labelled trees Γ \Gamma Γ to describe multiple compositions of maps f i , g j , m k , n l , o m f_i,g_j,m_k,n_l,o_m f i , g j , m k , n l , o m , in the obvious way. Also, if f , g : ( A , m ) → ( B , n ) \mathbf{f},\mathbf{g}:(A,\mathbf{m})\to(B,\mathbf{n}) f , g : ( A , m ) → ( B , n ) be morphisms of A ∞ A_\infty A ∞ -algebras, and H : f ⇒ g H:\mathbf{f}\Rightarrow\mathbf{g} H : f ⇒ g is a homotopy, we can use labelled trees Γ \Gamma Γ to describe multiple compositions of maps f i , g j , H k , m k , n l f_i,g_j,H_k,m_k,n_l f i , g j , H k , m k , n l .
Example 6.5. (a) We represent the defining equation (62) of A ∞ A_\infty A ∞ -algebras graphically in Figure 6.3 .
Figure 6.3. Equation (3.3): defining A ∞ A_\infty A ∞ -algebras ( A , m ) (A,\mathbf{m}) ( A , m )
(b) We represent the defining equation (66) of morphisms of A ∞ A_\infty A ∞ -algebras graphically in Figure 6.4 .
Figure 6.4. Equation (3.7): morphisms of A ∞ A_\infty A ∞ -algebras f : ( A , m ) → ( B , n ) f:(A,\mathfrak{m})\to(B,\mathfrak{n}) f : ( A , m ) → ( B , n )
Labelling graphs for simplices of morphisms
In our definition of ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras in §6.3–§6.4, the N N N -simplices involve graded R R R -modules A 0 , … , A N A^0,\ldots,A^N A 0 , … , A N and R R R -multilinear maps m k I : ( A min I ) ⊗ k → A max I m_k^I:(A^{\min I})^{\otimes k}\to A^{\max I} m k I : ( A m i n I ) ⊗ k → A m a x I for all subsets ∅ ≠ I ⊆ { 0 , 1 , … , N } \varnothing\ne I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } and k ≥ 0 k\ge0 k ≥ 0 . We extend the graphical notation of §6.1 to include multiple compositions of such maps m k I m_k^I m k I .
Definition 6.6. Let N ≥ 0 N\ge0 N ≥ 0 . A { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree Γ \Gamma Γ is a rooted planar tree Γ \Gamma Γ , as in Definition 6.1, together with a nonempty subset ∅ ≠ I v ⊆ { 0 , … , N } \varnothing\ne I_v\subseteq\{0,\ldots,N\} ∅ = I v ⊆ { 0 , … , N } for each vertex v v v in Γ \Gamma Γ , satisfying:
(a) If l l l is a leaf of Γ \Gamma Γ then I l = { 0 } I_l=\{0\} I l = { 0 } .
(b) The root r r r of Γ \Gamma Γ has I r = { N } I_r=\{N\} I r = { N } .
(c) Let v → e w v\xrightarrow{e}w v e w be a directed edge in Γ \Gamma Γ . Then max ( I v ) = min ( I w ) \max(I_v)=\min(I_w) max ( I v ) = min ( I w ) .
When we draw a { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree, we write the subset I v I_v I v next to the vertex v v v . We may omit labels of root and leaves as they are automatic by (a)–(b). For each edge v → e w v\xrightarrow{e}w v e w , we may also label e e e by I e : = max ( I v ) = min ( I w ) I_e:=\max(I_v)=\min(I_w) I e := max ( I v ) = min ( I w ) in { 0 , … , n } \{0,\ldots,n\} { 0 , … , n } . Figure 6.5 gives an example of a { 0 , … , 5 } \{0,\ldots,5\} { 0 , … , 5 } -labelled tree.
Figure 6.5. A { 0 , … , 5 } \{0,\ldots,5\} { 0 , … , 5 } -labelled tree
We do not need to take the labelling set to be { 0 , 1 , … , N } \{0,1,\ldots,N\} { 0 , 1 , … , N } . For example, we could replace it by { a 0 , a 1 , … , a N } \{a_0,a_1,\ldots,a_N\} { a 0 , a 1 , … , a N } for a 0 < a 1 < ⋯ < a N a_0<a_1<\cdots<a_N a 0 < a 1 < ⋯ < a N in N \mathbb{N} N , and get a notion of { a 0 , a 1 , … , a N } \{a_0,a_1,\ldots,a_N\} { a 0 , a 1 , … , a N } -labelled tree, where above we require that ∅ ≠ I v ⊆ { a 0 , a 1 , … , a N } \varnothing\ne I_v\subseteq\{a_0,a_1,\ldots,a_N\} ∅ = I v ⊆ { a 0 , a 1 , … , a N } , and in (a),(b) we take I l = { a 0 } I_l=\{a_0\} I l = { a 0 } and I r = { a N } I_r=\{a_N\} I r = { a N } . Or we could replace { 0 , 1 , … , N } \{0,1,\ldots,N\} { 0 , 1 , … , N } by any finite totally-ordered set.
Now let R R R be a commutative ring, and suppose we are given the data:
A i = ⨁ d ∈ Z A i , d A^i=\bigoplus_{d\in\mathbb{Z}}A^{i,d} A i = ⨁ d ∈ Z A i , d is a Z \mathbb{Z} Z -graded R R R -module for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M with M ≥ 0 M\ge0 M ≥ 0 , and each k ≥ 0 k\ge0 k ≥ 0 , we are given an R R R -linear map
m k I : ( A i 0 ) ⊗ k ⟶ A i M , m_k^I:(A^{i_0})^{\otimes k}\longrightarrow A^{i_M}, m k I : ( A i 0 ) ⊗ k ⟶ A i M ,
graded of degree 2 − M − k 2-M-k 2 − M − k . Here when k = 0 k=0 k = 0 , by convention ( A i 0 ) ⊗ 0 = R (A^{i_0})^{\otimes0}=R ( A i 0 ) ⊗ 0 = R , and we sometimes identify m 0 I : R → A i M m_0^I:R\to A^{i_M} m 0 I : R → A i M with m 0 I ( 1 ) ∈ A i M m_0^I(1)\in A^{i_M} m 0 I ( 1 ) ∈ A i M . Equivalently, we may regard m k I m_k^I m k I as an R R R -multilinear map
m k I : A i 0 × ⋯ × A i 0 ⏟ k copies ⟶ A i M . m_k^I:\underbrace{A^{i_0}\times\cdots\times A^{i_0}}_{k\text{ copies}}\longrightarrow A^{i_M}. m k I : k copies A i 0 × ⋯ × A i 0 ⟶ A i M .
If Γ \Gamma Γ is an n n n -leaved { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree, we define an R R R -linear map
C ( Γ ) : ( A 0 ) ⊗ n ⟶ A N (161) C(\Gamma):(A^0)^{\otimes n}\longrightarrow A^N \tag*{(161)} C ( Γ ) : ( A 0 ) ⊗ n ⟶ A N (161)
as in Definition 6.2, by associating the operator m k I v m_{k}^{I_v} m k I v to each internal vertex v v v with k ≥ 1 k\ge1 k ≥ 1 incoming edges, and m 0 I b ( 1 ) m_0^{I_b}(1) m 0 I b ( 1 ) to each bud b b b , and taking the composition over the diagram. (Note that we do not associate operators m 0 { 0 } m_0^{\{0\}} m 0 { 0 } with the leaves, or m 1 { N } m_1^{\{N\}} m 1 { N } with the root.) Conditions (a)–(c) above ensure that this composition is well-defined, that is, that the inputs to each m k I m_k^I m k I in the formula do lie in A i 0 A^{i_0} A i 0 for the correct i 0 i_0 i 0 , and the output lies in A N A^N A N . So, for example, Figure 6.5 yields the R R R -linear map
m 2 { 3 , 5 } ( m 2 { 0 , 1 , 3 } ⊗ m 2 { 2 , 3 } ( m 2 { 0 , 2 } ⊗ m 2 { 0 , 1 , 2 } ) ) : ( A 0 ) ⊗ 6 ⟶ A 5 . m_2^{\{3,5\}}\left(m_2^{\{0,1,3\}}\otimes m_2^{\{2,3\}}\left(m_2^{\{0,2\}}\otimes m_2^{\{0,1,2\}}\right)\right):(A^0)^{\otimes6}\longrightarrow A^5. m 2 { 3 , 5 } ( m 2 { 0 , 1 , 3 } ⊗ m 2 { 2 , 3 } ( m 2 { 0 , 2 } ⊗ m 2 { 0 , 1 , 2 } ) ) : ( A 0 ) ⊗ 6 ⟶ A 5 .
If we are given elements a 1 , … , a n ∈ A 0 a_1,\ldots,a_n \in A^0 a 1 , … , a n ∈ A 0 we can insert these as labels at the leaves l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n , and form C ( Γ ) ( a 1 ⊗ ⋯ ⊗ a n ) ∈ A N C(\Gamma)(a_1\otimes\cdots\otimes a_n)\in A^N C ( Γ ) ( a 1 ⊗ ⋯ ⊗ a n ) ∈ A N .
Since m k I m^I_k m k I has degree 3 − ∣ I ∣ − k 3-|I|-k 3 − ∣ I ∣ − k , an easy calculation shows C ( Γ ) C(\Gamma) C ( Γ ) has degree
deg C ( Γ ) = 1 − n + ∑ v internal vertex or bud of Γ ( 2 − ∣ I v ∣ ) . (162) \deg C(\Gamma)=1-n+\sum_{\substack{v\ \text{internal vertex or bud}\\\text{of }\Gamma}}(2-|I_v|). \tag*{(162)} deg C ( Γ ) = 1 − n + v internal vertex or bud of Γ ∑ ( 2 − ∣ I v ∣ ) . (162)
Balanced { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled trees will be important in our theory, as our definition of A ∞ A l g R ∞ A_\infty\mathrm{Alg}^{\infty}_R A ∞ Alg R ∞ involves sums over balanced trees.
Definition 6.7. Let Γ \Gamma Γ be a { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree with n ≥ 0 n\geq0 n ≥ 0 leaves and m ≥ 0 m\geq0 m ≥ 0 buds. As in Definition 6.1(x), let us write the m + n m+n m + n leaves and buds together as u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n , such that the root r r r and the leaves and buds appear in the anticlockwise cyclic order r , u 1 , … , u m + n r,u_1,\ldots,u_{m+n} r , u 1 , … , u m + n , going round the outside of Γ \Gamma Γ in R 2 \mathbb{R}^2 R 2 . We call Γ \Gamma Γ balanced if the following hold:
(a) For every internal vertex v v v in Γ \Gamma Γ we have ∣ I v ∣ ≥ 2 |I_v|\geq2 ∣ I v ∣ ≥ 2 .
(b) For every bud b b b in Γ \Gamma Γ we have ∣ I b ∣ ≥ 3 |I_b|\geq3 ∣ I b ∣ ≥ 3 .
(c) As extra data, each bud b b b of Γ \Gamma Γ should also be labelled by a nonempty subset P b P_b P b of { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } called the prefix of b b b , which should satisfy min P b = 0 \min P_b=0 min P b = 0 and max P b = min I b \max P_b=\min I_b max P b = min I b .
(d) For each i = 1 , … , m + n − 1 i=1,\ldots,m+n-1 i = 1 , … , m + n − 1 , consider the shortest paths in Γ \Gamma Γ from the leaves or buds u i u_i u i and u i + 1 u_{i+1} u i + 1 to the root r r r . Let w w w be the unique internal vertex that lies on both these paths and is closest to l i , l i + 1 l_i,l_{i+1} l i , l i + 1 . Write the shortest paths from u i , u i + 1 u_i,u_{i+1} u i , u i + 1 to w w w in Γ \Gamma Γ as u i = v 0 ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ w u_i=v^0\longrightarrow v^1\longrightarrow v^2\cdots v^k\longrightarrow w u i = v 0 ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ w and u i + 1 = v ~ 0 ⟶ v ~ 1 ⟶ v ~ 2 ⋯ v ~ k ~ ⟶ w u_{i+1}=\widetilde v^0\longrightarrow\widetilde v^1\longrightarrow\widetilde v^2\cdots\widetilde v^{\widetilde k}\longrightarrow w u i + 1 = v 0 ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ w , for k , k ~ ≥ 0 k,\widetilde k\geq0 k , k ≥ 0 .
Write I v j = { i 0 j , i 1 j , … , i a j j } I_{v^j}=\{i^j_0,i^j_1,\ldots,i^j_{a_j}\} I v j = { i 0 j , i 1 j , … , i a j j } for i 0 j < i 1 j < ⋯ < i a j j i^j_0<i^j_1<\cdots<i^j_{a_j} i 0 j < i 1 j < ⋯ < i a j j , j = 0 , … , k j=0,\ldots,k j = 0 , … , k , and I v ~ j = { i ~ 0 j , i ~ 1 j , … , i ~ a ~ j j } I_{\widetilde v^j}=\{\widetilde i^j_0,\widetilde i^j_1,\ldots,\widetilde i^j_{\widetilde a_j}\} I v j = { i 0 j , i 1 j , … , i a j j } for i ~ 0 j < i ~ 1 j < ⋯ < i ~ a ~ j j \widetilde i^j_0<\widetilde i^j_1<\cdots<\widetilde i^j_{\widetilde a_j} i 0 j < i 1 j < ⋯ < i a j j , j = 0 , … , k ~ j=0,\ldots,\widetilde k j = 0 , … , k . Then a j , a ~ j ≥ 1 a_j,\widetilde a_j\geq1 a j , a j ≥ 1 by (i), and Definition 6.6(c) gives
i a j j = i 0 j + 1 , i ~ a ~ j j = i ~ 0 j + 1 , i a k k = i ~ a ~ k ~ k ~ = min ( I w ) . i^{j}_{a_j}=i^{j+1}_0,\qquad\widetilde i^{j}_{\widetilde a_j}=\widetilde i^{j+1}_0,\qquad i^k_{a_k}=\widetilde i^{\widetilde k}_{\widetilde a_{\widetilde k}}=\min(I_w). i a j j = i 0 j + 1 , i a j j = i 0 j + 1 , i a k k = i a k k = min ( I w ) . If u i u_i u i is a bud, write the prefix P u i P_{u_i} P u i as P u i = { p 0 , … , p l } P_{u_i}=\{p_0,\ldots,p_l\} P u i = { p 0 , … , p l } with 0 = p 0 < p 1 < ⋯ < p l = min I u i = i 0 0 0=p_0<p_1<\cdots<p_l=\min I_{u_i}=i^0_0 0 = p 0 < p 1 < ⋯ < p l = min I u i = i 0 0 . Otherwise write l = − 1 l=-1 l = − 1 .
Similarly, if u i + 1 u_{i+1} u i + 1 is a bud, write P u i + 1 = { p ~ 0 , … , p ~ l ~ } P_{u_{i+1}}=\{\widetilde p_0,\ldots,\widetilde p_{\widetilde l}\} P u i + 1 = { p 0 , … , p l } with 0 = p ~ 0 < p ~ 1 < ⋯ < p ~ l ~ = min I u i + 1 = i ~ 0 0 0=\widetilde p_0<\widetilde p_1<\cdots<\widetilde p_{\widetilde l}=\min I_{u_{i+1}}=\widetilde i^0_0 0 = p 0 < p 1 < ⋯ < p l = min I u i + 1 = i 0 0 . Otherwise write l ~ = 0 \widetilde l=0 l = 0 and p ~ 0 = 0 \widetilde p_0=0 p 0 = 0 .
We require that in strictly increasing subsequences of { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } we have
( p 0 , p 1 , … , p l , i a 0 0 = i 0 1 , i a 1 1 = i 0 2 , … , i a k k ) = ( p ~ 0 , p ~ 1 , … , p ~ l ~ = i ~ 0 0 , i ~ a ~ 0 0 = i ~ 0 1 , i ~ a ~ 1 1 = i ~ 0 2 , … , i ~ a ~ k ~ − 1 k ~ − 1 = i ~ 0 k ~ , … , i ~ a ~ k ~ k ~ ) . (163) (p_0,p_1,\ldots,p_l,i^0_{a_0}=i^1_0,i^1_{a_1}=i^2_0,\ldots,i^k_{a_k})=
(\widetilde p_0,\widetilde p_1,\ldots,\widetilde p_{\widetilde l}=\widetilde i^0_0,\widetilde i^0_{\widetilde a_0}=\widetilde i^1_0,\widetilde i^1_{\widetilde a_1}=\widetilde i^2_0,\ldots,\widetilde i^{\widetilde k-1}_{\widetilde a_{\widetilde k-1}}=\widetilde i^{\widetilde k}_0,\ldots,\widetilde i^{\widetilde k}_{\widetilde a_{\widetilde k}}). \tag*{(163)} ( p 0 , p 1 , … , p l , i a 0 0 = i 0 1 , i a 1 1 = i 0 2 , … , i a k k ) = ( p 0 , p 1 , … , p l = i 0 0 , i a 0 0 = i 0 1 , i a 1 1 = i 0 2 , … , i a k − 1 k − 1 = i 0 k , … , i a k k ) . (163)
Suppose Γ \Gamma Γ is balanced. Write the shortest paths from u m + n , u 1 u_{m+n},u_1 u m + n , u 1 to the root r r r in Γ \Gamma Γ as u m + n ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ r u_{m+n}\longrightarrow v^1\longrightarrow v^2\cdots v^k\longrightarrow r u m + n ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ r and u 1 ⟶ v ~ 1 ⟶ v ~ 2 ⋯ v ~ k ~ ⟶ r u_1\longrightarrow\widetilde v^1\longrightarrow\widetilde v^2\cdots\widetilde v^{\widetilde k}\longrightarrow r u 1 ⟶ v 1 ⟶ v 2 ⋯ v k ⟶ r , for k , k ~ ≥ 0 k,\tilde{k} \ge0 k , k ~ ≥ 0 . If u m + n u_{m+n} u m + n is a bud, write the prefix P u m + n P_{u_{m+n}} P u m + n as P u m + n = { p 0 , … , p l } P_{u_{m+n}}=\{p_0,\ldots,p_l\} P u m + n = { p 0 , … , p l } with 0 = p 0 < p 1 < ⋯ < p l = min I u i = i 0 0 0=p_0<p_1<\cdots<p_l=\min I_{u_i}=i_0^0 0 = p 0 < p 1 < ⋯ < p l = min I u i = i 0 0 . Otherwise write l = − 1 l=-1 l = − 1 . Similarly, if u 1 u_1 u 1 is a bud, write P u 1 = { p ~ 0 , … , p ~ l ~ } P_{u_1}=\{\tilde{p}_0,\ldots,\tilde{p}_{\tilde{l}}\} P u 1 = { p ~ 0 , … , p ~ l ~ } with 0 = p ~ 0 < p ~ 1 < ⋯ < p ~ l ~ = min I u 1 = i ~ 0 0 0=\tilde{p}_0<\tilde{p}_1<\cdots<\tilde{p}_{\tilde{l}}=\min I_{u_1}=\tilde{i}_0^0 0 = p ~ 0 < p ~ 1 < ⋯ < p ~ l ~ = min I u 1 = i ~ 0 0 . Otherwise write l ~ = 0 \tilde{l}=0 l ~ = 0 and p ~ 0 = 0 \tilde{p}_0=0 p ~ 0 = 0 . Define the right and left sequences R ( Γ ) R(\Gamma) R ( Γ ) and L ( Γ ) L(\Gamma) L ( Γ ) of Γ \Gamma Γ to be
R ( Γ ) = { p 0 , p 1 , … , p l , i 0 0 = i 1 1 , i a 1 1 = i 0 2 , … , i a k k } , L ( Γ ) = { p ~ 0 , p ~ 1 , … , p ~ l ~ = i ~ 0 0 , i ~ 0 1 = i ~ 1 1 , … , i ~ a ~ k ~ − 1 k ~ − 1 = i ~ 0 k ~ , … , i ~ a ~ k ~ k ~ } . (164) \begin{aligned}
R(\Gamma)&=\{p_0,p_1,\ldots,p_l,i_0^0=i_1^1,i_{a_1}^1=i_0^2,\ldots,i_{a_k}^k\},\\
L(\Gamma)&=\{\tilde{p}_0,\tilde{p}_1,\ldots,\tilde{p}_{\tilde{l}}=\tilde{i}_0^0,\tilde{i}_0^1=\tilde{i}_1^1,\ldots,\tilde{i}_{\tilde{a}_{\tilde{k}-1}}^{\tilde{k}-1}=\tilde{i}_0^{\tilde{k}},\ldots,\tilde{i}_{\tilde{a}_{\tilde{k}}}^{\tilde{k}}\}. \tag*{(164)}
\end{aligned} R ( Γ ) L ( Γ ) = { p 0 , p 1 , … , p l , i 0 0 = i 1 1 , i a 1 1 = i 0 2 , … , i a k k } , = { p ~ 0 , p ~ 1 , … , p ~ l ~ = i ~ 0 0 , i ~ 0 1 = i ~ 1 1 , … , i ~ a ~ k ~ − 1 k ~ − 1 = i ~ 0 k ~ , … , i ~ a ~ k ~ k ~ } . (164)
Using (162) , we can show that if Γ \Gamma Γ is balanced then
deg C ( Γ ) = 1 − n + ∣ R ( Γ ) ∣ − ∣ L ( Γ ) ∣ . (165) \deg C(\Gamma)=1-n+|R(\Gamma)|-|L(\Gamma)|. \tag*{(165)} deg C ( Γ ) = 1 − n + ∣ R ( Γ ) ∣ − ∣ L ( Γ ) ∣. (165)
For example, Figure 6.5 is not balanced, as (163) for leaves l 2 , l 3 l_2,l_3 l 2 , l 3 requires that { 0 , 3 } = { 0 , 2 , 3 } \{0,3\}=\{0,2,3\} { 0 , 3 } = { 0 , 2 , 3 } , and (163) for leaves l 4 , l 5 l_4,l_5 l 4 , l 5 requires that { 0 , 2 } = { 0 , 1 , 2 } \{0,2\}=\{0,1,2\} { 0 , 2 } = { 0 , 1 , 2 } , both of which are false. Figure 6.6 gives an example of a balanced { 0 , … , 8 } \{0,\ldots,8\} { 0 , … , 8 } -labelled tree, with the values of R ( Γ ) R(\Gamma) R ( Γ ) and L ( Γ ) L(\Gamma) L ( Γ ) .
Figure 6.6. A balanced { 0 , … , 8 } \{0,\ldots,8\} { 0 , … , 8 } -labelled tree
Let Γ \Gamma Γ be a balanced n n n -leaved { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree, with leaves numbered l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n as in Definition 6.1(viii). For i = 1 , … , n i=1,\ldots,n i = 1 , … , n , let the shortest path from l i l_i l i to the root r r r in Γ \Gamma Γ be l i → e 1 ∙ → e 2 ∙ ⋯ ∙ → e k ▲ l_i \xrightarrow{e_1} \bullet\xrightarrow{e_2} \bullet\cdots\bullet\xrightarrow{e_k} \blacktriangle l i e 1 ∙ e 2 ∙ ⋯ ∙ e k ▲ . Then 0 = I e 1 < I e 2 < ⋯ < I e k = N 0=I_{e_1}<I_{e_2}<\cdots<I_{e_k}=N 0 = I e 1 < I e 2 < ⋯ < I e k = N by (a), where I e i I_{e_i} I e i is as in Definition 6.6. Define the edge sequence of l i l_i l i to be ES ( l i ) = { I e 1 , I e 2 , … , I e k } \operatorname{ES}(l_i)=\{I_{e_1},I_{e_2},\ldots,I_{e_k}\} ES ( l i ) = { I e 1 , I e 2 , … , I e k } . We can show from (d) and (164) that
L ( Γ ) ⊇ ES ( l 1 ) ⊇ ES ( l 2 ) ⊇ ES ( l 3 ) ⊇ ⋯ ⊇ ES ( l n ) ⊆ R ( Γ ) . L(\Gamma)\supseteq\operatorname{ES}(l_1)\supseteq\operatorname{ES}(l_2)\supseteq\operatorname{ES}(l_3)\supseteq\cdots\supseteq\operatorname{ES}(l_n)\subseteq R(\Gamma). L ( Γ ) ⊇ ES ( l 1 ) ⊇ ES ( l 2 ) ⊇ ES ( l 3 ) ⊇ ⋯ ⊇ ES ( l n ) ⊆ R ( Γ ) .
In particular, this implies that R ( Γ ) ⊆ ES ( l i ) R(\Gamma)\subseteq\operatorname{ES}(l_i) R ( Γ ) ⊆ ES ( l i ) for all i = 1 , … , n i=1,\ldots,n i = 1 , … , n .
Remark 6.8. Condition Definition 6.7(d) probably seems mysterious at this stage. However, in §7.1 we extend our notation to maps m k I , x 0 , … , x k m_k^{I,x_0,\ldots,x_k} m k I , x 0 , … , x k depending on a sequence of objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k , as for A ∞ A_\infty A ∞ -categories in Definition 4.1. Then part (d) will turn out to be necessary to match the domains of inputs and outputs of edges, so that the multiple compositions make sense.
The next lemma is elementary.
Lemma 6.9. Let Γ \Gamma Γ be a balanced { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree. For v v v an internal vertex or bud in Γ \Gamma Γ , let the interior int ( I v ) \operatorname{int}(I_v) int ( I v ) of I v I_v I v be
int ( I v ) = I v ∖ { min ( I v ) , max ( I v ) } . \operatorname{int}(I_v)=I_v\setminus\{\min(I_v),\max(I_v)\}. int ( I v ) = I v ∖ { min ( I v ) , max ( I v )} . Then int ( I v ) ⊆ L ( Γ ) ∖ R ( Γ ) \operatorname{int}(I_v)\subseteq L(\Gamma)\setminus R(\Gamma) int ( I v ) ⊆ L ( Γ ) ∖ R ( Γ ) . Furthermore, for each i i i in L ( Γ ) ∖ R ( Γ ) L(\Gamma)\setminus R(\Gamma) L ( Γ ) ∖ R ( Γ ) , there is a unique internal vertex or bud v v v of Γ \Gamma Γ with i ∈ int ( I v ) i\in\operatorname{int}(I_v) i ∈ int ( I v ) .
We define an operation of sawing a balanced { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree.
Definition 6.10. Let Γ \Gamma Γ be a balanced { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled tree, with right sequence R ( Γ ) = { i 0 , i 1 , … , i d } R(\Gamma)=\{i_0,i_1,\ldots,i_d\} R ( Γ ) = { i 0 , i 1 , … , i d } with 0 = i 0 < i 1 < ⋯ < i d = N 0=i_0<i_1<\cdots<i_d=N 0 = i 0 < i 1 < ⋯ < i d = N . Suppose d > 1 d>1 d > 1 , and let 0 < c < d 0<c<d 0 < c < d . We will explain how to saw Γ \Gamma Γ at level i c i_c i c .
As in Definition 6.6, each edge e e e of Γ \Gamma Γ is labelled by an element I e I_e I e in { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } . Form a new, disconnected labelled graph Γ ~ \widetilde{\Gamma} Γ by modifying Γ \Gamma Γ as follows: for each edge v ∙ → e u ∙ v\mathbin{\bullet}\xrightarrow{e}u\mathbin{\bullet} v ∙ e u ∙ of Γ \Gamma Γ with I e = i c I_e=i_c I e = i c , we ‘break e e e in the middle’ (or ‘saw through the branch e e e ’), replacing e e e by two new edges v ∙ → e ′ r e v\mathbin{\bullet}\xrightarrow{e'}r_e v ∙ e ′ r e and ℓ e → e ′ ′ u ∙ \ell_e\xrightarrow{e''}u\mathbin{\bullet} ℓ e e ′′ u ∙ , with two new vertices r e r_e r e and ℓ e \ell_e ℓ e . Here r e r_e r e is the root of a component of Γ ~ \widetilde{\Gamma} Γ , and ℓ e \ell_e ℓ e is a leaf of another component of Γ ~ \widetilde{\Gamma} Γ . We label r e r_e r e and ℓ e \ell_e ℓ e by I r e = I ℓ e = { i c } I_{r_e}=I_{\ell_e}=\{i_c\} I r e = I ℓ e = { i c } .
As R ( Γ ) ⊆ E S ( ℓ i ) R(\Gamma)\subseteq ES(\ell_i) R ( Γ ) ⊆ E S ( ℓ i ) , the shortest path in Γ \Gamma Γ from any leaf ℓ i \ell_i ℓ i to the root r r r contains a unique edge e e e with I e = i c I_e=i_c I e = i c . So cutting Γ \Gamma Γ into Γ ~ \widetilde{\Gamma} Γ disconnects r r r from every leaf ℓ i \ell_i ℓ i of Γ \Gamma Γ .
Write Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 for the connected component of Γ ~ \widetilde{\Gamma} Γ containing r r r . Then Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 is an { i c , … , N } \{i_c,\ldots,N\} { i c , … , N } -labelled tree, with leaves ℓ e \ell_e ℓ e for each edge e e e in Γ \Gamma Γ with I e = i c I_e=i_c I e = i c . Number this set of edges e e e in Γ \Gamma Γ as e 1 , … , e k e_1,\ldots,e_k e 1 , … , e k , such that the leaves of Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 in the order in Definition 6.1(viii) are ℓ e 1 , ℓ e 2 , … , ℓ e k \ell_{e_1},\ell_{e_2},\ldots,\ell_{e_k} ℓ e 1 , ℓ e 2 , … , ℓ e k . Define Γ ~ i \widetilde{\Gamma}_i Γ i to be the connected component of Γ ~ \widetilde{\Gamma} Γ containing r e i r_{e_i} r e i for i = 1 , … , k i=1,\ldots,k i = 1 , … , k . Write Γ ~ i \widetilde{\Gamma}_i Γ i for the labelled tree corresponding to Γ ~ i \widetilde{\Gamma}_i Γ i for i = 0 , … , k i=0,\ldots,k i = 0 , … , k .
To make Γ ~ 0 , … , Γ ~ k \widetilde{\Gamma}_0,\ldots,\widetilde{\Gamma}_k Γ 0 , … , Γ k into balanced labelled trees, we must explain how to define the extra data of the prefixes P ~ b \widetilde{P}_b P b of a bud b b b in some Γ ~ i \widetilde{\Gamma}_i Γ i from Definition 6.7(c), from the prefix P b P_b P b of the corresponding bud b b b in Γ \Gamma Γ . If b b b lies in Γ ~ i \widetilde{\Gamma}_i Γ i for i = 1 , … , k i=1,\ldots,k i = 1 , … , k , we just set P ~ b = P b \widetilde{P}_b=P_b P b = P b . If b b b lies in Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 , write P b = { j 0 , … , j l } P_b=\{j_0,\ldots,j_l\} P b = { j 0 , … , j l } for 0 = j 0 < j 1 < ⋯ < j l = min P v 0=j_0<j_1<\cdots<j_l=\min P_v 0 = j 0 < j 1 < ⋯ < j l = min P v . It is easy to see that I e = i c = j m I_e=i_c=j_m I e = i c = j m for some m = 0 , … , l m=0,\ldots,l m = 0 , … , l . Then we define the prefix of b b b in Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 to be P ~ b = { j m , … , j l } \widetilde{P}_b=\{j_m,\ldots,j_l\} P b = { j m , … , j l } .
We can now show that:
(i) Γ ~ \widetilde{\Gamma} Γ is the disjoint union of Γ ~ 0 , Γ ~ 1 , … , Γ ~ k \widetilde{\Gamma}_0,\widetilde{\Gamma}_1,\ldots,\widetilde{\Gamma}_k Γ 0 , Γ 1 , … , Γ k .
(ii) Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 is a balanced { i c , … , N } \{i_c,\ldots,N\} { i c , … , N } -labelled tree, with prefixes as above.
(iii) Γ ~ 1 , … , Γ ~ k \widetilde{\Gamma}_1,\ldots,\widetilde{\Gamma}_k Γ 1 , … , Γ k are balanced { 0 , … , i c } \{0,\ldots,i_c\} { 0 , … , i c } -labelled trees, with prefixes as above.
(iv) Write L ( Γ ) = { ı ~ 0 , ı ~ 1 , … , ı ~ d ~ } L(\Gamma)=\{\widetilde{\imath}_0,\widetilde{\imath}_1,\ldots,\widetilde{\imath}_{\widetilde{d}}\} L ( Γ ) = { 0 , 1 , … , d } with 0 = ı ~ 0 < ı ~ 1 < ⋯ < ı ~ d ~ = N 0=\widetilde{\imath}_0<\widetilde{\imath}_1<\cdots<\widetilde{\imath}_{\widetilde{d}}=N 0 = 0 < 1 < ⋯ < d = N and ı ~ c ~ = i c \widetilde{\imath}_{\widetilde{c}}=i_c c = i c . Then
L ( Γ ~ 0 ) = { ı ~ c ~ , ı ~ 1 , … , ı ~ d ~ } , R ( Γ ~ 0 ) = { i c , i c + 1 , … , i d } , L ( Γ ~ 1 ) = { ı ~ 0 , ı ~ 1 , … , ı ~ c ~ } , R ( Γ k ) = { i 0 , i 1 , … , i c } , R ( Γ ~ i ) = L ( Γ ~ i + 1 ) , i = 1 , 2 , … , k − 1. \begin{aligned}
L(\widetilde{\Gamma}_0)&=\{\widetilde{\imath}_{\widetilde{c}},\widetilde{\imath}_1,\ldots,\widetilde{\imath}_{\widetilde{d}}\},&
R(\widetilde{\Gamma}_0)&=\{i_c,i_{c+1},\ldots,i_d\},\\
L(\widetilde{\Gamma}_1)&=\{\widetilde{\imath}_0,\widetilde{\imath}_1,\ldots,\widetilde{\imath}_{\widetilde{c}}\},&
R(\Gamma_k)&=\{i_0,i_1,\ldots,i_c\},\\
R(\widetilde{\Gamma}_i)&=L(\widetilde{\Gamma}_{i+1}),&&i=1,2,\ldots,k-1.
\end{aligned} L ( Γ 0 ) L ( Γ 1 ) R ( Γ i ) = { c , 1 , … , d } , = { 0 , 1 , … , c } , = L ( Γ i + 1 ) , R ( Γ 0 ) R ( Γ k ) = { i c , i c + 1 , … , i d } , = { i 0 , i 1 , … , i c } , i = 1 , 2 , … , k − 1.
Note that as the added vertices are leaves and roots, the set of internal vertices of Γ \Gamma Γ is the disjoint union of the sets of internal vertices of Γ ~ 0 , … , Γ ~ k \widetilde{\Gamma}_0,\ldots,\widetilde{\Gamma}_k Γ 0 , … , Γ k , and the set of buds of Γ \Gamma Γ is the disjoint union of the sets of buds of Γ ~ 0 , … , Γ ~ k \widetilde{\Gamma}_0,\ldots,\widetilde{\Gamma}_k Γ 0 , … , Γ k .
We claim that the compositions satisfy
C ( Γ ) = ( ∏ v an internal vertex or bud of Γ ~ i , i > 0 , w an internal vertex or bud of Γ ~ 0 , v ≺ w in order ≺ of Γ , v has l incoming edges , w has m incoming edges ( − 1 ) ( 3 − ∣ I v ∣ − l ) ( 3 − ∣ I w ∣ − m ) ) ⋅ C ( Γ ~ 0 ) ∘ ( C ( Γ ~ 1 ) ⊗ C ( Γ ~ 2 ) ⊗ ⋯ ⊗ C ( Γ ~ k ) ) , (166) \begin{aligned}
C(\Gamma)=\left(\prod_{\substack{v\text{ an internal vertex or bud of }\widetilde{\Gamma}_i,\ i>0,\\
w\text{ an internal vertex or bud of }\widetilde{\Gamma}_0,\\
v\prec w\text{ in order }\prec\text{ of }\Gamma,\\
v\text{ has }l\text{ incoming edges},\\
w\text{ has }m\text{ incoming edges}}}
(-1)^{(3-|I_v|-l)(3-|I_w|-m)}\right)
\cdot C(\widetilde{\Gamma}_0)\circ
\bigl(C(\widetilde{\Gamma}_1)\otimes C(\widetilde{\Gamma}_2)\otimes\cdots\otimes C(\widetilde{\Gamma}_k)\bigr), \tag*{(166)}
\end{aligned} C ( Γ ) = v an internal vertex or bud of Γ i , i > 0 , w an internal vertex or bud of Γ 0 , v ≺ w in order ≺ of Γ , v has l incoming edges , w has m incoming edges ∏ ( − 1 ) ( 3 − ∣ I v ∣ − l ) ( 3 − ∣ I w ∣ − m ) ⋅ C ( Γ 0 ) ∘ ( C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k ) ) , (166)
where the total order ≺ \prec ≺ on the internal vertices of Γ \Gamma Γ is as in Definition 6.3. To see this, note that in C ( Γ ) C(\Gamma) C ( Γ ) , the operators m k v I v m_{k_v}^{I_v} m k v I v for the internal vertices v v v of Γ \Gamma Γ appear in order ≺ \prec ≺ . In C ( Γ ~ 0 ) ∘ ( C ( Γ ~ 1 ) ⊗ C ( Γ ~ 2 ) ⊗ ⋯ ⊗ C ( Γ ~ k ) ) C(\widetilde{\Gamma}_0)\circ(C(\widetilde{\Gamma}_1)\otimes C(\widetilde{\Gamma}_2)\otimes\cdots\otimes C(\widetilde{\Gamma}_k)) C ( Γ 0 ) ∘ ( C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k )) , the operators m k v I v m_{k_v}^{I_v} m k v I v appear firstly in the order of the tree Γ ~ 0 , Γ ~ 1 , … , Γ ~ k \widetilde{\Gamma}_0,\widetilde{\Gamma}_1,\ldots,\widetilde{\Gamma}_k Γ 0 , Γ 1 , … , Γ k containing v v v , and then for a fixed Γ ~ i \widetilde{\Gamma}_i Γ i in the order ≺ \prec ≺ restricted to Γ ~ i \widetilde{\Gamma}_i Γ i . Thus, two vertices v , w v,w v , w appear in a different order on the two sides of (6.7) if and only if v v v lies in Γ ~ i \widetilde{\Gamma}_i Γ i for i > 0 i>0 i > 0 , and w w w lies in Γ ~ 0 \widetilde{\Gamma}_0 Γ 0 , and v ≺ w v\prec w v ≺ w (or vice versa). As the degrees of m k v I v m_{k_v}^{I_v} m k v I v and m k w I w m_{k_w}^{I_w} m k w I w are 3 − ∣ I v ∣ − l 3-|I_v|-l 3 − ∣ I v ∣ − l , 3 − ∣ I w ∣ − m 3-|I_w|-m 3 − ∣ I w ∣ − m , the factor ( − 1 ) ( 3 − ∣ I v ∣ − l ) ( 3 − ∣ I w ∣ − m ) (-1)^{(3-|I_v|-l)(3-|I_w|-m)} ( − 1 ) ( 3 − ∣ I v ∣ − l ) ( 3 − ∣ I w ∣ − m ) is that prescribed in (3.2) for exchanging the order of m k v I v m_{k_v}^{I_v} m k v I v and m k w I w m_{k_w}^{I_w} m k w I w .
Figure 6.7 shows the result of sawing Figure 6.6 at level 6.
Figure 6.7. Sawing Figure 6.6 at level 6
Definition 6.11. Let R R R be a commutative ring, and suppose as in Definition 6.6 that we are given the data:
(i) A i = ⨁ d ∈ Z A i , d A^i=\bigoplus_{d\in\mathbb{Z}}A^{i,d} A i = ⨁ d ∈ Z A i , d is a Z \mathbb{Z} Z -graded R R R -module for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
(ii) For each nonempty subset I ⊂ { 0 , … , N } I\subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M with M ≥ 0 M\geq0 M ≥ 0 , and each k ≥ 0 k\geq0 k ≥ 0 , we are given an R R R -linear map m k I : ( A i 0 ) ⊗ k ⟶ A i M m_k^I:(A^{i_0})^{\otimes k}\longrightarrow A^{i_M} m k I : ( A i 0 ) ⊗ k ⟶ A i M , graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k .
Actually, because of Definition 6.7 (b), when k = 0 k=0 k = 0 we only use maps m 0 I m_0^I m 0 I for ∣ I ∣ ≥ 3 |I|\geq3 ∣ I ∣ ≥ 3 , so we need not define m 0 I m_0^I m 0 I when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . It is convenient to set m 0 I = 0 m_0^I=0 m 0 I = 0 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 .
For each nonempty subset I = { i 0 , i 1 , … , i M } ⊆ { 0 , … , N } I=\{i_0,i_1,\ldots,i_M\}\subseteq\{0,\ldots,N\} I = { i 0 , i 1 , … , i M } ⊆ { 0 , … , N } as in (ii) and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , define an R R R -linear map
d n , k I : ( A i 0 ) ⊗ n ⟶ ( A i M ) ⊗ k (167) d_{n,k}^I:(A^{i_0})^{\otimes n}\longrightarrow(A^{i_M})^{\otimes k} \tag*{(167)} d n , k I : ( A i 0 ) ⊗ n ⟶ ( A i M ) ⊗ k (167)
as follows:
d n , 0 I = { id R : ( A i 0 ) ⊗ 0 ⟶ ( A i M ) ⊗ 0 , n = 0 and M = 1 , 0 , otherwise. (168) d^I_{n,0} =
\begin{cases}
\operatorname{id}_R:(A^{i_0})^{\otimes0}\longrightarrow(A^{i_M})^{\otimes0}, & n=0\text{ and }M=1,\\
0, & \text{otherwise.}
\end{cases} \tag*{(168)} d n , 0 I = { id R : ( A i 0 ) ⊗ 0 ⟶ ( A i M ) ⊗ 0 , 0 , n = 0 and M = 1 , otherwise. (168)
d n , k { i 0 } = { ∑ j = 0 k − 1 ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id A i 0 ⊗ j ⊗ m n − k + 1 { i 0 } ⊗ id A i 0 ⊗ k − j − 1 , n ≥ k , 0 , n < k . (169) d^{\{i_0\}}_{n,k} =
\begin{cases}
\sum_{j=0}^{k-1}(-1)^{j+(n-k+1)(k-j-1)}\operatorname{id}_{A^{i_0}}^{\otimes j}\otimes m^{\{i_0\}}_{n-k+1}\otimes\operatorname{id}_{A^{i_0}}^{\otimes k-j-1}, & n\geq k,\\
0, & n<k.
\end{cases} \tag*{(169)} d n , k { i 0 } = { ∑ j = 0 k − 1 ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id A i 0 ⊗ j ⊗ m n − k + 1 { i 0 } ⊗ id A i 0 ⊗ k − j − 1 , 0 , n ≥ k , n < k . (169)
d n , k I = ∑ isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with n i ≥ 0 leaves , 1 ≤ i ≤ k , n = n 1 + ⋯ + n k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 ( − 1 ) ( k + 1 ) ( M + n + 1 ) ( ∏ v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ , v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) ) ⋅ C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k ) , (170) d^I_{n,k} =
\sum_{\substack{\text{isomorphism classes}\\
\text{of }\Gamma_1,\ldots,\Gamma_k:\\
\Gamma_i\text{ is a balanced}\\
I\text{-labelled tree}\\
\text{with }n_i\geq0\text{ leaves},\\
1\leq i\leq k,\\
n=n_1+\cdots+n_k,\\
L(\Gamma_1)=I,\\
R(\Gamma_k)=\{i_0,i_M\},\\
R(\Gamma_i)=L(\Gamma_{i+1}),\\
i=1,\ldots,k-1}}
(-1)^{(k+1)(M+n+1)}
\left(\prod_{\substack{
v\text{ internal vertex or bud of }\Gamma_i,\ 1\leq i\leq k,\\
w\text{ internal vertex or bud of }\Gamma_{i'},\ 1\leq i'\leq k,\\
\text{either }i<i',\text{ or }i=i',\\
v\prec w\text{ in order }\prec\text{ of }\Gamma_i,\\
v\text{ has }p\text{ incoming edges including }q\text{ leaves},\\
w\text{ has }r\text{ incoming edges}}}
(-1)^{N(v,w)+q(r+1)+(p+1)(|I_w|+r+1)}\right)
\cdot C(\Gamma_1)\otimes C(\Gamma_2)\otimes\cdots\otimes C(\Gamma_k), \tag*{(170)} d n , k I = isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with n i ≥ 0 leaves , 1 ≤ i ≤ k , n = n 1 + ⋯ + n k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 ∑ ( − 1 ) ( k + 1 ) ( M + n + 1 ) v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ , v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ∏ ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) ⋅ C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k ) , (170)
where we define
N ( v , w ) = ∣ { ( r , s ) ∈ int ( I v ) × int ( I w ) : r < s } ∣ . (171) N(v,w)=\left|\{(r,s)\in\operatorname{int}(I_v)\times\operatorname{int}(I_w):r<s\}\right|. \tag*{(171)} N ( v , w ) = ∣ {( r , s ) ∈ int ( I v ) × int ( I w ) : r < s } ∣ . (171)
with int ( I v ) = I v ∖ { min ( I v ) , max ( I v ) } \operatorname{int}(I_v)=I_v\setminus\{\min(I_v),\max(I_v)\} int ( I v ) = I v ∖ { min ( I v ) , max ( I v )} as in Lemma 6.9.
It follows from (165) and (168) –(170) that d n , k I d^{I}_{n,k} d n , k I is graded of degree
deg d n , k I = k − n + 2 − ∣ I ∣ . \deg d^{I}_{n,k}=k-n+2-|I|. deg d n , k I = k − n + 2 − ∣ I ∣.
Remark 6.12. The conditions Definition 6.7(a),(b) in the definition of balanced labelled trees Γ \Gamma Γ , which say that ∣ I v ∣ ≥ 2 |I_v|\ge2 ∣ I v ∣ ≥ 2 if v v v is an internal vertex and ∣ I b ∣ ≥ 3 |I_b|\ge3 ∣ I b ∣ ≥ 3 if b b b is a bud, are both necessary to ensure that (170) is a finite sum.
Example 6.13. (i) If k = 1 k=1 k = 1 then for any I , n I,n I , n we have
d n , 1 I = { m n I , n ≥ 1 or ∣ I ∣ ≥ 3 , 0 , n = 0 and ∣ I ∣ = 1 , 2. (172) d^{I}_{n,1}=
\begin{cases}
m^{I}_{n}, & n\ge1\text{ or }|I|\ge3,\\
0, & n=0\text{ and }|I|=1,2.
\end{cases} \tag*{(172)} d n , 1 I = { m n I , 0 , n ≥ 1 or ∣ I ∣ ≥ 3 , n = 0 and ∣ I ∣ = 1 , 2. (172)
This is because for n ≥ 1 n\ge1 n ≥ 1 the only balanced I I I -labelled graph Γ \Gamma Γ with L ( Γ ) = I L(\Gamma)=I L ( Γ ) = I and R ( Γ ) = { i 0 , i m } R(\Gamma)=\{i_0,i_m\} R ( Γ ) = { i 0 , i m } is the graph with only one internal vertex v v v with I v = I I_v=I I v = I and n n n leaves, and all the signs in (170) are trivial. If n = 0 n=0 n = 0 then v v v is a bud, and we require ∣ I v ∣ ≥ 3 |I_v|\ge3 ∣ I v ∣ ≥ 3 by Definition 6.7(b).
(ii) If I = { 0 , 1 } I=\{0,1\} I = { 0 , 1 } and k ≥ 1 k\ge1 k ≥ 1 then after simplifying signs
d n , k { 0 , 1 } = ∑ n = n 1 + ⋯ + n k , n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } . (173) d^{\{0,1\}}_{n,k}
=
\sum_{\substack{n=n_1+\cdots+n_k,\\ n_a\ge1}}
(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}
m^{\{0,1\}}_{n_1}\otimes m^{\{0,1\}}_{n_2}\otimes\cdots\otimes m^{\{0,1\}}_{n_k}. \tag*{(173)} d n , k { 0 , 1 } = n = n 1 + ⋯ + n k , n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } . (173)
(iii) If I = { 0 , 1 , 2 } I=\{0,1,2\} I = { 0 , 1 , 2 } and k ≥ 1 k\ge1 k ≥ 1 then after simplifying signs
d n , k { 0 , 1 , 2 } = ∑ n = n 1 + ⋯ + n k , 1 ≤ l ≤ k , n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 , p i j ≥ 1 , m i ≥ 1 , n i ≥ 1 , i ≠ l , n l ≥ 0 ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ ∏ i = 1 l − 1 ( − 1 ) ∑ a = 1 m i ( p i a + 1 ) ( m i − a ) ⋅ m p 1 1 { 1 , 2 } ( m p 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p 1 m 1 { 0 , 1 } ) ⊗ ⋯ ⊗ m p l − 1 1 { 1 , 2 } ( m p l − 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p l − 1 m l − 1 { 0 , 1 } ) ⊗ m n l { 0 , 1 , 2 } ⊗ m n l + 1 { 0 , 2 } ⊗ ⋯ ⊗ m n k { 0 , 2 } . \begin{aligned}
d^{\{0,1,2\}}_{n,k}
={}&
\sum_{\substack{
n=n_1+\cdots+n_k,\\
1\le l\le k,\\
n_i=p_i^1+\cdots+p_i^{m_i},\quad i=1,\ldots,l-1,\\
p_i^j\ge1,\ m_i\ge1,\\
n_i\ge1,\ i\ne l,\\
n_l\ge0
}}
(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}
\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}
\\
&\quad\cdot\prod_{i=1}^{l-1}(-1)^{\sum_{a=1}^{m_i}(p_i^a+1)(m_i-a)}
\\
&\quad\cdot m^{\{1,2\}}_{p_1^1}
\left(m^{\{0,1\}}_{p_1^2}\otimes\cdots\otimes m^{\{0,1\}}_{p_1^{m_1}}\right)
\otimes\cdots\otimes
m^{\{1,2\}}_{p_{l-1}^1}
\left(m^{\{0,1\}}_{p_{l-1}^2}\otimes\cdots\otimes m^{\{0,1\}}_{p_{l-1}^{m_{l-1}}}\right)
\\
&\quad\otimes m^{\{0,1,2\}}_{n_l}\otimes m^{\{0,2\}}_{n_{l+1}}\otimes\cdots\otimes m^{\{0,2\}}_{n_k}.
\end{aligned} d n , k { 0 , 1 , 2 } = n = n 1 + ⋯ + n k , 1 ≤ l ≤ k , n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 , p i j ≥ 1 , m i ≥ 1 , n i ≥ 1 , i = l , n l ≥ 0 ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ i = 1 ∏ l − 1 ( − 1 ) ∑ a = 1 m i ( p i a + 1 ) ( m i − a ) ⋅ m p 1 1 { 1 , 2 } ( m p 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p 1 m 1 { 0 , 1 } ) ⊗ ⋯ ⊗ m p l − 1 1 { 1 , 2 } ( m p l − 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p l − 1 m l − 1 { 0 , 1 } ) ⊗ m n l { 0 , 1 , 2 } ⊗ m n l + 1 { 0 , 2 } ⊗ ⋯ ⊗ m n k { 0 , 2 } .
Lemma 6.14. In Definition 6.11 , d n , k I = 0 d^{I}_{n,k}=0 d n , k I = 0 unless n ≥ min ( k , k + 2 − ∣ I ∣ ) n\ge\min(k,k+2-|I|) n ≥ min ( k , k + 2 − ∣ I ∣ ) .
Proof. In (168) –(169) this is obvious. Suppose k > n k>n k > n . For the sum in (170) to be nonempty, there must exist Γ 1 , … , Γ k \Gamma_1,\ldots,\Gamma_k Γ 1 , … , Γ k as in (170) , where at least k − n k-n k − n of the Γ i \Gamma_i Γ i have no leaves, and so must have at least one bud b b b . Then ∣ I b ∣ ≥ 3 |I_b|\ge3 ∣ I b ∣ ≥ 3 by Definition 6.7(b), so ∣ int ( I b ) ∣ ≥ 1 |\operatorname{int}(I_b)|\ge1 ∣ int ( I b ) ∣ ≥ 1 . But the elements of int ( I b ) \operatorname{int}(I_b) int ( I b ) for the ≥ k − n \ge k-n ≥ k − n buds b b b are disjoint elements of I ∖ { i 0 , i M } I\setminus\{i_0,i_M\} I ∖ { i 0 , i M } . Thus k − n ≥ ∣ I ∣ − 2 k-n\ge|I|-2 k − n ≥ ∣ I ∣ − 2 . □
The next theorem is proved in §10.1.
Theorem 6.15. In Definition 6.11 , let k = k ′ + k ′ ′ k=k'+k'' k = k ′ + k ′′ with k ′ , k ′ ′ ≥ 1 k',k''\ge1 k ′ , k ′′ ≥ 1 . Then
(a) If M = 0 M=0 M = 0 and n ≥ k n\geq k n ≥ k then
d n , k { i 0 } = ( − 1 ) k ′ ′ ( k ′ + n ) d n − k ′ ′ , k ′ { i 0 } ⊗ id A i 0 ⊗ k ′ ′ + ( − 1 ) k ′ id A i 0 ⊗ k ′ ⊗ d n − k ′ , k ′ ′ { i 0 } . (174) d^{\{i_0\}}_{n,k}=(-1)^{k^{\prime\prime}(k^\prime+n)}d^{\{i_0\}}_{n-k^{\prime\prime},k^\prime}\otimes\operatorname{id}_{A_{i_0}}^{\otimes k^{\prime\prime}}+(-1)^{k^\prime}\operatorname{id}_{A_{i_0}}^{\otimes k^\prime}\otimes d^{\{i_0\}}_{n-k^\prime,k^{\prime\prime}}. \tag*{(174)} d n , k { i 0 } = ( − 1 ) k ′′ ( k ′ + n ) d n − k ′′ , k ′ { i 0 } ⊗ id A i 0 ⊗ k ′′ + ( − 1 ) k ′ id A i 0 ⊗ k ′ ⊗ d n − k ′ , k ′′ { i 0 } . (174)
If M = 0 M=0 M = 0 and n < k n<k n < k then d n , k { i 0 } = 0 d^{\{i_0\}}_{n,k}=0 d n , k { i 0 } = 0 by (169) .
(b) If M > 0 M>0 M > 0 then
d n , k I = ∑ n = n ′ + n ′ ′ , n ′ , n ′ ′ ≥ 0 , 0 < l ≤ M , 0 = j 0 < j 1 < ⋯ < j l = M , k 0 , … , k l ≥ 0 , k 0 = n ′ , k l = k ′ , k a − 1 ≥ k a + 1 − j a + j a − 1 , a = 1 , … , l ( − 1 ) k ′ ′ ( k ′ + l + M ) + n ′ ′ ( k ′ ′ + l + 1 ) + ∑ a = 1 l − 1 ( j a − a ) ( d k l − 1 , k l { i j l − 1 , … , i j l } ∘ d k l − 2 , k l − 1 { i j l − 2 , … , i j l − 1 } ∘ ⋯ ∘ d k 0 , k 1 { i j 0 , … , i j 1 } ) ⊗ d n ′ ′ , k ′ ′ { i j 0 , i j 1 , i j 2 , … , i j l } , (175) \begin{aligned}
d^I_{n,k}
={}&\sum_{\substack{n=n^\prime+n^{\prime\prime},\ n^\prime,n^{\prime\prime}\geq0,\\
0<l\leq M,\ 0=j_0<j_1<\cdots<j_l=M,\\
k_0,\ldots,k_l\geq0,\ k_0=n^\prime,\ k_l=k^\prime,\\
k_{a-1}\geq k_a+1-j_a+j_{a-1},\ a=1,\ldots,l}}
(-1)^{k^{\prime\prime}(k^\prime+l+M)+n^{\prime\prime}(k^{\prime\prime}+l+1)+\sum_{a=1}^{l-1}(j_a-a)}\\
&\qquad\left(d^{\{i_{j_{l-1}},\ldots,i_{j_l}\}}_{k_{l-1},k_l}\circ d^{\{i_{j_{l-2}},\ldots,i_{j_{l-1}}\}}_{k_{l-2},k_{l-1}}\circ\cdots\circ d^{\{i_{j_0},\ldots,i_{j_1}\}}_{k_0,k_1}\right)\otimes d^{\{i_{j_0},i_{j_1},i_{j_2},\ldots,i_{j_l}\}}_{n^{\prime\prime},k^{\prime\prime}}, \tag*{(175)}
\end{aligned} d n , k I = n = n ′ + n ′′ , n ′ , n ′′ ≥ 0 , 0 < l ≤ M , 0 = j 0 < j 1 < ⋯ < j l = M , k 0 , … , k l ≥ 0 , k 0 = n ′ , k l = k ′ , k a − 1 ≥ k a + 1 − j a + j a − 1 , a = 1 , … , l ∑ ( − 1 ) k ′′ ( k ′ + l + M ) + n ′′ ( k ′′ + l + 1 ) + ∑ a = 1 l − 1 ( j a − a ) ( d k l − 1 , k l { i j l − 1 , … , i j l } ∘ d k l − 2 , k l − 1 { i j l − 2 , … , i j l − 1 } ∘ ⋯ ∘ d k 0 , k 1 { i j 0 , … , i j 1 } ) ⊗ d n ′′ , k ′′ { i j 0 , i j 1 , i j 2 , … , i j l } , (175)
where the sum is finite. Here the condition k a − 1 ≥ k a + 1 − j a + j a − 1 k_{a-1}\geq k_a+1-j_a+j_{a-1} k a − 1 ≥ k a + 1 − j a + j a − 1 is unnecessary, as if it does not hold then d k a − 1 , k a { i j a − 1 , … , i j a } = 0 d^{\{i_{j_{a-1}},\ldots,i_{j_a}\}}_{k_{a-1},k_a}=0 d k a − 1 , k a { i j a − 1 , … , i j a } = 0 by Lemma 6.14, but without it the sum (175) would contain infinitely many zero terms.
The ∞ \infty ∞ -categories A ∞ − A l g R u n , ∞ \mathrm{A}_{\infty}\mathrm{-Alg}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ , A ∞ − A l g R , K u n , ∞ \mathrm{A}_{\infty}\mathrm{-Alg}^{\mathrm{un},\infty}_{R,K} A ∞ − Alg R , K un , ∞
We now define an ∞ \infty ∞ -category A ∞ − A l g R u n , ∞ \mathrm{A}_{\infty}\mathrm{-Alg}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ of unital A ∞ \mathrm{A}_{\infty} A ∞ -algebras. We first define a simplicial set A ∞ − A l g ^ R u n , ∞ \widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ which is not an ∞ \infty ∞ -category, and then define our ∞ \infty ∞ -category as a simplicial subset A ∞ − A l g R u n , ∞ ⊂ A ∞ − A l g ^ R u n , ∞ \mathrm{A}_{\infty}\mathrm{-Alg}^{\mathrm{un},\infty}_R\subset\widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ ⊂ A ∞ − Alg R un , ∞ . Theorem 6.23 extends A ∞ − A l g ^ R u n , ∞ \widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ to an ∞ \infty ∞ -bicategory ( A ∞ − A l g ^ R u n , ∞ , T ) (\widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R,T) ( A ∞ − Alg R un , ∞ , T ) in the sense of § 2.8 , with A ∞ − A l g R u n , ∞ = Core ( A ∞ − A l g ^ R u n , ∞ , T ) \mathrm{A}_{\infty}\mathrm{-Alg}^{\mathrm{un},\infty}_R=\operatorname{Core}(\widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R,T) A ∞ − Alg R un , ∞ = Core ( A ∞ − Alg R un , ∞ , T ) .
Definition 6.16. Use the notation of Chapter 2 and § 6.1 –§ 6.2 . Let R R R be a commutative ring. We will define a simplicial set A ∞ − A l g ^ R u n , ∞ \widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ of unital A ∞ \mathrm{A}_{\infty} A ∞ -algebras over R R R .
(a) For N ≥ 0 N\geq0 N ≥ 0 , define the N N N -simplices ( A ∞ − A l g ^ R u n , ∞ ) N (\widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R)_N ( A ∞ − Alg R un , ∞ ) N in A ∞ − A l g ^ R u n , ∞ \widehat{\mathrm{A}_{\infty}\mathrm{-Alg}}^{\mathrm{un},\infty}_R A ∞ − Alg R un , ∞ to be the set of pairs ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^N,(m^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where:
(i) A i = ⨁ d ∈ Z A i , d A^i=\bigoplus_{d\in\mathbb{Z}}A^{i,d} A i = ⨁ d ∈ Z A i , d is a Z \mathbb{Z} Z -graded R R R -module for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
(ii) For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M with M ≥ 0 M\geq0 M ≥ 0 , we have m I = ( m k I ) k ≥ 0 m^I=(m^I_k)_{k\geq0} m I = ( m k I ) k ≥ 0 , where m k I m^I_k m k I is an R R R -linear map
m k I : ( A i 0 ) ⊗ k ⟶ A i M , (176) m^I_k:(A^{i_0})^{\otimes k}\longrightarrow A^{i_M}, \tag*{(176)} m k I : ( A i 0 ) ⊗ k ⟶ A i M , (176)
graded of degree $2-M-k=3-|I|-k$. Here when k = 0 k=0 k = 0 , by convention ( A i 0 ) ⊗ 0 = R (A^{i_0})^{\otimes0}=R ( A i 0 ) ⊗ 0 = R , and we sometimes identify m 0 I : R → A i M m^I_0:R\to A^{i_M} m 0 I : R → A i M with m 0 I ( 1 ) ∈ A i M m^I_0(1)\in A^{i_M} m 0 I ( 1 ) ∈ A i M . In fact when k = 0 k=0 k = 0 we only want maps m 0 I m^I_0 m 0 I for ∣ I ∣ ≥ 3 |I|\geq3 ∣ I ∣ ≥ 3 , so we require that m 0 I = 0 m^I_0=0 m 0 I = 0 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . Alternatively, we can say that m k I m^I_k m k I in (176) should only be defined for k ≥ 1 k\geq1 k ≥ 1 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 and for k ≥ 0 k\geq0 k ≥ 0 when $|I|\geq 3, and then we should exclude terms in (177) and other equations below involving m 0 I m^I_0 m 0 I for ∣ I ∣ = 1 , 2 |I| = 1, 2 ∣ I ∣ = 1 , 2 .
We require that for all I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I = \{i_0 < i_1 < \cdots< i_M\} \subseteq\{0, \ldots, N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } as above and all n ≥ 0 n \ge0 n ≥ 0 , the maps m j I m^I_j m j I should satisfy
∑ k = 1 n m k { i 0 } ∘ d n , k { i 0 } = 0 if M = 0 , \sum_{k=1}^{n} m^{\{i_0\}}_k \circ d^{\{i_0\}}_{n,k} = 0 \quad\text{if } M = 0, k = 1 ∑ n m k { i 0 } ∘ d n , k { i 0 } = 0 if M = 0 ,
∑ k = 1 n m k { i 0 } ∘ d n , k { i 0 } + ( − 1 ) M ∑ k = 1 n + M − 1 m k { i M } ∘ d n , k I = 0 if M > 0. (177) \sum_{k=1}^{n} m^{\{i_0\}}_k \circ d^{\{i_0\}}_{n,k} + (-1)^M \sum_{k=1}^{n+M-1} m^{\{i_M\}}_k \circ d^I_{n,k} = 0 \quad\text{if } M > 0. \tag*{(177)} k = 1 ∑ n m k { i 0 } ∘ d n , k { i 0 } + ( − 1 ) M k = 1 ∑ n + M − 1 m k { i M } ∘ d n , k I = 0 if M > 0. (177)
where d n , k I d^I_{n,k} d n , k I is defined from the m j I m^I_j m j I as in (167) –(170) . Here the upper limits k = n k = n k = n and k = n + M − 1 k = n+M-1 k = n + M − 1 in the sums in (177) are unnecessary as d n , k { i 0 } = 0 d^{\{i_0\}}_{n,k} = 0 d n , k { i 0 } = 0 if k > n k > n k > n and d n , k I = 0 d^I_{n,k} = 0 d n , k I = 0 when k > n + M − 1 k > n+M-1 k > n + M − 1 for M > 0 M > 0 M > 0 by Lemma 6.14, so we could replace the sums by ∑ k = 0 ∞ \sum_{k=0}^{\infty} ∑ k = 0 ∞ , but we write it as in (177) to get finite sums. The lower limits k = 1 k = 1 k = 1 are as m 0 { i } = 0 m^{\{i\}}_0 = 0 m 0 { i } = 0 .
(iii) As in Example 6.20(a), parts (i)–(ii) imply that ( A i , m { i } ) (A^i,m^{\{i\}}) ( A i , m { i } ) is an A ∞ A_\infty A ∞ -algebra for each i = 0 , … , N i = 0,\ldots,N i = 0 , … , N . We require this A ∞ A_\infty A ∞ -algebra ( A i , m { i } ) (A^i,m^{\{i\}}) ( A i , m { i } ) to be unital , with unique unit 1 i ∈ A 0 i 1^i \in A^i_0 1 i ∈ A 0 i . That is, we have
m 2 { i } ∘ ( 1 i ⊗ id A i ) = m 2 { i } ∘ ( id A i ⊗ 1 i ) = id A i , m^{\{i\}}_2 \circ(1^i \otimes\operatorname{id}_{A_i}) = m^{\{i\}}_2 \circ(\operatorname{id}_{A_i} \otimes1^i) = \operatorname{id}_{A_i}, m 2 { i } ∘ ( 1 i ⊗ id A i ) = m 2 { i } ∘ ( id A i ⊗ 1 i ) = id A i ,
m k { i } ∘ ( id A i ⊗ l − 1 ⊗ 1 i ⊗ id A i ⊗ k − l ) = 0 , all 1 ≤ l ≤ k , k ≠ 2. (178) m^{\{i\}}_k \circ(\operatorname{id}_{A_i}^{\otimes l-1} \otimes1^i \otimes\operatorname{id}_{A_i}^{\otimes k-l}) = 0,\quad\text{all } 1 \le l \le k,\ k \ne2. \tag*{(178)} m k { i } ∘ ( id A i ⊗ l − 1 ⊗ 1 i ⊗ id A i ⊗ k − l ) = 0 , all 1 ≤ l ≤ k , k = 2. (178)
(iv) As in Example 6.20(b), parts (i)–(iii) imply that m { i , j } : ( A i , m { i } ) ⟶ ( A j , m { j } ) m^{\{i,j\}}\colon(A^i,m^{\{i\}}) \longrightarrow(A^j,m^{\{j\}}) m { i , j } : ( A i , m { i } ) ⟶ ( A j , m { j } ) is a morphism of unital A ∞ A_\infty A ∞ -algebras for all $0 \le i < j \le N$. We require this to be a unital morphism. That is, we have
m 1 { i , j } ( 1 i ) = 1 j , m^{\{i,j\}}_1(1^i) = 1^j, m 1 { i , j } ( 1 i ) = 1 j ,
m k { i , j } ∘ ( id A i ⊗ l − 1 ⊗ 1 i ⊗ id A i ⊗ k − l ) = 0 , all 1 ≤ l ≤ k , k ≠ 1. (179) m^{\{i,j\}}_k \circ(\operatorname{id}_{A_i}^{\otimes l-1} \otimes1^i \otimes\operatorname{id}_{A_i}^{\otimes k-l}) = 0,\quad\text{all } 1 \le l \le k,\ k \ne1. \tag*{(179)} m k { i , j } ∘ ( id A i ⊗ l − 1 ⊗ 1 i ⊗ id A i ⊗ k − l ) = 0 , all 1 ≤ l ≤ k , k = 1. (179)
(v) For all I ⊆ { 0 , … , N } I \subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } with ∣ I ∣ ≥ 3 |I| \ge3 ∣ I ∣ ≥ 3 , which we write as I = { i 0 , i 1 , … , i M } I = \{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0 < i_1 < \cdots< i_M i 0 < i 1 < ⋯ < i M with M ≥ 2 M \ge2 M ≥ 2 , we have
m k I ∘ ( id A i 0 ⊗ l − 1 ⊗ 1 i 0 ⊗ id A i 0 ⊗ k − l ) = 0 , all 1 ≤ l ≤ k . (180) m^I_k \circ(\operatorname{id}_{A_{i_0}}^{\otimes l-1} \otimes1^{i_0} \otimes\operatorname{id}_{A_{i_0}}^{\otimes k-l}) = 0,\quad\text{all } 1 \le l \le k. \tag*{(180)} m k I ∘ ( id A i 0 ⊗ l − 1 ⊗ 1 i 0 ⊗ id A i 0 ⊗ k − l ) = 0 , all 1 ≤ l ≤ k . (180)
(b) If α : [ N ‾ ] → [ N ] \alpha\colon[\overline{N}] \to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , that is, α : { 0 , … , N ‾ } → { 0 , … , N } \alpha\colon\{0,\ldots,\overline{N}\} \to\{0,\ldots,N\} α : { 0 , … , N } → { 0 , … , N } is a nondecreasing function, define
A ∞ Alg ^ R u n , ∞ ( α ) : ( A ∞ Alg ^ R u n , ∞ ) N ⟶ ( A ∞ Alg ^ R u n , ∞ ) N ‾ \widehat{A_\infty\operatorname{Alg}}^{\mathrm{un},\infty}_{R}(\alpha)\colon\left(\widehat{A_\infty\operatorname{Alg}}^{\mathrm{un},\infty}_{R}\right)_N \longrightarrow\left(\widehat{A_\infty\operatorname{Alg}}^{\mathrm{un},\infty}_{R}\right)_{\overline{N}} A ∞ Alg R un , ∞ ( α ) : ( A ∞ Alg R un , ∞ ) N ⟶ ( A ∞ Alg R un , ∞ ) N
to map ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) to ( ( A α ( i ) ) i = 0 N ‾ , ( m ‾ I ‾ ) ∅ ≠ I ‾ ⊆ { 0 , … , N ‾ } ) \left((A^{\alpha(i)})_{i=0}^{\overline{N}},(\overline{m}^{\overline{I}})_{\varnothing\ne\overline{I}\subseteq\{0,\ldots,\overline{N}\}}\right) ( ( A α ( i ) ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where m ‾ I ‾ = ( m ‾ k I ‾ ) k ≥ 0 \overline{m}^{\overline{I}} = (\overline{m}^{\overline{I}}_k)_{k\ge0} m I = ( m k I ) k ≥ 0 with
m ‾ k I ‾ = { m k α ( I ‾ ) , if α ∣ I ‾ is injective , id A i , if k = 1 and I ‾ = { i ‾ 0 , i ‾ 1 } with i ‾ 0 ≠ i ‾ 1 , α ( i ‾ 0 ) = α ( i ‾ 1 ) = i , 1 2 i , if k = 0 , I ‾ = { i ‾ 0 < i ‾ 1 < i ‾ 2 } and α ∣ I ‾ is not injective , 0 , otherwise. (181) \overline{m}^{\overline{I}}_k =
\begin{cases}
m^{\alpha(\overline{I})}_k, & \text{if }\alpha|_{\overline{I}}\text{ is injective},\\
\operatorname{id}_{A_i}, & \text{if }k=1\text{ and }\overline{I}=\{\overline{i}_0,\overline{i}_1\}\text{ with }\overline{i}_0\ne\overline{i}_1,\ \alpha(\overline{i}_0)=\alpha(\overline{i}_1)=i,\\
1^i_2, & \text{if }k=0,\ \overline{I}=\{\overline{i}_0<\overline{i}_1<\overline{i}_2\}\text{ and }\alpha|_{\overline{I}}\text{ is not injective},\\
0, & \text{otherwise.}
\end{cases} \tag*{(181)} m k I = ⎩ ⎨ ⎧ m k α ( I ) , id A i , 1 2 i , 0 , if α ∣ I is injective , if k = 1 and I = { i 0 , i 1 } with i 0 = i 1 , α ( i 0 ) = α ( i 1 ) = i , if k = 0 , I = { i 0 < i 1 < i 2 } and α ∣ I is not injective , otherwise. (181)
Here by m ‾ 0 { ı ˉ 0 , ı ˉ 1 , ı ˉ 2 } = l ‾ ı ˉ 2 \overline{m}^{\{\bar{\imath}_0,\bar{\imath}_1,\bar{\imath}_2\}}_0=\overline{l}^{\bar{\imath}_2} m 0 { ˉ 0 , ˉ 1 , ˉ 2 } = l ˉ 2 we mean that m ‾ 0 { ı ˉ 0 , ı ˉ 1 , ı ˉ 2 } : R → A ı ˉ 2 \overline{m}^{\{\bar{\imath}_0,\bar{\imath}_1,\bar{\imath}_2\}}_0:R\to A^{\bar{\imath}_2} m 0 { ˉ 0 , ˉ 1 , ˉ 2 } : R → A ˉ 2 maps λ ↦ λ 1 ı ˉ 2 \lambda\mapsto\lambda1^{\bar{\imath}_2} λ ↦ λ 1 ˉ 2 .
Proposition 6.18 shows that A ^ ∞ A l g R u n , ∞ ( α ) \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R(\alpha) A ∞ Alg R un , ∞ ( α ) does map to ( A ^ ∞ A l g R u n , ∞ ) N ˉ (\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R)_{\bar{N}} ( A ∞ Alg R un , ∞ ) N ˉ . It is easy to check that if β : [ N ˇ ] → [ N ˉ ] \beta:[\check{N}]\to[\bar{N}] β : [ N ˇ ] → [ N ˉ ] is another morphism in Δ \Delta Δ then
A ^ ∞ A l g R u n , ∞ ( α ∘ β ) = A ^ ∞ A l g R u n , ∞ ( β ) ∘ A ^ ∞ A l g R u n , ∞ ( α ) : \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R(\alpha\circ\beta)=\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R(\beta)\circ\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R(\alpha): A ∞ Alg R un , ∞ ( α ∘ β ) = A ∞ Alg R un , ∞ ( β ) ∘ A ∞ Alg R un , ∞ ( α ) :
( A ^ ∞ A l g R u n , ∞ ) N ⟶ ( A ^ ∞ A l g R u n , ∞ ) N ˇ . (\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R)_N\longrightarrow(\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R)_{\check{N}}. ( A ∞ Alg R un , ∞ ) N ⟶ ( A ∞ Alg R un , ∞ ) N ˇ .
Also A ^ ∞ A l g R u n , ∞ ( id [ N ] ) = id ( A ^ ∞ A l g R u n , ∞ ) N \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R(\operatorname{id}_{[N]})=\operatorname{id}_{(\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R)_N} A ∞ Alg R un , ∞ ( id [ N ] ) = id ( A ∞ Alg R un , ∞ ) N . Hence A ^ ∞ A l g R u n , ∞ : Δ o p → S e t \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R:\Delta^{\mathrm{op}}\to\mathrm{Set} A ∞ Alg R un , ∞ : Δ op → Set is a functor, that is, A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ is a simplicial set.
The next proposition will be proved in §10.2.
Proposition 6.18. In Definition 6.16(b), ( ( A α ( i ) ) i = 0 N ˉ , ( m ˉ I ) ∅ ≠ I ⊆ { 0 , … , N ˉ } ) ((A^{\alpha(i)})_{i=0}^{\bar{N}},(\bar{m}^{I})_{\varnothing\neq I\subseteq\{0,\ldots,\bar{N}\}}) (( A α ( i ) ) i = 0 N ˉ , ( m ˉ I ) ∅ = I ⊆ { 0 , … , N ˉ } ) lies in ( A ^ ∞ A l g R u n , ∞ ) N ˉ (\widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R)_{\bar{N}} ( A ∞ Alg R un , ∞ ) N ˉ .
The next theorem will be proved in §10.3.
Theorem 6.19. Let ( ( A i ) i = 0 N , ( m I ) I ⊆ { 0 , … , N } ) ((A^i)_{i=0}^{N},(m^I)_{I\subseteq\{0,\ldots,N\}}) (( A i ) i = 0 N , ( m I ) I ⊆ { 0 , … , N } ) be an N N N -simplex in A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ as in Definition 6.16, and define maps d n , k I d^I_{n,k} d n , k I from the m k I m^I_k m k I as in (167) –(170) . Then for each ∅ ≠ I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } \varnothing\neq I=\{i_0<i_1<\cdots<i_M\}\subseteq\{0,\ldots,N\} ∅ = I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , in R R R -linear maps ( A i 0 ) ⊗ n → ( A i M ) ⊗ k (A^{i_0})^{\otimes n}\to(A^{i_M})^{\otimes k} ( A i 0 ) ⊗ n → ( A i M ) ⊗ k we have
∑ l = k n d l , k { i 0 } ∘ d n , l { i 0 } = 0 if M = 0 , \sum_{l=k}^{n}d^{\{i_0\}}_{l,k}\circ d^{\{i_0\}}_{n,l}=0\qquad\text{if }M=0, l = k ∑ n d l , k { i 0 } ∘ d n , l { i 0 } = 0 if M = 0 , ∑ l = 1 n d l , k I ∘ d n , l I + ( − 1 ) M ∑ l = k n + M − 1 d l , k { i M } ∘ d n , l I = 0 if M > 0. (182) \sum_{l=1}^{n}d^{I}_{l,k}\circ d^{I}_{n,l}+(-1)^M\sum_{l=k}^{n+M-1}d^{\{i_M\}}_{l,k}\circ d^{I}_{n,l}=0\qquad\text{if }M>0. \tag*{(182)} l = 1 ∑ n d l , k I ∘ d n , l I + ( − 1 ) M l = k ∑ n + M − 1 d l , k { i M } ∘ d n , l I = 0 if M > 0. (182)
Here the limits in the sums in (182) are unnecessary and could be replaced by ∑ l = 0 ∞ ⋯ \sum_{l=0}^{\infty}\cdots ∑ l = 0 ∞ ⋯ , as for l l l outside these limits the summands are zero by (167) –(170) and Lemma 6.14.
Example 6.20. (a) Let ( ( A 0 ) , ( m { 0 } ) ) ((A^0),(m^{\{0\}})) (( A 0 ) , ( m { 0 } )) be a 0-simplex of A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ . Then (169) and (177) for I = { 0 } I=\{0\} I = { 0 } say that
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) = 0. \sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk}m^{\{0\}}_{i+k+1}\circ\left(\operatorname{id}_{A^0}^{\otimes i}\otimes m^{\{0\}}_j\otimes\operatorname{id}_{A^0}^{\otimes k}\right)=0. i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) = 0. Thus by (62) , ( A 0 , m { 0 } ) (A^0,m^{\{0\}}) ( A 0 , m { 0 } ) is an A ∞ A_{\infty} A ∞ -algebra over R R R . Definition 6.16(a)(iii) says that ( A 0 , m { 0 } ) (A^0,m^{\{0\}}) ( A 0 , m { 0 } ) is unital.
(b) A 1-simplex ( ( A 0 , A 1 ) , ( m { 0 } , m { 1 } , m { 0 , 1 } ) ) \left((A^0,A^1),(\mathfrak m^{\{0\}},\mathfrak m^{\{1\}},\mathfrak m^{\{0,1\}})\right) ( ( A 0 , A 1 ) , ( m { 0 } , m { 1 } , m { 0 , 1 } ) ) of A ^ ∞ Alg R u n , ∞ \widehat{\mathcal A}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ consists of unital A ∞ A_\infty A ∞ -algebras ( A 0 , m { 0 } ) , ( A 1 , m { 1 } ) (A^0,\mathfrak m^{\{0\}}),(A^1,\mathfrak m^{\{1\}}) ( A 0 , m { 0 } ) , ( A 1 , m { 1 } ) over R R R by (a), and R R R -linear maps
m k { 0 , 1 } : ( A 0 ) ⊗ k ⟶ A 1 \mathfrak m_k^{\{0,1\}}:(A^0)^{\otimes k}\longrightarrow A^1 m k { 0 , 1 } : ( A 0 ) ⊗ k ⟶ A 1
for all k ≥ 1 k\geq1 k ≥ 1 , graded of degree $1-k, s a t i s f y i n g b y ( 6.10 ) , ( 6.11 ) , ( 6.18 ) f o r , satisfying by (6.10), (6.11), (6.18) for , s a t i s f y in g b y ( 6.10 ) , ( 6.11 ) , ( 6.18 ) f or n\geq 1
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 } ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) − ∑ n = n 1 + ⋯ + n k , n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 1 } ( m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } ) = 0 , \sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}\mathfrak m_{i+k+1}^{\{0,1\}}\left(\operatorname{id}_{A^0}^{\otimes i}\otimes\mathfrak m_j^{\{0\}}\otimes\operatorname{id}_{A^0}^{\otimes k}\right)
-\sum_{\substack{n=n_1+\cdots+n_k,\ n_a\geq1}}(-1)^{\sum_{i=1}^k(n_i+1)(k-i)}\mathfrak m_k^{\{1\}}\left(\mathfrak m_{n_1}^{\{0,1\}}\otimes\mathfrak m_{n_2}^{\{0,1\}}\otimes\cdots\otimes\mathfrak m_{n_k}^{\{0,1\}}\right)=0, i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 } ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) − n = n 1 + ⋯ + n k , n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 1 } ( m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } ) = 0 ,
after simplifying signs. By (3.7), this says m { 0 , 1 } \mathfrak m^{\{0,1\}} m { 0 , 1 } is a morphism ( A 0 , m { 0 } ) ⟶ ( A 1 , m { 1 } ) (A^0,\mathfrak m^{\{0\}})\longrightarrow(A^1,\mathfrak m^{\{1\}}) ( A 0 , m { 0 } ) ⟶ ( A 1 , m { 1 } ) of A ∞ A_\infty A ∞ -algebras over R R R . Definition 6.16(a)(iv) says that m { 0 , 1 } \mathfrak m^{\{0,1\}} m { 0 , 1 } is unital. Thus 1-simplices of A ^ ∞ Alg R u n , ∞ \widehat{\mathcal A}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ are equivalent to unital morphisms of unital A ∞ A_\infty A ∞ -algebras over R R R .
(c) A 2-simplex of A ^ ∞ Alg R u n , ∞ \widehat{\mathcal A}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ consists of unital A ∞ A_\infty A ∞ -algebras ( A 0 , m { 0 } ) , ( A 1 , m { 1 } ) , ( A 2 , m { 2 } ) (A^0,\mathfrak m^{\{0\}}),(A^1,\mathfrak m^{\{1\}}),(A^2,\mathfrak m^{\{2\}}) ( A 0 , m { 0 } ) , ( A 1 , m { 1 } ) , ( A 2 , m { 2 } ) over R R R as in (a), and unital morphisms m { i , j } : ( A i , m { i } ) ⟶ ( A j , m { j } ) \mathfrak m^{\{i,j\}}:(A^i,\mathfrak m^{\{i\}})\longrightarrow(A^j,\mathfrak m^{\{j\}}) m { i , j } : ( A i , m { i } ) ⟶ ( A j , m { j } ) for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), and R R R -linear maps
m k { 0 , 1 , 2 } : ( A 0 ) ⊗ k ⟶ A 2 \mathfrak m_k^{\{0,1,2\}}:(A^0)^{\otimes k}\longrightarrow A^2 m k { 0 , 1 , 2 } : ( A 0 ) ⊗ k ⟶ A 2
for k = 0 , 1 , 2 , … k=0,1,2,\ldots k = 0 , 1 , 2 , … , graded of degree − k -k − k , satisfying (6.21) and for all $n\geq 0
\begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}\mathfrak m_{i+k+1}^{\{0,1,2\}}\circ\left(\operatorname{id}_{A^0}^{\otimes i}\otimes\mathfrak m_j^{\{0\}}\otimes\operatorname{id}_{A^0}^{\otimes k}\right)\\
&+\sum_{\substack{1\leq l\leq k\leq n+1,\ n=n_1+\cdots+n_k\\ n_i=p_i^1+\cdots+p_i^{m_i},\ i=1,\ldots,l-1\\ p_i^j\geq1,\ m_i\geq1,\ n_l\geq0,\ n_i\geq1,\ i\ne l}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\cdot\prod_{a=1}^{l-1}(-1)^{\sum_{a=1}^{m_i}(p_i^a+1)(m_i-a)}\\
&\qquad\mathfrak m_k^{\{2\}}\left[\mathfrak m_{m_1}^{\{1,2\}}\left(\mathfrak m_{p_1^1}^{\{0,1\}}\otimes\mathfrak m_{p_1^2}^{\{0,1\}}\otimes\cdots\otimes\mathfrak m_{p_1^{m_1}}^{\{0,1\}}\right)\otimes\cdots\otimes\mathfrak m_{m_{l-1}}^{\{1,2\}}\left(\mathfrak m_{p_{l-1}^1}^{\{0,1\}}\otimes\mathfrak m_{p_{l-1}^2}^{\{0,1\}}\otimes\cdots\otimes\mathfrak m_{p_{l-1}^{m_{l-1}}}^{\{0,1\}}\right)\\
&\qquad\otimes\mathfrak m_{n_l}^{\{0,1,2\}}\otimes\mathfrak m_{n_{l+1}}^{\{0,2\}}\otimes\cdots\otimes\mathfrak m_{n_k}^{\{0,2\}}\right]=0, \tag*{(183)}
\end{aligned}
after simplifying signs. If we write f = m { 1 , 2 } ∘ m { 0 , 1 } f=\mathfrak m^{\{1,2\}}\circ\mathfrak m^{\{0,1\}} f = m { 1 , 2 } ∘ m { 0 , 1 } and g = m { 0 , 2 } g=\mathfrak m^{\{0,2\}} g = m { 0 , 2 } as morphisms f , g : ( A 0 , m { 0 } ) ⟶ ( A 2 , m { 2 } ) f,g:(A^0,\mathfrak m^{\{0\}})\longrightarrow(A^2,\mathfrak m^{\{2\}}) f , g : ( A 0 , m { 0 } ) ⟶ ( A 2 , m { 2 } ) , then we can rewrite (6.24) as
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) + ∑ 1 ≤ l ≤ k ≤ n + 1 , n = n 1 + ⋯ + n k n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 p i j ≥ 1 , m i ≥ 1 , n l ≥ 0 , n i ≥ 1 , i ≠ l ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 2 } [ f n 1 ⊗ ⋯ ⊗ f n l − 1 ⊗ m n l { 0 , 1 , 2 } ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ] = 0. (184) \begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}\mathfrak m_{i+k+1}^{\{0,1,2\}}\circ\left(\operatorname{id}_{A^0}^{\otimes i}\otimes\mathfrak m_j^{\{0\}}\otimes\operatorname{id}_{A^0}^{\otimes k}\right)\\
&+\sum_{\substack{1\leq l\leq k\leq n+1,\ n=n_1+\cdots+n_k\\ n_i=p_i^1+\cdots+p_i^{m_i},\ i=1,\ldots,l-1\\ p_i^j\geq1,\ m_i\geq1,\ n_l\geq0,\ n_i\geq1,\ i\ne l}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\\
&\qquad\mathfrak m_k^{\{2\}}\left[f_{n_1}\otimes\cdots\otimes f_{n_{l-1}}\otimes\mathfrak m_{n_l}^{\{0,1,2\}}\otimes g_{n_{l+1}}\otimes\cdots\otimes g_{n_k}\right]=0. \tag*{(184)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) + 1 ≤ l ≤ k ≤ n + 1 , n = n 1 + ⋯ + n k n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 p i j ≥ 1 , m i ≥ 1 , n l ≥ 0 , n i ≥ 1 , i = l ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 2 } [ f n 1 ⊗ ⋯ ⊗ f n l − 1 ⊗ m n l { 0 , 1 , 2 } ⊗ g n l + 1 ⊗ ⋯ ⊗ g n k ] = 0. (184)
Regarding ( A 0 , m { 0 } ) (A^0,\mathsf{m}^{\{0\}}) ( A 0 , m { 0 } ) , ( A 2 , m { 2 } ) (A^2,\mathsf{m}^{\{2\}}) ( A 2 , m { 2 } ) as unital A ∞ A_\infty A ∞ -categories with one object, this is a unital A ∞ A_\infty A ∞ -natural transformation η : f ⇒ g \eta:\mathsf{f}\Rightarrow\mathsf{g} η : f ⇒ g as in Definition 4.22.
In particular, if ( ( A i ) i = 0 2 , ( m I ) ∅ ≠ I ⊆ { 0 , 1 , 2 } ) \left((A^i)_{i=0}^2,(\mathsf{m}^I)_{\varnothing\ne I\subseteq\{0,1,2\}}\right) ( ( A i ) i = 0 2 , ( m I ) ∅ = I ⊆ { 0 , 1 , 2 } ) is a 2-simplex in A ∞ A l g ^ R u n , ∞ \widehat{A_\infty\mathrm{Alg}}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ with m 0 { 0 , 1 , 2 } = 1 2 \mathsf{m}_0^{\{0,1,2\}}=\mathbb{1}^2 m 0 { 0 , 1 , 2 } = 1 2 and m k { 0 , 1 , 2 } = 0 \mathsf{m}_k^{\{0,1,2\}}=0 m k { 0 , 1 , 2 } = 0 for k > 0 k>0 k > 0 , so that η = id m { 0 , 2 } \eta=\operatorname{id}_{\mathsf{m}^{\{0,2\}}} η = id m { 0 , 2 } , then m { 0 , 2 } = m { 1 , 2 } ∘ m { 0 , 1 } \mathsf{m}^{\{0,2\}}=\mathsf{m}^{\{1,2\}}\circ\mathsf{m}^{\{0,1\}} m { 0 , 2 } = m { 1 , 2 } ∘ m { 0 , 1 } as morphisms of A ∞ A_\infty A ∞ -algebras ( A 0 , m 0 ) ⟶ ( A 2 , m 2 ) (A^0,\mathsf{m}^0)\longrightarrow(A^2,\mathsf{m}^2) ( A 0 , m 0 ) ⟶ ( A 2 , m 2 ) . This is how the notion of composition of morphisms of A ∞ A_\infty A ∞ -algebras in Definition 3.8 is encoded in our theory.
When n = 0 n=0 n = 0 , equation (6.24) becomes
m 1 { 2 } ∘ m 0 { 0 , 1 , 2 } = 0 : R ⟶ A 2 , 1 . \mathsf{m}^{\{2\}}_1\circ\mathsf{m}^{\{0,1,2\}}_0=0:R\longrightarrow A^{2,1}. m 1 { 2 } ∘ m 0 { 0 , 1 , 2 } = 0 : R ⟶ A 2 , 1 .
Thus m 0 { 0 , 1 , 2 } ( 1 ) ∈ A 2 , 0 \mathsf{m}^{\{0,1,2\}}_0(1)\in A^{2,0} m 0 { 0 , 1 , 2 } ( 1 ) ∈ A 2 , 0 defines a cohomology class
[ m 0 { 0 , 1 , 2 } ( 1 ) ] ∈ H 0 ( A 2 , m 1 { 2 } ) . (185) [\mathsf{m}^{\{0,1,2\}}_0(1)]\in H^0(A^2,\mathsf{m}^{\{2\}}_1). \tag*{(185)} [ m 0 { 0 , 1 , 2 } ( 1 )] ∈ H 0 ( A 2 , m 1 { 2 } ) . (185)
Note that H ∗ ( A 2 , m 1 { 2 } ) H^*(A^2,\mathsf{m}^{\{2\}}_1) H ∗ ( A 2 , m 1 { 2 } ) is a graded unital R R R -algebra, as in §3.2.
When n = 1 n=1 n = 1 , equation (6.24) becomes
m 1 { 0 , 1 , 2 } ∘ m 1 { 0 } + m 1 { 2 } ∘ m 1 { 0 , 1 , 2 } − m 2 { 2 } ( ( m 1 { 1 , 2 } ∘ m 1 { 0 , 1 } ) ⊗ m 0 { 0 , 1 , 2 } ( 1 ) ) + m 2 { 2 } ( m 0 { 0 , 1 , 2 } ( 1 ) ⊗ m 1 { 0 , 2 } ) = 0. \mathsf{m}^{\{0,1,2\}}_1\circ\mathsf{m}^{\{0\}}_1+\mathsf{m}^{\{2\}}_1\circ\mathsf{m}^{\{0,1,2\}}_1-\mathsf{m}^{\{2\}}_2\left((\mathsf{m}^{\{1,2\}}_1\circ\mathsf{m}^{\{0,1\}}_1)\otimes\mathsf{m}^{\{0,1,2\}}_0(1)\right)+\mathsf{m}^{\{2\}}_2\left(\mathsf{m}^{\{0,1,2\}}_0(1)\otimes\mathsf{m}^{\{0,2\}}_1\right)=0. m 1 { 0 , 1 , 2 } ∘ m 1 { 0 } + m 1 { 2 } ∘ m 1 { 0 , 1 , 2 } − m 2 { 2 } ( ( m 1 { 1 , 2 } ∘ m 1 { 0 , 1 } ) ⊗ m 0 { 0 , 1 , 2 } ( 1 ) ) + m 2 { 2 } ( m 0 { 0 , 1 , 2 } ( 1 ) ⊗ m 1 { 0 , 2 } ) = 0.
By taking cohomology, we can deduce that
( H ∗ ( m { 1 , 2 } ) ∘ H ∗ ( m { 0 , 1 } ) ) ⋅ [ m 0 { 0 , 1 , 2 } ( 1 ) ] = [ m 0 { 0 , 1 , 2 } ( 1 ) ] ⋅ H ∗ ( m { 0 , 2 } ) \left(H^*(\mathsf{m}^{\{1,2\}})\circ H^*(\mathsf{m}^{\{0,1\}})\right)\cdot[\mathsf{m}^{\{0,1,2\}}_0(1)]=[\mathsf{m}^{\{0,1,2\}}_0(1)]\cdot H^*(\mathsf{m}^{\{0,2\}}) ( H ∗ ( m { 1 , 2 } ) ∘ H ∗ ( m { 0 , 1 } ) ) ⋅ [ m 0 { 0 , 1 , 2 } ( 1 )] = [ m 0 { 0 , 1 , 2 } ( 1 )] ⋅ H ∗ ( m { 0 , 2 } )
in maps H ∗ ( A 0 , m { 0 } ) ⟶ H ∗ ( A 2 , m { 2 } ) H^*(A^0,\mathsf{m}^{\{0\}})\longrightarrow H^*(A^2,\mathsf{m}^{\{2\}}) H ∗ ( A 0 , m { 0 } ) ⟶ H ∗ ( A 2 , m { 2 } ) ,
where we first map H ∗ ( A 0 ) ⟶ H ∗ ( A 2 ) H^*(A^0)\longrightarrow H^*(A^2) H ∗ ( A 0 ) ⟶ H ∗ ( A 2 ) , and then multiply on the right or left with [ m 0 { 0 , 1 , 2 } ( 1 ) ] [\mathsf{m}^{\{0,1,2\}}_0(1)] [ m 0 { 0 , 1 , 2 } ( 1 )] in the R R R -algebra H ∗ ( A 2 ) H^*(A^2) H ∗ ( A 2 ) .
(d) For a 3-simplex, equation (6.18) for I = { 0 , 1 , 2 , 3 } I=\{0,1,2,3\} I = { 0 , 1 , 2 , 3 } and n = 0 n=0 n = 0 becomes
m 1 { 3 } ∘ m 0 { 0 , 1 , 2 , 3 } + m 2 { 3 } ( ( m 1 { 2 , 3 } ∘ m 0 { 0 , 1 , 2 } ( 1 ) ) ⊗ m 0 { 0 , 2 , 3 } ( 1 ) ) − m 2 { 3 } ( m 0 { 1 , 2 , 3 } ( 1 ) ⊗ m 0 { 0 , 1 , 3 } ( 1 ) ) = 0. \mathsf{m}^{\{3\}}_1\circ\mathsf{m}^{\{0,1,2,3\}}_0+\mathsf{m}^{\{3\}}_2\left((\mathsf{m}^{\{2,3\}}_1\circ\mathsf{m}^{\{0,1,2\}}_0(1))\otimes\mathsf{m}^{\{0,2,3\}}_0(1)\right)-\mathsf{m}^{\{3\}}_2\left(\mathsf{m}^{\{1,2,3\}}_0(1)\otimes\mathsf{m}^{\{0,1,3\}}_0(1)\right)=0. m 1 { 3 } ∘ m 0 { 0 , 1 , 2 , 3 } + m 2 { 3 } ( ( m 1 { 2 , 3 } ∘ m 0 { 0 , 1 , 2 } ( 1 )) ⊗ m 0 { 0 , 2 , 3 } ( 1 ) ) − m 2 { 3 } ( m 0 { 1 , 2 , 3 } ( 1 ) ⊗ m 0 { 0 , 1 , 3 } ( 1 ) ) = 0.
By taking cohomology, we see that in the R R R -algebra H 0 ( A 3 , m 1 { 3 } ) H^0(A^3,\mathsf{m}^{\{3\}}_1) H 0 ( A 3 , m 1 { 3 } ) we have
H ∗ ( m { 2 , 3 } ) ( [ m 0 { 0 , 1 , 2 } ( 1 ) ] ) ⋅ [ m 0 { 0 , 2 , 3 } ( 1 ) ] = [ m 0 { 1 , 2 , 3 } ( 1 ) ] ⋅ [ m 0 { 0 , 1 , 3 } ( 1 ) ] . H^*(\mathsf{m}^{\{2,3\}})([\mathsf{m}^{\{0,1,2\}}_0(1)])\cdot[\mathsf{m}^{\{0,2,3\}}_0(1)]=[\mathsf{m}^{\{1,2,3\}}_0(1)]\cdot[\mathsf{m}^{\{0,1,3\}}_0(1)]. H ∗ ( m { 2 , 3 } ) ([ m 0 { 0 , 1 , 2 } ( 1 )]) ⋅ [ m 0 { 0 , 2 , 3 } ( 1 )] = [ m 0 { 1 , 2 , 3 } ( 1 )] ⋅ [ m 0 { 0 , 1 , 3 } ( 1 )] .
Definition 6.21. Define a simplicial subset A ∞ A l g R u n , ∞ ⊂ A ∞ A l g ^ R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}\subset\widehat{A_{\infty}\mathrm{Alg}}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ ⊂ A ∞ Alg R un , ∞ by an N N N -simplex ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^N,(\mathsf{m}^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ∞ A l g ^ R u n , ∞ \widehat{A_{\infty}\mathrm{Alg}}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ lies in A ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ if and only if whenever 0 ≤ i 0 < i 1 < i 2 ≤ N 0\le i_0<i_1<i_2\le N 0 ≤ i 0 < i 1 < i 2 ≤ N then the cohomology class [ m 0 { i 0 , i 1 , i 2 } ( 1 ) ] [\mathsf{m}^{\{i_0,i_1,i_2\}}_0(1)] [ m 0 { i 0 , i 1 , i 2 } ( 1 )] from (6.26) is invertible in the unital R R R -algebra H 0 ( A i 2 , m 1 { i 2 } ) H^0(A^{i_2},\mathsf{m}^{\{i_2\}}_1) H 0 ( A i 2 , m 1 { i 2 } ) .
If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ then A ∞ A l g ^ R u n , ∞ ( α ) \widehat{A_{\infty}\mathrm{Alg}}_{R}^{\mathrm{un},\infty}(\alpha) A ∞ Alg R un , ∞ ( α ) in Definition 6.16(b) maps ( A ∞ A l g R u n , ∞ ) N ⟶ ( A ∞ A l g R u n , ∞ ) N ‾ (A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty})_N\longrightarrow(A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty})_{\overline{N}} ( A ∞ Alg R un , ∞ ) N ⟶ ( A ∞ Alg R un , ∞ ) N , so A ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ is a simplicial set. We also write A ∞ A l g R u n , ∞ = A ∞ A l g R , 2 u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}=A_{\infty}\mathrm{Alg}_{R,2}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ = A ∞ Alg R , 2 un , ∞ to fit with the ∞ \infty ∞ -categories A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ in Definition 6.25.
The next theorem is proved in §10.4.
Theorem 6.22. In Definition 6.21, A ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ is an ∞ \infty ∞ -category.
In §2.8, following Lurie [62, §4], we defined ∞ \infty ∞ -bicategories, a kind of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category. The next theorem will be proved in §10.5.
Theorem 6.23. In the simplicial set A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ in Definition 6.16, define a 2-simplex to be thin if the cohomology class [ m 0 { 0 , 1 , 2 } ( 1 ) ] [m_{0}^{\{0,1,2\}}(1)] [ m 0 { 0 , 1 , 2 } ( 1 )] from (6.26) is invertible in the unital R R R -algebra H 0 ( A 2 , m 0 { 2 } ) H^{0}(A^{2},m_{0}^{\{2\}}) H 0 ( A 2 , m 0 { 2 } ) , that is, if it is a 2-simplex in A ∞ A l g R u n , ∞ ⊂ A ^ ∞ C a t R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}\subset\widehat{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ ⊂ A ∞ Cat R un , ∞ . Write T ⊂ A ^ ∞ A l g R u n , ∞ ( [ 2 ] ) T\subset\widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}([2]) T ⊂ A ∞ Alg R un , ∞ ([ 2 ]) for the set of thin 2-simplices. Then ( A ^ ∞ A l g R u n , ∞ , T ) (\widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty},T) ( A ∞ Alg R un , ∞ , T ) is an ∞ \infty ∞ -bicategory in the sense of §2.8. We have A ∞ A l g R u n , ∞ = Core ( A ^ ∞ A l g R u n , ∞ , T ) A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}=\operatorname{Core}(\widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty},T) A ∞ Alg R un , ∞ = Core ( A ∞ Alg R un , ∞ , T ) , as in Definition 2.41.
Remark 6.24. Theorem 6.22 follows trivially from Theorem 6.23, as the core of an ∞ \infty ∞ -bicategory is an ∞ \infty ∞ -category. But we use the proof of Theorem 6.22 in that of Theorem 6.23, so we prove Theorem 6.22 first.
Definition 6.25. Let K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Define simplicial subsets
A ∞ A l g R , K u n , ∞ ⊂ A ∞ A l g R u n , ∞ ⊂ A ^ ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty}\subset A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}\subset\widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ ⊂ A ∞ Alg R un , ∞ ⊂ A ∞ Alg R un , ∞
such that an N N N -simplex ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((A^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ in Definition 6.16(a) lies in A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ if for all 0 ≤ i 0 < i 1 < i 2 ≤ N 0\le i_{0}<i_{1}<i_{2}\le N 0 ≤ i 0 < i 1 < i 2 ≤ N we have m 0 { i 0 , i 1 , i 2 } : λ ↦ λ 1 i 2 m_{0}^{\{i_{0},i_{1},i_{2}\}}:\lambda\mapsto\lambda1_{i_{2}} m 0 { i 0 , i 1 , i 2 } : λ ↦ λ 1 i 2 , and m 0 I = 0 m_{0}^{I}=0 m 0 I = 0 for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } with ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K and ∣ I ∣ ≠ 3 |I|\ne3 ∣ I ∣ = 3 .
This condition is preserved by the maps in Definition 6.16(b), and so defines a simplicial subset A ∞ A l g R , K u n , ∞ ⊂ A ^ ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty}\subset\widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ ⊂ A ∞ Alg R un , ∞ . As [ m 0 { i 0 , i 1 , i 2 } ( 1 ) ] = [ 1 i 2 ] [m_{0}^{\{i_{0},i_{1},i_{2}\}}(1)]=[1_{i_{2}}] [ m 0 { i 0 , i 1 , i 2 } ( 1 )] = [ 1 i 2 ] is invertible in H 0 ( A i 2 , m 0 { i 2 } ) H^{0}(A^{i_{2}},m_{0}^{\{i_{2}\}}) H 0 ( A i 2 , m 0 { i 2 } ) we see that A ∞ A l g R , K u n , ∞ ⊂ A ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty}\subset A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ ⊂ A ∞ Alg R un , ∞ . Clearly A ∞ A l g R , K u n , ∞ ⊂ A ∞ A l g R , L u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty}\subset A_{\infty}\mathrm{Alg}_{R,L}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ ⊂ A ∞ Alg R , L un , ∞ if L ≤ K L\le K L ≤ K in { 3 , 4 , 5 , … , ∞ } \{3,4,5,\ldots,\infty\} { 3 , 4 , 5 , … , ∞ } , giving the inclusion relations in the first column of (6.1). Theorem 6.35 will show that A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ is an ∞ \infty ∞ -category. We say that A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ has type K K K .
Remark 6.26. It is not difficult to rewrite the definitions of A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ , A ∞ A l g R u n , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ , A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ above in terms of bar complexes , as in §3.6. Implicitly we do this already, as the operators in the bar complex picture are essentially d I = ∏ k , n d n , k I d^{I}=\prod_{k,n}d_{n,k}^{I} d I = ∏ k , n d n , k I as in Definition 6.11, but with degrees shifted by 1, and Theorem 6.19 gives identities satisfied by these d I d^{I} d I . In the bar complex version some of the signs become simpler, for example in (6.11). But other than this, the author did not find the exercise very illuminating, so we will not write it out here.
Other ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -algebras
We wish to generalize the definition of A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ in §6.3 for K ≥ 3 K\ge3 K ≥ 3 to A ∞ A_{\infty} A ∞ -algebras that are not unital, but only cohomologically unital, or non-unital. The main difficulty in doing this is that the degeneracy maps in the simplicial set A ^ ∞ A l g R u n , ∞ \widehat{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ in (6.22) use units 1 2 ‾ 1_{\overline{2}} 1 2 in m 0 { 0 ‾ , 1 ‾ , 2 ‾ } m_{0}^{\{\overline{0},\overline{1},\overline{2}\}} m 0 { 0 , 1 , 2 } .
Our idea is simple. Suppose ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^{i})_{i=0}^{N},(\mathfrak{m}^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) is an N N N -simplex in A ∞ A l g R , K u n , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ in Definitions 6.16 and 6.21 . We define new maps m ˙ k { i 0 , … , i M } \dot{\mathfrak{m}}_{k}^{\{i_{0},\ldots,i_{M}\}} m ˙ k { i 0 , … , i M } by
m ˙ k { i 0 , … , i M } = { m 0 { i 0 , i 1 , i 2 } − 1 i 2 , k = 0 and M = 2 , m k { i 0 , … , i M } , otherwise . (186) \dot{\mathfrak{m}}_{k}^{\{i_{0},\ldots,i_{M}\}}=
\begin{cases}
\mathfrak{m}_{0}^{\{i_{0},i_{1},i_{2}\}}-\mathord{\mathbf{1}}^{i_{2}}, & k=0\text{ and }M=2,\\
\mathfrak{m}_{k}^{\{i_{0},\ldots,i_{M}\}}, & \text{otherwise}.
\end{cases} \tag*{(186)} m ˙ k { i 0 , … , i M } = { m 0 { i 0 , i 1 , i 2 } − 1 i 2 , m k { i 0 , … , i M } , k = 0 and M = 2 , otherwise . (186)
where 1 i 2 \mathord{\mathbf{1}}^{i_{2}} 1 i 2 means the map R → A i 2 , 0 R\to A^{i_{2},0} R → A i 2 , 0 , λ ↦ λ 1 i 2 \lambda\mapsto\lambda\mathord{\mathbf{1}}^{i_{2}} λ ↦ λ 1 i 2 . Then we rewrite the definition of A ∞ A l g R , K u n , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ in terms of the maps m ˙ k \dot{\mathfrak{m}}_{k} m ˙ k in Definition 6.27, to give A ˙ ∞ A l g R , K u n , ∞ \dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Alg R , K un , ∞ . The result no longer uses units in degeneracy maps, so is suitable for generalizing to the non-unital case.
We will do this only for A ∞ A l g R , K u n , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ for K ≥ 3 K\ge3 K ≥ 3 in Definition 6.25, and not for A ∞ A l g R u n , ∞ = A ∞ A l g R , 2 u n , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}=\mathrm{A}_{\infty}\mathrm{Alg}_{R,2}^{\mathrm{un},\infty} A ∞ Alg R un , ∞ = A ∞ Alg R , 2 un , ∞ in Definition 6.21, so in (6.27) we automatically have m 0 { i 0 , i 1 , i 2 } = 1 i 2 \mathfrak{m}_{0}^{\{i_{0},i_{1},i_{2}\}}=\mathord{\mathbf{1}}^{i_{2}} m 0 { i 0 , i 1 , i 2 } = 1 i 2 and m ˙ 0 { i 0 , i 1 , i 2 } = 0 \dot{\mathfrak{m}}_{0}^{\{i_{0},i_{1},i_{2}\}}=0 m ˙ 0 { i 0 , i 1 , i 2 } = 0 . In fact the approach of this section would also suffice to define an ∞ \infty ∞ -category A ˙ ∞ A l g R u n , ∞ = A ˙ ∞ A l g R , 2 u n , ∞ \dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R}^{\mathrm{un},\infty}=\dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,2}^{\mathrm{un},\infty} A ˙ ∞ Alg R un , ∞ = A ˙ ∞ Alg R , 2 un , ∞ . However, the generalization of this to A ∞ A_{\infty} A ∞ -categories does not work, and as A ∞ A_{\infty} A ∞ -categories are our primary interest, we will not bother.
Here is the analogue of Definitions 6.16 and 6.25.
Definition 6.27. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . We will define a simplicial set A ˙ ∞ A l g R , K u n , ∞ \dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Alg R , K un , ∞ of unital A ∞ A_{\infty} A ∞ -algebras over R R R .
(a) For N ≥ 0 N\ge0 N ≥ 0 , define the N N N -simplices ( A ˙ ∞ A l g R , K u n , ∞ ) N (\dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty})_{N} ( A ˙ ∞ Alg R , K un , ∞ ) N in A ˙ ∞ A l g R , K u n , ∞ \dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Alg R , K un , ∞ to be the set of pairs ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^{i})_{i=0}^{N},(\mathfrak{m}^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where:
(i) A i = ⨁ d ∈ Z A i , d A^{i}=\bigoplus_{d\in\mathbb{Z}}A^{i,d} A i = ⨁ d ∈ Z A i , d is a Z \mathbb{Z} Z -graded R R R -module for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N .
(ii) For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M\ge0 M ≥ 0 , we have m I = ( m k I ) k ≥ 0 \mathfrak{m}^{I}=(\mathfrak{m}_{k}^{I})_{k\ge0} m I = ( m k I ) k ≥ 0 , where m k I \mathfrak{m}_{k}^{I} m k I is an R R R -linear map m k I : ( A i 0 ) ⊗ k → A i M \mathfrak{m}_{k}^{I}:(A^{i_{0}})^{\otimes k}\to A^{i_{M}} m k I : ( A i 0 ) ⊗ k → A i M , graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k . Here when k = 0 k=0 k = 0 , by convention ( A i 0 ) ⊗ 0 = R (A^{i_{0}})^{\otimes0}=R ( A i 0 ) ⊗ 0 = R , and we sometimes identify m 0 I : R → A i M \mathfrak{m}_{0}^{I}:R\to A^{i_{M}} m 0 I : R → A i M with m 0 I ( 1 ) ∈ A i M \mathfrak{m}_{0}^{I}(1)\in A^{i_{M}} m 0 I ( 1 ) ∈ A i M . When k = 0 k=0 k = 0 we require that m 0 I = 0 \mathfrak{m}_{0}^{I}=0 m 0 I = 0 for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } with ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K .
We require that for all I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_{0}<i_{1}<\cdots<i_{M}\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } as above and all n ≥ 0 n\ge0 n ≥ 0 , the maps m k I \mathfrak{m}_{k}^{I} m k I should satisfy
∑ j = 0 M ∑ 0 ≤ k ≤ max ( n , n + j − 1 ) ( − 1 ) j m k { i j , … , i M } d n , k { i 0 , … , i j } = ∑ j = 1 M − 1 ( − 1 ) j m n I ∖ { i j } , (187) \sum_{j=0}^{M}\sum_{0\le k\le\max(n,n+j-1)}(-1)^{j}\mathfrak{m}_{k}^{\{i_{j},\ldots,i_{M}\}}d_{n,k}^{\{i_{0},\ldots,i_{j}\}}
=
\sum_{j=1}^{M-1}(-1)^{j}\mathfrak{m}_{n}^{I\setminus\{i_{j}\}}, \tag*{(187)} j = 0 ∑ M 0 ≤ k ≤ m a x ( n , n + j − 1 ) ∑ ( − 1 ) j m k { i j , … , i M } d n , k { i 0 , … , i j } = j = 1 ∑ M − 1 ( − 1 ) j m n I ∖ { i j } , (187) where d n , k { i 0 , … , i j } d_{n,k}^{\{i_{0},\ldots,i_{j}\}} d n , k { i 0 , … , i j } is defined from the m k I \mathfrak{m}_{k}^{I} m k I as in (6.8)–(6.11), using sums over { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled rooted trees Γ i \Gamma_{i} Γ i . Here the upper limit on k k k is not really necessary, as if it does not hold then d n , k { i 0 , … , i j } = 0 d_{n,k}^{\{i_{0},\ldots,i_{j}\}}=0 d n , k { i 0 , … , i j } = 0 by Lemma 6.14.
(iii) Definition 6.16(iii)–(v) hold.
(b) If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , that is, α : { 0 , … , N ‾ } → { 0 , … , N } \alpha:\{0,\ldots,\overline{N}\}\to\{0,\ldots,N\} α : { 0 , … , N } → { 0 , … , N } is a nondecreasing function, define
A ˙ ∞ A l g R , K u n , ∞ ( α ) : ( A ˙ ∞ A l g R , K u n , ∞ ) N ⟶ ( A ˙ ∞ A l g R , K u n , ∞ ) N ‾ \dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty}(\alpha):
(\dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty})_{N}
\longrightarrow
(\dot{\mathrm{A}}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty})_{\overline{N}} A ˙ ∞ Alg R , K un , ∞ ( α ) : ( A ˙ ∞ Alg R , K un , ∞ ) N ⟶ ( A ˙ ∞ Alg R , K un , ∞ ) N
to map ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^{N},(\mathfrak{m}^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) to ( ( A α ( i ) ) i = 0 N ‾ , ( m ‾ I ‾ ) ∅ ≠ I ‾ ⊆ { 0 , … , N ‾ } ) \left((A^{\alpha(i)})_{i=0}^{\overline{N}},(\overline{\mathfrak{m}}^{\overline{I}})_{\varnothing\ne\overline{I}\subseteq\{0,\ldots,\overline{N}\}}\right) ( ( A α ( i ) ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where m ‾ = ( m ‾ k ) k ≥ 0 \overline{\mathfrak{m}}=(\overline{\mathfrak{m}}_k)_{k\geq0} m = ( m k ) k ≥ 0 with
m ‾ k I ‾ = { m k α ( I ‾ ) if α ∣ I ‾ is injective , id A i if k = 1 and I ‾ = { i ‾ 0 , i ‾ 1 } with i ‾ 0 ≠ i ‾ 1 , α ( i ‾ 0 ) = α ( i ‾ 1 ) = i , 0 otherwise. (188) \overline{\mathfrak{m}}_k^{\overline{I}}=
\begin{cases}
\mathfrak{m}_k^{\alpha(\overline{I})} & \text{if }\alpha|_{\overline{I}}\text{ is injective},\\
\operatorname{id}_{A^i} & \text{if }k=1\text{ and }\overline{I}=\{\overline{i}_0,\overline{i}_1\}\text{ with }\overline{i}_0\ne\overline{i}_1,\ \alpha(\overline{i}_0)=\alpha(\overline{i}_1)=i,\\
0 & \text{otherwise.}
\end{cases} \tag*{(188)} m k I = ⎩ ⎨ ⎧ m k α ( I ) id A i 0 if α ∣ I is injective , if k = 1 and I = { i 0 , i 1 } with i 0 = i 1 , α ( i 0 ) = α ( i 1 ) = i , otherwise. (188)
Proposition 6.28 shows A ˙ ∞ A l g R , K u n , ∞ ( α ) \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K}(\alpha) A ˙ ∞ Alg R , K un , ∞ ( α ) does map to ( A ˙ ∞ A l g R , K u n , ∞ ) N ‾ \left(\dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K}\right)_{\overline{N}} ( A ˙ ∞ Alg R , K un , ∞ ) N , and thus A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is a simplicial set as in Definition 6.16. Theorem 6.35 will show A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is an ∞ \infty ∞ -category. We say that A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ has type K K K . Clearly A ˙ ∞ A l g R , K u n , ∞ ⊂ A ˙ ∞ A l g R , L u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K}\subset\dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,L} A ˙ ∞ Alg R , K un , ∞ ⊂ A ˙ ∞ Alg R , L un , ∞ if L ≤ K L\leq K L ≤ K in { 3 , 4 , 5 , … , ∞ } \{3,4,5,\ldots,\infty\} { 3 , 4 , 5 , … , ∞ } , giving the inclusion relations in the second column of (160) .
Note that (177) and (187) are different, and (181) and (188) are different. The following analogue of Proposition 6.18 is proved in §10.6.
Proposition 6.28. In Definition 6.27 (b), ( ( A α ( i ) ) i = 0 N ‾ , ( m ‾ I ‾ ) ∅ ≠ I ‾ ⊆ { 0 , … , N ‾ } ) \left((A^{\alpha(i)})_{i=0}^{\overline{N}},(\overline{\mathfrak{m}}^{\overline{I}})_{\varnothing\ne\overline{I}\subseteq\{0,\ldots,\overline{N}\}}\right) ( ( A α ( i ) ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) lies in ( A ˙ ∞ A l g R , K u n , ∞ ) N ‾ \left(\dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K}\right)_{\overline{N}} ( A ˙ ∞ Alg R , K un , ∞ ) N .
The next theorem, an analogue of Theorem 6.19 , will be proved in §10.7 using equation (187) .
Theorem 6.29. Let ( ( A i ) , ( m I ) ) \left((A^i),(\mathfrak{m}^I)\right) ( ( A i ) , ( m I ) ) be an N N N -simplex in A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ , as in Definition 6.27 , and define maps d n , k d_{n,k} d n , k from the m k I \mathfrak{m}^I_k m k I as in (167) –(170) . Then for each ∅ ≠ I = { i 0 < ⋯ < i M } ⊆ { 0 , … , N } \varnothing\ne I=\{i_0<\cdots<i_M\}\subseteq\{0,\ldots,N\} ∅ = I = { i 0 < ⋯ < i M } ⊆ { 0 , … , N } and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , in R R R -linear maps ( A i 0 ) ⊗ n → ( A i M ) ⊗ k (A^{i_0})^{\otimes n}\to(A^{i_M})^{\otimes k} ( A i 0 ) ⊗ n → ( A i M ) ⊗ k we have
∑ j = 0 M ∑ 0 ≤ l ≤ max ( n , n + j − 1 ) ( − 1 ) j d l , k { i j , … , i M } ∘ d n , l { i 0 , … , i j } = ∑ j = 1 M − 1 ( − 1 ) j d n , k I ∖ { i j } . (189) \sum_{j=0}^{M}\sum_{0\leq l\leq\max(n,n+j-1)}(-1)^j d_{l,k}^{\{i_j,\ldots,i_M\}}\circ d_{n,l}^{\{i_0,\ldots,i_j\}}=\sum_{j=1}^{M-1}(-1)^j d_{n,k}^{I\setminus\{i_j\}}. \tag*{(189)} j = 0 ∑ M 0 ≤ l ≤ m a x ( n , n + j − 1 ) ∑ ( − 1 ) j d l , k { i j , … , i M } ∘ d n , l { i 0 , … , i j } = j = 1 ∑ M − 1 ( − 1 ) j d n , k I ∖ { i j } . (189) Here the upper limit on l l l is not really necessary, as if it does not hold then d n , l { i 0 , … , i j } = 0 d_{n,l}^{\{i_0,\ldots,i_j\}}=0 d n , l { i 0 , … , i j } = 0 by Lemma 6.14.
Here is the analogue of Example 6.20 .
Example 6.30. (a) A 0-simplex ( ( A 0 ) , ( m { 0 } ) ) \left((A^0),(\mathfrak{m}^{\{0\}})\right) ( ( A 0 ) , ( m { 0 } ) ) of A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is a unital A ∞ A_\infty A ∞ -algebra over R R R .
(b) A 1-simplex ( ( A 0 , A 1 ) , ( m { 0 } , m { 1 } , m { 0 , 1 } ) ) \left((A^0,A^1),(\mathfrak{m}^{\{0\}},\mathfrak{m}^{\{1\}},\mathfrak{m}^{\{0,1\}})\right) ( ( A 0 , A 1 ) , ( m { 0 } , m { 1 } , m { 0 , 1 } ) ) of A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is equivalent to a unital morphism m { 0 , 1 } : ( A 0 , m { 0 } ) → ( A 1 , m { 1 } ) \mathfrak{m}^{\{0,1\}}:(A^0,\mathfrak{m}^{\{0\}})\to(A^1,\mathfrak{m}^{\{1\}}) m { 0 , 1 } : ( A 0 , m { 0 } ) → ( A 1 , m { 1 } ) of unital A ∞ A_\infty A ∞ -algebras.
(c) A 2-simplex of A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is equivalent to unital A ∞ A_\infty A ∞ -algebras ( A 0 , m { 0 } ) (A^0,\mathfrak{m}^{\{0\}}) ( A 0 , m { 0 } ) , ( A 1 , m { 1 } ) (A^1,\mathfrak{m}^{\{1\}}) ( A 1 , m { 1 } ) , ( A 2 , m { 2 } ) (A^2,\mathfrak{m}^{\{2\}}) ( A 2 , m { 2 } ) as in (a), and unital morphisms m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) \mathfrak{m}^{\{i,j\}}:(A^i,\mathfrak{m}^{\{i\}})\to(A^j,\mathfrak{m}^{\{j\}}) m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), and R R R -linear maps
m k { 0 , 1 , 2 } : ( A 0 ) ⊗ k ⟶ A 2 \mathfrak{m}^{\{0,1,2\}}_k:(A^0)^{\otimes k}\longrightarrow A^2 m k { 0 , 1 , 2 } : ( A 0 ) ⊗ k ⟶ A 2
for k = 1 , 2 , … k=1,2,\ldots k = 1 , 2 , … , graded of degree − k -k − k , with m 0 { 0 , 1 , 2 } = 0 m^{\{0,1,2\}}_0=0 m 0 { 0 , 1 , 2 } = 0 , satisfying for all n ≥ 0 n\geq0 n ≥ 0
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) − ∑ n = n 1 + ⋯ + n k n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 1 , 2 } ( m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } ) + ∑ n = n 1 + ⋯ + n k , 1 ≤ l ≤ k n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 p i j ≥ 1 , m i ≥ 1 , all i , j n i ≥ 1 , i = 1 , … , k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ ∏ i = 1 l − 1 ( − 1 ) ∑ a = 1 m i ( p i a + 1 ) ( m i − a ) m k { 2 } [ m p 1 1 { 1 , 2 } ( m p 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p 1 m 1 { 0 , 1 } ) ⊗ ⋯ ⊗ m p l − 1 1 { 0 , 1 } ( m p l − 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p l − 1 m l − 1 { 0 , 1 } ) ⊗ m n l { 0 , 1 , 2 } ⊗ m n l + 1 { 0 , 2 } ⊗ ⋯ ⊗ m n k { 0 , 2 } ] = − m n { 0 , 2 } . (190) \begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\ i+j+k=n}}(-1)^{i+jk}m^{\{0,1,2\}}_{i+k+1}\circ\left(\operatorname{id}_{A_0}^{\otimes i}\otimes m^{\{0\}}_j\otimes\operatorname{id}_{A_0}^{\otimes k}\right)\\
&-\sum_{\substack{n=n_1+\cdots+n_k\\ n_a\geq1}}(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}m^{\{1,2\}}_k\left(m^{\{0,1\}}_{n_1}\otimes m^{\{0,1\}}_{n_2}\otimes\cdots\otimes m^{\{0,1\}}_{n_k}\right)\\
&+\sum_{\substack{n=n_1+\cdots+n_k,\ 1\leq l\leq k\\ n_i=p_i^1+\cdots+p_i^{m_i},\ i=1,\ldots,l-1\\ p_i^j\geq1,\ m_i\geq1,\ \text{all }i,j\\ n_i\geq1,\ i=1,\ldots,k}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\cdot\prod_{i=1}^{l-1}(-1)^{\sum_{a=1}^{m_i}(p_i^a+1)(m_i-a)}\\
&\qquad m_k^{\{2\}}\left[m_{p_1^1}^{\{1,2\}}\left(m_{p_1^2}^{\{0,1\}}\otimes\cdots\otimes m_{p_1^{m_1}}^{\{0,1\}}\right)\otimes\cdots\otimes m_{p_{l-1}^1}^{\{0,1\}}\left(m_{p_{l-1}^2}^{\{0,1\}}\otimes\cdots\otimes m_{p_{l-1}^{m_{l-1}}}^{\{0,1\}}\right)\right.\\
&\qquad\left.\otimes m_{n_l}^{\{0,1,2\}}\otimes m_{n_{l+1}}^{\{0,2\}}\otimes\cdots\otimes m_{n_k}^{\{0,2\}}\right]\\
&=-m_n^{\{0,2\}}. \tag*{(190)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ∘ ( id A 0 ⊗ i ⊗ m j { 0 } ⊗ id A 0 ⊗ k ) − n = n 1 + ⋯ + n k n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 1 , 2 } ( m n 1 { 0 , 1 } ⊗ m n 2 { 0 , 1 } ⊗ ⋯ ⊗ m n k { 0 , 1 } ) + n = n 1 + ⋯ + n k , 1 ≤ l ≤ k n i = p i 1 + ⋯ + p i m i , i = 1 , … , l − 1 p i j ≥ 1 , m i ≥ 1 , all i , j n i ≥ 1 , i = 1 , … , k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ i = 1 ∏ l − 1 ( − 1 ) ∑ a = 1 m i ( p i a + 1 ) ( m i − a ) m k { 2 } [ m p 1 1 { 1 , 2 } ( m p 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p 1 m 1 { 0 , 1 } ) ⊗ ⋯ ⊗ m p l − 1 1 { 0 , 1 } ( m p l − 1 2 { 0 , 1 } ⊗ ⋯ ⊗ m p l − 1 m l − 1 { 0 , 1 } ) ⊗ m n l { 0 , 1 , 2 } ⊗ m n l + 1 { 0 , 2 } ⊗ ⋯ ⊗ m n k { 0 , 2 } ] = − m n { 0 , 2 } . (190)
Note that we require n l ≥ 1 n_l\geq1 n l ≥ 1 in the third sum rather than n l ≥ 0 n_l\geq0 n l ≥ 0 , as m 0 { 0 , 1 , 2 } = 0 m^{\{0,1,2\}}_0=0 m 0 { 0 , 1 , 2 } = 0 . Equation (6.31) is trivial when n = 0 n=0 n = 0 . By equations (3.10), (3.12), (3.13), (6.21), and (6.31), m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 1 m^{\{0,1,2\}}=(m_k^{\{0,1,2\}})_{k\geq1} m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 1 is a unital homotopy m { 1 , 2 } ∘ m { 0 , 1 } ⇒ m { 0 , 2 } m^{\{1,2\}}\circ m^{\{0,1\}}\Rightarrow m^{\{0,2\}} m { 1 , 2 } ∘ m { 0 , 1 } ⇒ m { 0 , 2 } . Hence, a 2-simplex in A ˙ ∞ A l g R , K u n , ∞ \dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Alg R , K un , ∞ is a 2-commutative diagram of A ∞ A_{\infty} A ∞ -algebras
( A 1 , m { 1 } ) m { 0 , 1 } ↗ ↓ m { 0 , 1 , 2 } ↘ m { 1 , 2 } ( A 0 , m { 0 } ) → m { 0 , 2 } ( A 2 , m { 2 } ) . \begin{array}{ccccc}
& (A^1,m^{\{1\}}) & & \\
{}^{m^{\{0,1\}}}\nearrow& \downarrow^{m^{\{0,1,2\}}} & \searrow^{m^{\{1,2\}}} & \\
(A^0,m^{\{0\}}) & \xrightarrow{\ m^{\{0,2\}}\ } & (A^2,m^{\{2\}}).
\end{array} m { 0 , 1 } ↗ ( A 0 , m { 0 } ) ( A 1 , m { 1 } ) ↓ m { 0 , 1 , 2 } m { 0 , 2 } ↘ m { 1 , 2 } ( A 2 , m { 2 } ) .
In particular, if ( A i ) i = 0 2 , ( m I ) ∅ ≠ I ⊆ { 0 , 1 , 2 } (A^i)_{i=0}^{2},(m^I)_{\varnothing\neq I\subseteq\{0,1,2\}} ( A i ) i = 0 2 , ( m I ) ∅ = I ⊆ { 0 , 1 , 2 } is a 2-simplex in A ˙ ∞ A l g R , K ∞ \dot{A}_{\infty}\mathrm{Alg}^{\infty}_{R,K} A ˙ ∞ Alg R , K ∞ and m { 0 , 1 , 2 } = 0 m^{\{0,1,2\}}=0 m { 0 , 1 , 2 } = 0 then m { 0 , 2 } = m { 1 , 2 } ∘ m { 0 , 1 } m^{\{0,2\}}=m^{\{1,2\}}\circ m^{\{0,1\}} m { 0 , 2 } = m { 1 , 2 } ∘ m { 0 , 1 } as morphisms of A ∞ A_{\infty} A ∞ -algebras ( A 0 , m 0 ) → ( A 2 , m 2 ) (A^0,m^0)\to(A^2,m^2) ( A 0 , m 0 ) → ( A 2 , m 2 ) . This is how the notion of composition of morphisms of A ∞ A_{\infty} A ∞ -algebras in Definition 3.8 is encoded in our theory.
The next theorem is proved in §10.8.
Theorem 6.31. There is a strict isomorphism of ∞ \infty ∞ -categories
I R , K u n : A ∞ A l g R , K u n , ∞ ⟶ A ˙ ∞ A l g R , K u n , ∞ , K ∈ { 3 , 4 , 5 , … , ∞ } , (191) I^{\mathrm{un}}_{R,K}:A_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K}\longrightarrow\dot{A}_{\infty}\mathrm{Alg}^{\mathrm{un},\infty}_{R,K},\qquad K\in\{3,4,5,\ldots,\infty\}, \tag*{(191)} I R , K un : A ∞ Alg R , K un , ∞ ⟶ A ˙ ∞ Alg R , K un , ∞ , K ∈ { 3 , 4 , 5 , … , ∞ } , (191) from Definitions 6.25 and 6.27, which acts on N N N -simplices by
I R , K u n : ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ⟼ ( ( A i ) i = 0 N , ( m ˙ I ) ∅ ≠ I ⊆ { 0 , … , N } ) , I^{\mathrm{un}}_{R,K}:\left((A^i)_{i=0}^{N},(m^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right)\longmapsto\left((A^i)_{i=0}^{N},(\dot{m}^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right), I R , K un : ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) ⟼ ( ( A i ) i = 0 N , ( m ˙ I ) ∅ = I ⊆ { 0 , … , N } ) , where m ˙ k { i 0 , … , i M } \dot{m}^{\{i_0,\ldots,i_M\}}_k m ˙ k { i 0 , … , i M } is defined from m k { i 0 , … , i M } m^{\{i_0,\ldots,i_M\}}_k m k { i 0 , … , i M } by
m ˙ k { i 0 , … , i M } = { 0 , k = 0 and M = 2 , m k { i 0 , … , i M } , otherwise . (192) \dot{m}^{\{i_0,\ldots,i_M\}}_k=
\begin{cases}
0, & k=0\text{ and }M=2,\\
m^{\{i_0,\ldots,i_M\}}_k, & \text{otherwise}.
\end{cases} \tag*{(192)} m ˙ k { i 0 , … , i M } = { 0 , m k { i 0 , … , i M } , k = 0 and M = 2 , otherwise . (192) This is equivalent to (6.27) in this case, as m 0 { i 0 , i 1 , i 2 } = 1 i 2 m^{\{i_0,i_1,i_2\}}_0=\mathbb{1}^{i_2} m 0 { i 0 , i 1 , i 2 } = 1 i 2 .
We now define the remaining ∞ \infty ∞ -categories in (6.1).
Definition 6.33. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Generalizing Definition 6.27, define a simplicial set A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ of cohomologically unital A ∞ A_{\infty} A ∞ -algebras over R R R such that the N N N -simplices ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) for N ≥ 0 N\geq0 N ≥ 0 satisfy Definition 6.27(a)(i)–(ii), but instead of Definition 6.27(a)(iii), we require that
(iii)′ ^{\prime} ′ As in Example 6.30(a), Definition 6.27(a)(i)–(ii) imply that ( A i , m { i } ) (A^{i},m^{\{i\}}) ( A i , m { i } ) is an A ∞ A_{\infty} A ∞ -algebra for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N . We require this A ∞ A_{\infty} A ∞ -algebra ( A i , m { i } ) (A^{i},m^{\{i\}}) ( A i , m { i } ) to be cohomologically unital, as in Definition 3.6.
(iv)′ ^{\prime} ′ As in Example 6.30(b), Definition 6.27(a)(i)–(ii) mean m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) m^{\{i,j\}}\colon(A^{i},m^{\{i\}})\to(A^{j},m^{\{j\}}) m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) is a morphism of A ∞ A_{\infty} A ∞ -algebras for all 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N . We require this to be a cohomologically unital morphism, as in Definition 3.8.
If α : [ N ‾ ] → [ N ] \alpha\colon[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , define A ∞ A l g R , K c u , ∞ ( α ) : ( A ∞ A l g R , K c u , ∞ ) N → ( A ∞ A l g R , K c u , ∞ ) N ‾ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty}(\alpha)\colon(A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty})^{N}\to(A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty})^{\overline{N}} A ∞ Alg R , K cu , ∞ ( α ) : ( A ∞ Alg R , K cu , ∞ ) N → ( A ∞ Alg R , K cu , ∞ ) N as in Definition 6.27(b). The proof of Proposition 6.28 also works in the cohomologically unital case, and shows A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ is a simplicial set. It contains A ∞ A l g R , K u n , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty} A ∞ Alg R , K un , ∞ in Definition 6.27 as a simplicial subset.
Theorem 6.29 also holds in A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ , as the proof does not use unitality.
Theorem 6.35 will show that A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ is an ∞ \infty ∞ -category. We say that A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ has type K K K . We also write A ∞ A l g R c u , ∞ = A ∞ A l g R , 3 c u , ∞ A_{\infty}\mathrm{Alg}_{R}^{\mathrm{cu},\infty}=A_{\infty}\mathrm{Alg}_{R,3}^{\mathrm{cu},\infty} A ∞ Alg R cu , ∞ = A ∞ Alg R , 3 cu , ∞ .
Definition 6.34. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Generalizing Definition 6.27, define a simplicial set A ∞ A l g R , K ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ of (possibly non-unital) A ∞ A_{\infty} A ∞ -algebras over R R R whose N N N -simplices ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) for N ≥ 0 N\geq0 N ≥ 0 satisfy Definition 6.27(a)(i)–(ii), but need not satisfy (a)(iii). If α : [ N ‾ ] → [ N ] \alpha\colon[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , define A ∞ A l g R , K ∞ ( α ) : ( A ∞ A l g R , K ∞ ) N → ( A ∞ A l g R , K ∞ ) N ‾ A_{\infty}\mathrm{Alg}_{R,K}^{\infty}(\alpha)\colon(A_{\infty}\mathrm{Alg}_{R,K}^{\infty})^{N}\to(A_{\infty}\mathrm{Alg}_{R,K}^{\infty})^{\overline{N}} A ∞ Alg R , K ∞ ( α ) : ( A ∞ Alg R , K ∞ ) N → ( A ∞ Alg R , K ∞ ) N as in Definition 6.27(b). The proof of Proposition 6.28 also works in the non-unital case, and shows A ∞ A l g R , K ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ is a simplicial set. It contains A ∞ A l g R , K c u , ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ in Definition 6.33 as a simplicial subset.
Theorem 6.29 also holds in A ∞ A l g R , K ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ , as the proof does not use unitality.
Theorem 6.35 will show that A ∞ A l g R , K ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ is an ∞ \infty ∞ -category. We say that A ∞ A l g R , K ∞ A_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ has type K K K . We also write A ∞ A l g R ∞ = A ∞ A l g R , 3 ∞ A_{\infty}\mathrm{Alg}_{R}^{\infty}=A_{\infty}\mathrm{Alg}_{R,3}^{\infty} A ∞ Alg R ∞ = A ∞ Alg R , 3 ∞ .
The next theorem, one of our main results, is a special case of Theorem 7.29.
Theorem 6.35. The following simplicial sets defined in Definitions 6.25, 6.27, 6.33, and 6.34, for all K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } , are all ∞ \infty ∞ -categories:
A ∞ A l g R , K u n , ∞ , A ˙ ∞ A l g R , K u n , ∞ , A ∞ A l g R , K c u , ∞ , A ∞ A l g R , K ∞ . A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty},\ \dot{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{un},\infty},\ A_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty},\ A_{\infty}\mathrm{Alg}_{R,K}^{\infty}. A ∞ Alg R , K un , ∞ , A ˙ ∞ Alg R , K un , ∞ , A ∞ Alg R , K cu , ∞ , A ∞ Alg R , K ∞ .
The next definition is an analogue of Lurie [61, Rem. 1.3.1.9] for dg-categories. It completes our diagram (160) of inclusions of ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -algebras.
Definition 6.36. Let A ∞ A l g R u n \mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un}}_{R} A ∞ Alg R un be the ordinary category of unital A ∞ A_\infty A ∞ -algebras over R R R from Definition 3.8, and N ( A ∞ A l g R u n ) \mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un}}_{R}) N ( A ∞ Alg R un ) be its nerve from Example 2.11, which is an ∞ \infty ∞ -category. Then N N N -simplices in N ( A ∞ A l g R u n ) \mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un}}_{R}) N ( A ∞ Alg R un ) are of the form ( ( A i , m i ) i = 0 N , ( f i ) i = 1 N ) \left((A^i,\mathbf{m}^i)_{i=0}^{N},(f^i)_{i=1}^{N}\right) ( ( A i , m i ) i = 0 N , ( f i ) i = 1 N ) , where ( A i , m i ) (A^i,\mathbf{m}^i) ( A i , m i ) is a unital A ∞ A_\infty A ∞ -algebra over R R R for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and f i : ( A i − 1 , m i − 1 ) → ( A i , m i ) f^i:(A^{i-1},\mathbf{m}^{i-1})\to(A^i,\mathbf{m}^i) f i : ( A i − 1 , m i − 1 ) → ( A i , m i ) is a unital A ∞ A_\infty A ∞ -morphism for i = 1 , … , N i=1,\ldots,N i = 1 , … , N .
Define an ∞ \infty ∞ -functor inc R u n : N ( A ∞ A l g R u n ) → A ˙ ∞ A l g R , ∞ u n , ∞ \operatorname{inc}^{\mathrm{un}}_{R}:\mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un}}_{R})\to\dot{\mathbf{A}}_\infty\mathrm{Alg}^{\mathrm{un},\infty}_{R,\infty} inc R un : N ( A ∞ Alg R un ) → A ˙ ∞ Alg R , ∞ un , ∞ to act on N N N -simplices taking ( ( A i , m i ) i = 0 N , ( f i ) i = 1 N ) \left((A^i,\mathbf{m}^i)_{i=0}^{N},(f^i)_{i=1}^{N}\right) ( ( A i , m i ) i = 0 N , ( f i ) i = 1 N ) to ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^{N},(\mathbf{m}^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where we set m { i } = m i \mathbf{m}^{\{i\}}=\mathbf{m}^i m { i } = m i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and m { i , j } = f j ∘ f j − 1 ∘ ⋯ ∘ f i + 1 : ( A i , m { i } ) → ( A j , m { j } ) \mathbf{m}^{\{i,j\}}=f^j\circ f^{j-1}\circ\cdots\circ f^{i+1}:(A^i,\mathbf{m}^{\{i\}})\to(A^j,\mathbf{m}^{\{j\}}) m { i , j } = f j ∘ f j − 1 ∘ ⋯ ∘ f i + 1 : ( A i , m { i } ) → ( A j , m { j } ) if 0 ≤ i < j ≤ N 0\le i<j\le N 0 ≤ i < j ≤ N , and m k I = 0 \mathbf{m}^{I}_k=0 m k I = 0 if ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 and k ≥ 0 k\ge0 k ≥ 0 .
We claim ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^{N},(\mathbf{m}^{I}_k)_{k\ge0,\ \varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ = I ⊆ { 0 , … , N } ) is an N N N -simplex of A ˙ ∞ A l g R , ∞ u n , ∞ \dot{\mathbf{A}}_\infty\mathrm{Alg}^{\mathrm{un},\infty}_{R,\infty} A ˙ ∞ Alg R , ∞ un , ∞ . The condition that m 0 I = 0 \mathbf{m}^{I}_0=0 m 0 I = 0 for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } with ∣ I ∣ ≤ ∞ |I|\le\infty ∣ I ∣ ≤ ∞ is clear. We must verify (187) for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } and n ≥ 0 n\ge0 n ≥ 0 . Divide into cases:
If I = { i } I=\{i\} I = { i } , this follows from ( A i , m { i } ) (A^i,\mathbf{m}^{\{i\}}) ( A i , m { i } ) being an A ∞ A_\infty A ∞ -algebra, as in Example 6.30(a).
If I = { i , j } I=\{i,j\} I = { i , j } with i < j i<j i < j , this follows from m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) \mathbf{m}^{\{i,j\}}:(A^i,\mathbf{m}^{\{i\}})\to(A^j,\mathbf{m}^{\{j\}}) m { i , j } : ( A i , m { i } ) → ( A j , m { j } ) being a morphism of A ∞ A_\infty A ∞ -algebras, as in Example 6.30(b).
If I = { i , j , k } I=\{i,j,k\} I = { i , j , k } with i < j < k i<j<k i < j < k , this follows from
m { i , k } = f k ∘ f k − 1 ∘ ⋯ ∘ f j + 1 ∘ f j ∘ ⋯ ∘ f i + 1 = m { j , k } ∘ m { i , j } \mathbf{m}^{\{i,k\}}=f^k\circ f^{k-1}\circ\cdots\circ f^{j+1}\circ f^j\circ\cdots\circ f^{i+1}=\mathbf{m}^{\{j,k\}}\circ\mathbf{m}^{\{i,j\}} m { i , k } = f k ∘ f k − 1 ∘ ⋯ ∘ f j + 1 ∘ f j ∘ ⋯ ∘ f i + 1 = m { j , k } ∘ m { i , j } and m { i , j , k } = 0 \mathbf{m}^{\{i,j,k\}}=0 m { i , j , k } = 0 , as in Example 6.30(c).
If ∣ I ∣ ≥ 4 |I|\ge4 ∣ I ∣ ≥ 4 then on the left hand side of (187) we have m k { i j , … , i M } = 0 \mathbf{m}^{\{i_j,\ldots,i_M\}}_k=0 m k { i j , … , i M } = 0 if j < M − 1 j<M-1 j < M − 1 , as then ∣ { i j , … , i M } ∣ ≥ 3 |\{i_j,\ldots,i_M\}|\ge3 ∣ { i j , … , i M } ∣ ≥ 3 , and d n , k { i 0 , … , i j } = 0 d^{\{i_0,\ldots,i_j\}}_{n,k}=0 d n , k { i 0 , … , i j } = 0 if j ≥ M − 1 ≥ 2 j\ge M-1\ge2 j ≥ M − 1 ≥ 2 , as every term of d n , k { i 0 , … , i j } d^{\{i_0,\ldots,i_j\}}_{n,k} d n , k { i 0 , … , i j } must contain a factor m k J \mathbf{m}^{J}_k m k J with ∣ J ∣ ≥ 3 |J|\ge3 ∣ J ∣ ≥ 3 , and so must be zero. Also the right hand side of (187) is zero as ∣ I ∖ { i j } ∣ ≥ 3 |I\setminus\{i_j\}|\ge3 ∣ I ∖ { i j } ∣ ≥ 3 .
Definition 6.27(a)(iii) follows from Definition 3.8.
It is now easy to see that inc R u n \operatorname{inc}^{\mathrm{un}}_{R} inc R un is a morphism of simplicial sets, that is, an ∞ \infty ∞ -functor. It is a strict isomorphism with the ∞ \infty ∞ -subcategory of A ˙ ∞ A l g R , ∞ u n , ∞ \dot{\mathbf{A}}_\infty\mathrm{Alg}^{\mathrm{un},\infty}_{R,\infty} A ˙ ∞ Alg R , ∞ un , ∞ consisting of N N N -simplices ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^{N},(\mathbf{m}^{I}_k)_{k\ge0,\ \varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ = I ⊆ { 0 , … , N } ) with m k I = 0 \mathbf{m}^{I}_k=0 m k I = 0 if ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 and k ≥ 0 k\ge0 k ≥ 0 . In the same way we define ∞ \infty ∞ -functors inc R c u : N ( A ∞ A l g R c u ) ↪ A ∞ A l g R , ∞ c u , ∞ \operatorname{inc}^{\mathrm{cu}}_{R}:\mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{cu}}_{R})\hookrightarrow\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{cu},\infty}_{R,\infty} inc R cu : N ( A ∞ Alg R cu ) ↪ A ∞ Alg R , ∞ cu , ∞ for cohomologically unital A ∞ A_\infty A ∞ -algebras, and inc R : N ( A ∞ A l g R ) ↪ A ∞ A l g R , ∞ ∞ \operatorname{inc}_{R}:\mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}_{R})\hookrightarrow\mathbf{A}_\infty\mathrm{Alg}^{\infty}_{R,\infty} inc R : N ( A ∞ Alg R ) ↪ A ∞ Alg R , ∞ ∞ for general A ∞ A_\infty A ∞ -algebras.
We can also define an ∞ \infty ∞ -functor inc R u n : N ( A ∞ A l g R u n ) → A ∞ A l g R , ∞ u n , ∞ \operatorname{inc}^{\mathrm{un}}_{R}:\mathcal{N}(\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un}}_{R})\to\mathbf{A}_\infty\mathrm{Alg}^{\mathrm{un},\infty}_{R,\infty} inc R un : N ( A ∞ Alg R un ) → A ∞ Alg R , ∞ un , ∞ which is identified with inc R u n \operatorname{inc}^{\mathrm{un}}_{R} inc R un under the isomorphism I R , ∞ u n , ∞ I^{\mathrm{un},\infty}_{R,\infty} I R , ∞ un , ∞ in (191) . The difference is that as in (186) , when we define ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ ≠ I ⊆ { 0 , … , N } ) \left((A^i)_{i=0}^{N},(\mathbf{m}^{I}_k)_{k\ge0,\ \varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( A i ) i = 0 N , ( m k I ) k ≥ 0 , ∅ = I ⊆ { 0 , … , N } ) above we define m 0 { i 0 , i 1 , i 2 } = 1 i 2 \mathbf{m}^{\{i_0,i_1,i_2\}}_0=\mathbb{1}^{i_2} m 0 { i 0 , i 1 , i 2 } = 1 i 2 rather than m 0 { i 0 , i 1 , i 2 } = 0 \mathbf{m}^{\{i_0,i_1,i_2\}}_0=0 m 0 { i 0 , i 1 , i 2 } = 0 .
∞-categories of A∞-categories
We now define many different ∞-categories of A ∞ A_\infty A ∞ -categories over a commutative ring R R R , which fit into a diagram of inclusions of ∞-categories:
\begin{CD}
A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} = A_\infty\operatorname{Cat}_{R,2}^{\mathrm{un},\infty} \\
@AAA @AAA \\
A_\infty\operatorname{Cat}_{R,3}^{\mathrm{un},\infty} @= \dot{A}_\infty\operatorname{Cat}_{R,3}^{\mathrm{un},\infty} @>>> A_\infty\operatorname{Cat}_{R,3}^{\mathrm{cu},\infty} @>>> A_\infty\operatorname{Cat}_{R,3}^{\infty} \\
@AAA @AAA @AAA @AAA \\
A_\infty\operatorname{Cat}_{R,4}^{\mathrm{un},\infty} @= \dot{A}_\infty\operatorname{Cat}_{R,4}^{\mathrm{un},\infty} @>>> A_\infty\operatorname{Cat}_{R,4}^{\mathrm{cu},\infty} @>>> A_\infty\operatorname{Cat}_{R,4}^{\infty} \\
@AAA @AAA @AAA @AAA \\
A_\infty\operatorname{Cat}_{R,5}^{\mathrm{un},\infty} @= \dot{A}_\infty\operatorname{Cat}_{R,5}^{\mathrm{un},\infty} @>>> A_\infty\operatorname{Cat}_{R,5}^{\mathrm{cu},\infty} @>>> A_\infty\operatorname{Cat}_{R,5}^{\infty} \\
@AAA @AAA @AAA @AAA \\
\vdots@. \vdots@. \vdots@. \vdots\\
@AAA @AAA @AAA @AAA \\
A_\infty\operatorname{Cat}_{R,\infty}^{\mathrm{un},\infty} @= \dot{A}_\infty\operatorname{Cat}_{R,\infty}^{\mathrm{un},\infty} @>>> A_\infty\operatorname{Cat}_{R,\infty}^{\mathrm{cu},\infty} @>>> A_\infty\operatorname{Cat}_{R,\infty}^{\infty} \\
@AAA @AAA @AAA @AAA \\
\mathcal{N}(A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}) @= \mathcal{N}(\dot{A}_\infty\operatorname{Cat}_{R}^{\mathrm{un}}) @>>> \mathcal{N}(A_\infty\operatorname{Cat}_{R}^{\mathrm{cu}}) @>>> \mathcal{N}(A_\infty\operatorname{Cat}_{R}) \tag*{(193)}
\end{CD}
This is the exact analogue of our diagram (160) of ∞-categories of A ∞ A_\infty A ∞ -algebras, and reduces to (160) if we restrict to A ∞ A_\infty A ∞ -categories with one object *, which are equivalent to A ∞ A_\infty A ∞ -algebras.
As in Chapter 6, we note that:
The ∞-categories in the first two columns, with superscripts ‘un’, have objects unital A ∞ A_\infty A ∞ -categories and 1-morphisms unital A ∞ A_\infty A ∞ -functors.
The ∞-categories in the third column, with superscript ‘cu’, have objects cohomologically unital A ∞ A_\infty A ∞ -categories and 1-morphisms cohomologically unital A ∞ A_\infty A ∞ -functors.
The ∞-categories in the fourth column have objects all A ∞ A_\infty A ∞ -categories, and 1-morphisms all A ∞ A_\infty A ∞ -functors.
The ∞-categories in the first and second columns, linked by ‘=’, are strictly isomorphic, but with definitions written in a different way.
In each row we have subscripts ‘R R R , K K K ’ for K = 2 , 3 , 4 , … , ∞ K = 2, 3, 4, \ldots, \infty K = 2 , 3 , 4 , … , ∞ , where we call K K K the type of the ∞-category.
N ( A ∞ Cat R u n ) , … , N ( A ∞ Cat R ) \mathcal{N}(A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}), \ldots, \mathcal{N}(A_\infty\operatorname{Cat}_{R}) N ( A ∞ Cat R un ) , … , N ( A ∞ Cat R ) are the nerves of the ordinary categories A ∞ Cat R u n , … , A ∞ Cat R A_\infty\operatorname{Cat}_{R}^{\mathrm{un}}, \ldots, A_\infty\operatorname{Cat}_{R} A ∞ Cat R un , … , A ∞ Cat R from Definition 4.9.
The 2-morphisms in A ∞ C a t R u n , ∞ \mathop{\mathrm{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ of type 2 in the first row of (193) correspond to unital A ∞ A_{\infty} A ∞ -natural isomorphisms in Definition 4.22. But in the ∞ \infty ∞ -categories of type K ≥ 3 K \ge3 K ≥ 3 in the subsequent rows, the 2-morphisms correspond to (unital ) homotopies of A ∞ A_{\infty} A ∞ -functors, as in Definition 4.21. In the author’s view, A ∞ C a t R u n , ∞ \mathop{\mathrm{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ will be the best ∞ \infty ∞ -category of A ∞ A_{\infty} A ∞ -categories for most applications, as A ∞ A_{\infty} A ∞ -natural isomorphisms are the best notion of when two A ∞ A_{\infty} A ∞ -functors are ‘the same’. It is a pity that this only works for unital A ∞ A_{\infty} A ∞ -categories.
Labelling graphs by objects in categories
Sections 6.1–6.2 set up notation for defining ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -algebras, involving R R R -linear maps m k I : ( A i 0 ) ⊗ k → A i M m_k^I\colon(A^{i_0})^{\otimes k}\to A^{i_M} m k I : ( A i 0 ) ⊗ k → A i M . To extend this to A ∞ A_{\infty} A ∞ -categories, as in Chapter 4 we must extend these to maps m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k depending on objects x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k in an A ∞ A_{\infty} A ∞ -category. When we deal with simplices of morphisms, as in §6.2, it will turn out that for I = { i 0 < i 1 < ⋯ < i M } I=\{i_0<i_1<\cdots<i_M\} I = { i 0 < i 1 < ⋯ < i M } we must work with graded R R R -linear maps of the form
m k I ; x 0 , … , x k : Hom A i 0 ( x 0 , x 1 ) ⊗ R Hom A i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A i 0 ( x k − 1 , x k ) → Hom A i M ( f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 ( x 0 ) , f i 0 i M ( x k ) ) . m_k^{I;x_0,\ldots,x_k}\colon\operatorname{Hom}_{\mathcal A^{i_0}}(x_0,x_1)\otimes_R\operatorname{Hom}_{\mathcal A^{i_0}}(x_1,x_2)\otimes_R\cdots\otimes_R\operatorname{Hom}_{\mathcal A^{i_0}}(x_{k-1},x_k)\to\operatorname{Hom}_{\mathcal A^{i_M}}\left(f^{i_{M-1}i_M}\circ f^{i_{M-2}i_{M-1}}\circ\cdots\circ f^{i_0i_1}(x_0),f^{i_0i_M}(x_k)\right). m k I ; x 0 , … , x k : Hom A i 0 ( x 0 , x 1 ) ⊗ R Hom A i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R Hom A i 0 ( x k − 1 , x k ) → Hom A i M ( f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 ( x 0 ) , f i 0 i M ( x k ) ) .
where A i \mathcal A^i A i are A ∞ A_{\infty} A ∞ -categories and f i j = F O b j i j : Obj A i → Obj A j f^{ij}=\mathfrak F_{\mathrm{Obj}}^{ij}\colon\operatorname{Obj}\mathcal A^i\to\operatorname{Obj}\mathcal A^j f ij = F Obj ij : Obj A i → Obj A j for A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathfrak F^{ij}\colon\mathcal A^i\to\mathcal A^j F ij : A i → A j .
We now extend the graphical notation of §6.1–§6.2 to cover multiple compositions of such maps m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k . To do this we add labels of ordered pairs ( y e , z e ) (y_e,z_e) ( y e , z e ) of objects in A i \mathcal A^i A i to each edge e e e in Γ \Gamma Γ , such that e e e represents an input/output in Hom A i ( y e , z e ) \operatorname{Hom}_{\mathcal A^i}(y_e,z_e) Hom A i ( y e , z e ) . These pairs ( y e , z e ) (y_e,z_e) ( y e , z e ) must satisfy consistency conditions to match the domains and ranges of the maps m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k .
We first define maps C ( Γ ) x 0 , … , x n \mathcal C(\Gamma)^{x_0,\ldots,x_n} C ( Γ ) x 0 , … , x n , generalizing C ( Γ ) \mathcal C(\Gamma) C ( Γ ) in Definition 6.6. We restrict to balanced Γ \Gamma Γ , as in Definition 6.7, as being balanced automatically implies the required consistency conditions on pairs ( y e , z e ) (y_e,z_e) ( y e , z e ) .
Definition 7.1. Let R R R be a commutative ring. Let N ≥ 0 N\ge0 N ≥ 0 , and suppose we are given data ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } (\mathcal O^i)_{i=0}^N,(f^{ij})_{0\le i\le j\le N},(\mathcal H^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}} ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } , where:
O i \mathcal O^i O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N . Elements of O i \mathcal O^i O i are called objects . Later we will take O i = Obj A i \mathcal O^i=\operatorname{Obj}\mathcal A^i O i = Obj A i for an A ∞ A_{\infty} A ∞ -category A i \mathcal A^i A i .
f i j : O i → O j f^{ij}\colon\mathcal O^i\to\mathcal O^j f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\le i\le j\le N 0 ≤ i ≤ j ≤ N . We require these maps to satisfy f i i = id O i f^{ii}=\operatorname{id}_{\mathcal O^i} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N . Later we will take f i j = F O b j i j f^{ij}=\mathfrak F_{\mathrm{Obj}}^{ij} f ij = F Obj ij for an A ∞ A_{\infty} A ∞ -functor F i j : A i → A j \mathfrak F^{ij}\colon\mathcal A^i\to\mathcal A^j F ij : A i → A j . For the ∞ \infty ∞ -category A ∞ C a t R u n , ∞ \mathop{\mathrm{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ we do not require that f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik , so for the present we do not assume this, though for the other ∞ \infty ∞ -categories in (193) we do require it.
Let ∅ ≠ I ⊆ { 0 , 1 , … , N } \varnothing\ne I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } with I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M . Define
f I : O i 0 → O i M f^I\colon\mathcal O^{i_0}\to\mathcal O^{i_M} f I : O i 0 → O i M
f I = { f i 0 i 0 = id O i 0 , M = 0 , f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 , M > 0. (194) f^I=
\begin{cases}
f^{i_0i_0}=\operatorname{id}_{\mathcal O^{i_0}}, & M=0,\\
f^{i_{M-1}i_M}\circ f^{i_{M-2}i_{M-1}}\circ\cdots\circ f^{i_0i_1}, & M>0.
\end{cases} \tag*{(194)} f I = { f i 0 i 0 = id O i 0 , f i M − 1 i M ∘ f i M − 2 i M − 1 ∘ ⋯ ∘ f i 0 i 1 , M = 0 , M > 0. (194)
(iii) For all i = 0 , 1 , … , N i = 0, 1, \ldots, N i = 0 , 1 , … , N and x , y ∈ O i x,y \in\mathcal{O}^{i} x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded R R R -module H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d \mathcal{H}^{i}(x,y) = \bigoplus_{d\in\mathbb{Z}} \mathcal{H}^{i}(x,y)^{d} H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d . We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^{i} = (\mathcal{H}^{i}(x,y))_{x,y\in\mathcal{O}^{i}} H i = ( H i ( x , y ) ) x , y ∈ O i . Later we will take H i ( x , y ) = Hom A i ( x , y ) \mathcal{H}^{i}(x,y) = \operatorname{Hom}_{\mathcal{A}^{i}}(x,y) H i ( x , y ) = Hom A i ( x , y ) for the A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i .
(iv) For each nonempty subset I ⊂ { 0 , … , N } I \subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I = \{i_{0}, i_{1}, \ldots, i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0} < i_{1} < \cdots< i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M \ge0 M ≥ 0 , and for all k ≥ 0 k \ge0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_{0},\ldots,x_{k} \in\mathcal{O}^{i_{0}} x 0 , … , x k ∈ O i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) , (195) m^{I;x_{0},\ldots,x_{k}}_{k}\colon\mathcal{H}^{i_{0}}(x_{0},x_{1}) \otimes_{R} \mathcal{H}^{i_{0}}(x_{1},x_{2}) \otimes_{R} \cdots\otimes_{R} \mathcal{H}^{i_{0}}(x_{k-1},x_{k}) \longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k})), \tag*{(195)} m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k )) , (195)
graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k = 3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k .
Actually, for the case k = 0 k=0 k = 0 , our definition of A ∞ Cat R u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ only uses maps m 0 I ; x 0 m^{I;x_{0}}_{0} m 0 I ; x 0 when ∣ I ∣ ≥ 3 |I| \ge3 ∣ I ∣ ≥ 3 . So we could say we want maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k when either ∣ I ∣ ≥ 3 |I| \ge3 ∣ I ∣ ≥ 3 and k ≥ 0 k \ge0 k ≥ 0 , or when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 and k ≥ 1 k \ge1 k ≥ 1 . It is convenient to set m 0 I ; x 0 = 0 m^{I;x_{0}}_{0}=0 m 0 I ; x 0 = 0 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . For the ∞ \infty ∞ -category of filtered A ∞ A_{\infty} A ∞ -categories in Chapter 8 we will use maps m 0 I ; x 0 m^{I;x_{0}}_{0} m 0 I ; x 0 for all ∣ I ∣ ≥ 1 |I| \ge1 ∣ I ∣ ≥ 1 .
Let ∅ ≠ I ⊂ { 0 , … , N } \varnothing\ne I \subset\{0,\ldots,N\} ∅ = I ⊂ { 0 , … , N } , written as I = { i 0 , i 1 , … , i M } I = \{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M . Let Γ \Gamma Γ be a balanced n n n -leaved I I I -labelled tree, as in Definition 6.7. Let x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n} \in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 . Generalizing C ( Γ ) C(\Gamma) C ( Γ ) in (6.2), we will define an R R R -linear map
C ( Γ ) x 0 , … , x n : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x n − 1 , x n ) ⟶ H i M ( f L ( Γ ) ( x 0 ) , f R ( Γ ) ( x n ) ) , (196) C(\Gamma)^{x_{0},\ldots,x_{n}}\colon\mathcal{H}^{i_{0}}(x_{0},x_{1}) \otimes_{R} \mathcal{H}^{i_{0}}(x_{1},x_{2}) \otimes_{R} \cdots\otimes_{R} \mathcal{H}^{i_{0}}(x_{n-1},x_{n}) \longrightarrow\mathcal{H}^{i_{M}}(f^{L(\Gamma)}(x_{0}),f^{R(\Gamma)}(x_{n})), \tag*{(196)} C ( Γ ) x 0 , … , x n : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x n − 1 , x n ) ⟶ H i M ( f L ( Γ ) ( x 0 ) , f R ( Γ ) ( x n )) , (196)
where L ( Γ ) ⊇ R ( Γ ) L(\Gamma) \supseteq R(\Gamma) L ( Γ ) ⊇ R ( Γ ) are the left and right sequences of Γ \Gamma Γ in Definition 6.7.
We do this by associating operators m k I v ; y 0 , … , y k m^{I_{v};y_{0},\ldots,y_{k}}_{k} m k I v ; y 0 , … , y k from (7.3) to each internal vertex or bud v v v in Γ \Gamma Γ with k k k incoming edges, and taking the composition over the diagram, as in Definition 6.2. The new feature (compared to Definition 6.6, which involved maps m k I v m^{I_{v}}_{k} m k I v rather than m k I v ; y 0 , … , y k m^{I_{v};y_{0},\ldots,y_{k}}_{k} m k I v ; y 0 , … , y k ) is that we have to explain how to define the objects y 0 , … , y k y_{0},\ldots,y_{k} y 0 , … , y k in O min I v \mathcal{O}^{\min I_{v}} O m i n I v , in such a way that the maps m k I v ; y 0 , … , y k m^{I_{v};y_{0},\ldots,y_{k}}_{k} m k I v ; y 0 , … , y k are composable, that is, the output H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) \mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k})) H i M ( f I ( x 0 ) , f i 0 i M ( x k )) from one map (7.3) matches an input H i ( y a , y a + 1 ) \mathcal{H}^{i}(y_{a},y_{a+1}) H i ( y a , y a + 1 ) of another map (7.3) into which it is substituted.
We associate additional object labels to the tree Γ \Gamma Γ : to each directed edge e e e from v v v to w w w , we associate an ordered pair ( y e , z e ) (y_{e},z_{e}) ( y e , z e ) with y e , z e ∈ O max ( I v ) = min ( I w ) y_{e},z_{e} \in\mathcal{O}^{\max(I_{v})=\min(I_{w})} y e , z e ∈ O m a x ( I v ) = m i n ( I w ) , and to each bud b b b we associate an object x j b x_{j_{b}} x j b , according to the following rules:
(a) Let l j l_{j} l j be the j t h j^{\mathrm{th}} j th leaf of Γ \Gamma Γ for j = 1 , … , n j=1,\ldots,n j = 1 , … , n , as in Definition 6.1(viii), and let e e e be the unique edge in Γ \Gamma Γ starting at l j l_{j} l j . We set ( y e , z e ) = ( x j − 1 , x j ) (y_{e},z_{e})=(x_{j-1},x_{j}) ( y e , z e ) = ( x j − 1 , x j ) , for x k ∈ O i 0 x_{k}\in\mathcal{O}^{i_{0}} x k ∈ O i 0 as given.
(b) Let r r r be the root of Γ \Gamma Γ , and let e e e be the unique edge in Γ \Gamma Γ ending at r r r . We set ( y e , z e ) = ( f L ( Γ ) ( x 0 ) , f R ( Γ ) ( x n ) ) (y_{e},z_{e})=(f^{L(\Gamma)}(x_{0}),f^{R(\Gamma)}(x_{n})) ( y e , z e ) = ( f L ( Γ ) ( x 0 ) , f R ( Γ ) ( x n )) .
(c) Let v v v be an internal vertex of Γ \Gamma Γ . As in Definition 6.1(vi), write the edges of Γ \Gamma Γ in anticlockwise cyclic order as e v , 0 , e v , 1 , … , e v , k e_{v,0},e_{v,1},\ldots,e_{v,k} e v , 0 , e v , 1 , … , e v , k , where e v , 0 e_{v,0} e v , 0 is the outgoing edge. We require that (i) z e v , i = y e v , i + 1 z_{e_v,i}=y_{e_v,i+1} z e v , i = y e v , i + 1 in O min I v \mathcal{O}^{\min I_v} O m i n I v for all 1 ≤ i < k 1\le i<k 1 ≤ i < k .
(ii) y e v , 0 = f I v ( y e v , 1 ) y_{e_v,0}=f^{I_v}(y_{e_v,1}) y e v , 0 = f I v ( y e v , 1 ) and z e v , 0 = f min I v max I v ( z e v , k ) z_{e_v,0}=f^{\min I_v\max I_v}(z_{e_v,k}) z e v , 0 = f m i n I v m a x I v ( z e v , k ) in O max I v \mathcal{O}^{\max I_v} O m a x I v .
(d) Let b b b be a bud of Γ \Gamma Γ . As in Definition 6.1(x), write the leaves and buds of Γ \Gamma Γ as u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n , ordered anticlockwise around the outside of Γ \Gamma Γ starting at the root r r r . Then b b b lies in the list u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n , and the leaves l 1 , … , l n l_1,\ldots,l_n l 1 , … , l n appear in this order in the list u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n . Let j b = 0 , … , n j_b=0,\ldots,n j b = 0 , … , n be unique such that j b = 0 j_b=0 j b = 0 when b b b lies before l 1 l_1 l 1 in u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n , and b b b lies between l j b l_{j_b} l j b and l j b + 1 l_{j_b+1} l j b + 1 in u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n when 1 ≤ j b < n 1\le j_b<n 1 ≤ j b < n , and j b = n j_b=n j b = n when b b b lies after l n l_n l n in u 1 , … , u m + n u_1,\ldots,u_{m+n} u 1 , … , u m + n . We label b b b by the object x j b x_{j_b} x j b . Let e : b → v e\colon b\to v e : b → v be the unique edge in Γ \Gamma Γ starting at b b b , and P b P_b P b be the prefix of b b b . We require that
y e = f I b ∘ f P b ( x j b ) , z e = f min I b max I b ∘ f P b ( x j b ) . y_e=f^{I_b}\circ f^{P_b}(x_{j_b}),\qquad z_e=f^{\min I_b\max I_b}\circ f^{P_b}(x_{j_b}). y e = f I b ∘ f P b ( x j b ) , z e = f m i n I b m a x I b ∘ f P b ( x j b ) .
We can now show that there are unique labels ( y e , z e ) (y_e,z_e) ( y e , z e ) for each edge e e e and x j b x_{j_b} x j b for each bud b b b satisfying (a)–(d). This works because the condition (6.4) in the definition of Γ \Gamma Γ balanced is exactly what is required to ensure that z e v , i = y e v , i + 1 z_{e_v,i}=y_{e_v,i+1} z e v , i = y e v , i + 1 as in (c)(i), and the definitions of L ( Γ ) L(\Gamma) L ( Γ ) , R ( Γ ) R(\Gamma) R ( Γ ) are what is required to describe the target of (7.4) at the root r r r . We call these labels ( y e , z e ) (y_e,z_e) ( y e , z e ) , x j b x_{j_b} x j b object labels, and Γ \Gamma Γ with its object labels an object-labelled n n n -leaved I I I -labelled tree.
Figure 7.1 shows the tree Γ \Gamma Γ of Figure 6.6 labelled with objects.
Figure 7.1. An object-labelled, balanced { 0 , … , 8 } \{0,\ldots,8\} { 0 , … , 8 } -labelled tree
Now we have defined the labels ( y e , z e ) (y_e,z_e) ( y e , z e ) for each edge e e e of Γ \Gamma Γ , to each internal vertex v v v of Γ \Gamma Γ with ordered edges e v , 0 , e v , 1 , … , e v , k e_{v,0},e_{v,1},\ldots,e_{v,k} e v , 0 , e v , 1 , … , e v , k as in (c), we associate the list of objects w 0 , … , w k w_0,\ldots,w_k w 0 , … , w k in O min I v \mathcal{O}^{\min} I_v O m i n I v where w 0 = y e v , 1 w_0=y_{e_{v,1}} w 0 = y e v , 1 , w 1 = z e v , 1 = y e v , 2 , … , w k − 1 = z e v , k − 1 = y e v , k w_1=z_{e_{v,1}}=y_{e_{v,2}},\ldots,w_{k-1}=z_{e_{v,k-1}}=y_{e_{v,k}} w 1 = z e v , 1 = y e v , 2 , … , w k − 1 = z e v , k − 1 = y e v , k , w k = z e v , k w_k=z_{e_{v,k}} w k = z e v , k , and then we associate the R R R -linear map m k I v ; w 0 , … , w k m^{I_v;w_0,\ldots,w_k}_k m k I v ; w 0 , … , w k to v v v . To each bud b b b with label x j b x_{j_b} x j b and prefix P b P_b P b we associate the element
m 0 I b , f P b ( x j b ) ( 1 ) ∈ H max I b ( f I b ∘ f P b ( x j b ) , f min I b max I b ∘ f P b ( x j b ) ) . (197) m^{I_b,f^{P_b}(x_{j_b})}_0(1)\in\mathcal{H}^{\max I_b}\left(f^{I_b}\circ f^{P_b}(x_{j_b}),f^{\min I_b\max I_b\circ f^{P_b}}(x_{j_b})\right). \tag*{(197)} m 0 I b , f P b ( x j b ) ( 1 ) ∈ H m a x I b ( f I b ∘ f P b ( x j b ) , f m i n I b m a x I b ∘ f P b ( x j b ) ) . (197)
We define C ( Γ ) x 0 , … , x n C(\Gamma)^{x_0,\ldots,x_n} C ( Γ ) x 0 , … , x n in (7.4) to be the composition over all internal vertices v v v of Γ \Gamma Γ of the maps m k I v ; w 0 , … , w k m^{I_v;w_0,\ldots,w_k}_k m k I v ; w 0 , … , w k , and all buds b b b of the elements (7.5), as in Definition 6.2. As in Definition 6.6, C ( Γ ) x 0 , … , x n C(\Gamma)^{x_0,\ldots,x_n} C ( Γ ) x 0 , … , x n is graded of degree (6.3).
Definition 7.2. Let R R R , N N N , ( O i ) i = 0 N (\mathcal{O}^i)_{i=0}^N ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N (f^{ij})_{0\leq i\leq j\leq N} ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N (\mathcal{H}^i)_{i=0}^N ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } (m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}} ( m I ) ∅ = I ⊆ { 0 , … , N } be as in Definition 7.1. Choose ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M with M ≥ 0 M\geq0 M ≥ 0 . Suppose k , n ≥ 0 k,n\geq0 k , n ≥ 0 , and x 0 , … , x n ∈ A i 0 x_0,\ldots,x_n\in\mathcal{A}^{i_0} x 0 , … , x n ∈ A i 0 , and y 0 , … , y k ∈ A i M y_0,\ldots,y_k\in\mathcal{A}^{i_M} y 0 , … , y k ∈ A i M . Generalizing Definition 6.11, we will define an R R R -linear map graded of degree k − n + 2 − ∣ I ∣ k-n+2-|I| k − n + 2 − ∣ I ∣
d n , k ; y 0 , … , y k I ; x 0 , … , x n : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x n − 1 , x n ) ⟶ H i M ( y 0 , y 1 ) ⊗ R H i M ( y 1 , y 2 ) ⊗ R ⋯ ⊗ R H i M ( y k − 1 , y k ) . (198) d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}\colon
\mathcal{H}^{i_0}(x_0,x_1)\otimes_R\mathcal{H}^{i_0}(x_1,x_2)\otimes_R\cdots\otimes_R\mathcal{H}^{i_0}(x_{n-1},x_n)
\longrightarrow
\mathcal{H}^{i_M}(y_0,y_1)\otimes_R\mathcal{H}^{i_M}(y_1,y_2)\otimes_R\cdots\otimes_R\mathcal{H}^{i_M}(y_{k-1},y_k). \tag*{(198)} d n , k ; y 0 , … , y k I ; x 0 , … , x n : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x n − 1 , x n ) ⟶ H i M ( y 0 , y 1 ) ⊗ R H i M ( y 1 , y 2 ) ⊗ R ⋯ ⊗ R H i M ( y k − 1 , y k ) . (198)
Here if n = 0 n=0 n = 0 the domain H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x n − 1 , x n ) \mathcal{H}^{i_0}(x_0,x_1)\otimes\cdots\otimes\mathcal{H}^{i_0}(x_{n-1},x_n) H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x n − 1 , x n ) is R R R , as it is a tensor product of zero factors which is R R R by convention, and similarly if k = 0 k=0 k = 0 the target H i M ( y 0 , y 1 ) ⊗ ⋯ ⊗ H i M ( y k − 1 , y k ) \mathcal{H}^{i_M}(y_0,y_1)\otimes\cdots\otimes\mathcal{H}^{i_M}(y_{k-1},y_k) H i M ( y 0 , y 1 ) ⊗ ⋯ ⊗ H i M ( y k − 1 , y k ) is R R R .
(a) Suppose k = n = 0 k=n=0 k = n = 0 , and M = 1 M=1 M = 1 so I = { i 0 , i 1 } I=\{i_0,i_1\} I = { i 0 , i 1 } , and y 0 = f i 0 i 1 ( x 0 ) y_0=f^{i_0i_1}(x_0) y 0 = f i 0 i 1 ( x 0 ) . As in (6.9), we define
d 0 , 0 ; f i 0 i 1 ( x 0 ) { i 0 , i 1 } ; x 0 = id R : R ⟶ R . (199) d^{\{i_0,i_1\};x_0}_{0,0;f^{i_0i_1}(x_0)}=\operatorname{id}_R\colon R\longrightarrow R. \tag*{(199)} d 0 , 0 ; f i 0 i 1 ( x 0 ) { i 0 , i 1 } ; x 0 = id R : R ⟶ R . (199)
In all other cases when k = 0 k=0 k = 0 — if n ≠ 0 n\ne0 n = 0 , or M ≠ 1 M\ne1 M = 1 , or n = 0 n=0 n = 0 , M = 1 M=1 M = 1 and y 0 ≠ f i 0 i 1 ( x 0 ) y_0\ne f^{i_0i_1}(x_0) y 0 = f i 0 i 1 ( x 0 ) — we define d n , 0 ; y 0 I ; x 0 , … , x n = 0 d^{I;x_0,\ldots,x_n}_{n,0;y_0}=0 d n , 0 ; y 0 I ; x 0 , … , x n = 0 , as in (6.9).
(b) If M = 0 M=0 M = 0 and 1 ≤ k ≤ n 1\leq k\leq n 1 ≤ k ≤ n , generalizing (6.10) we define
d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = ∑ 0 ≤ j < k y i = x i , 0 ≤ i ≤ j y i = x n − k + i , j < i ≤ k ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) . (200) d^{\{i_0\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{\substack{0\leq j<k\\ y_i=x_i,\ 0\leq i\leq j\\ y_i=x_{n-k+i},\ j<i\leq k}}
(-1)^{j+(n-k+1)(k-j-1)}
\operatorname{id}_{\mathcal{H}^{i_0}(x_0,x_1)}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j-1},x_j)}
\otimes m^{\{i_0\};x_j,x_{j+1},\ldots,x_{j+n-k+1}}_{n-k+1}
\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j+n-k+1},x_{j+n-k+2})}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{n-1},x_n)}. \tag*{(200)} d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = 0 ≤ j < k y i = x i , 0 ≤ i ≤ j y i = x n − k + i , j < i ≤ k ∑ ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) . (200)
If M = 0 M=0 M = 0 and k > n k>n k > n we define d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = 0 d^{\{i_0\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}=0 d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = 0 . (c) If M > 0 M>0 M > 0 and k ≥ 1 k\ge1 k ≥ 1 then generalizing (170) , define
d n , k ; y 0 , … , y k I ; x 0 , … , x n = ∑ 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with l i − l i − 1 leaves , 1 ≤ i ≤ k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 , y i = f R ( Γ i ) ( x l i ) , i = 1 , … , k , y 0 = f I ( x 0 ) ( − 1 ) ( k + 1 ) ( M + n + 1 ) ( ∏ v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ and v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) C ( Γ 1 ) x l 0 , … , x l 1 ⊗ C ( Γ 2 ) x l 1 , … , x l 2 ⊗ ⋯ ⊗ C ( Γ k ) x l k − 1 , … , x l k . (201) d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{\substack{0=l_0\le l_1\le\cdots\le l_k=n,\\
\text{isomorphism classes}\\
\text{of }\Gamma_1,\ldots,\Gamma_k:\\
\Gamma_i\text{ is a balanced}\\
I\text{-labelled tree}\\
\text{with }l_i-l_{i-1}\\
\text{leaves},\ 1\le i\le k,\\
L(\Gamma_1)=I,\\
R(\Gamma_k)=\{i_0,i_M\},\\
R(\Gamma_i)=L(\Gamma_{i+1}),\\
i=1,\ldots,k-1,\\
y_i=f^{R(\Gamma_i)}(x_{l_i}),\\
i=1,\ldots,k,\ y_0=f^I(x_0)}}
(-1)^{(k+1)(M+n+1)}
(\prod_{\substack{v\text{ internal vertex or bud of }\Gamma_i,\ 1\le i\le k,\\
w\text{ internal vertex or bud of }\Gamma_{i'},\ 1\le i'\le k,\\
\text{either }i<i',\text{ or }i=i'\text{ and }v\prec w\text{ in order }\prec\text{ of }\Gamma_i,\\
v\text{ has }p\text{ incoming edges including }q\text{ leaves},\ w\text{ has }r\text{ incoming edges}}}
(-1)^{N(v,w)+q(r+1)+(p+1)(|I_w|+r+1)}
C(\Gamma_1)_{x_{l_0},\ldots,x_{l_1}}\otimes C(\Gamma_2)_{x_{l_1},\ldots,x_{l_2}}\otimes\cdots\otimes C(\Gamma_k)_{x_{l_{k-1}},\ldots,x_{l_k}} . \tag*{(201)} d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 = l 0 ≤ l 1 ≤ ⋯ ≤ l k = n , isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with l i − l i − 1 leaves , 1 ≤ i ≤ k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 , y i = f R ( Γ i ) ( x l i ) , i = 1 , … , k , y 0 = f I ( x 0 ) ∑ ( − 1 ) ( k + 1 ) ( M + n + 1 ) ( v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ and v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ∏ ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) C ( Γ 1 ) x l 0 , … , x l 1 ⊗ C ( Γ 2 ) x l 1 , … , x l 2 ⊗ ⋯ ⊗ C ( Γ k ) x l k − 1 , … , x l k . (201)
where we use the notation of §6.1–§6.2, and N ( v , w ) N(v,w) N ( v , w ) is as in (171) .
Here is the analogue of Example 6.13(i),(ii).
Example 7.3. (i) If k = 1 k=1 k = 1 then for any I I I , n n n , as in (172) we have
d n , 1 ; y 0 , y 1 I ; x 0 , … , x n = { m n I ; x 0 , … , x n , n ≥ 1 or ∣ I ∣ ≥ 3 , and y 0 = f I ( x 0 ) , y 1 = f i 0 i M ( x n ) , 0 , otherwise. (202) d^{I;x_0,\ldots,x_n}_{n,1;y_0,y_1}
=
\begin{cases}
m^{I;x_0,\ldots,x_n}_n, & n\ge1\text{ or }|I|\ge3,\text{ and }y_0=f^I(x_0),\ y_1=f^{i_0i_M}(x_n),\\
0, & \text{otherwise.}
\end{cases} \tag*{(202)} d n , 1 ; y 0 , y 1 I ; x 0 , … , x n = { m n I ; x 0 , … , x n , 0 , n ≥ 1 or ∣ I ∣ ≥ 3 , and y 0 = f I ( x 0 ) , y 1 = f i 0 i M ( x n ) , otherwise. (202)
(ii) If I = { 0 , 1 } I=\{0,1\} I = { 0 , 1 } and k ≥ 1 k\ge1 k ≥ 1 then as in (173) we have
d n , k ; y 0 , … , y k { 0 , 1 } ; x 0 , … , x n = ∑ n = n 1 + ⋯ + n k , n a ≥ 1 ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m n 1 { 0 , 1 } ; x p 0 , … , x p 1 ⊗ m n 2 { 0 , 1 } ; x p 1 , … , x p 2 ⊗ ⋯ ⊗ m n k { 0 , 1 } ; x p k − 1 , … , x p k , (203) d^{\{0,1\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{\substack{n=n_1+\cdots+n_k,\ n_a\ge1}}
(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}
m^{\{0,1\};x_{p_0},\ldots,x_{p_1}}_{n_1}\otimes
m^{\{0,1\};x_{p_1},\ldots,x_{p_2}}_{n_2}\otimes\cdots\otimes
m^{\{0,1\};x_{p_{k-1}},\ldots,x_{p_k}}_{n_k}, \tag*{(203)} d n , k ; y 0 , … , y k { 0 , 1 } ; x 0 , … , x n = n = n 1 + ⋯ + n k , n a ≥ 1 ∑ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m n 1 { 0 , 1 } ; x p 0 , … , x p 1 ⊗ m n 2 { 0 , 1 } ; x p 1 , … , x p 2 ⊗ ⋯ ⊗ m n k { 0 , 1 } ; x p k − 1 , … , x p k , (203)
where p j = ∑ i = 0 j n i p_j=\sum_{i=0}^{j}n_i p j = ∑ i = 0 j n i , j = 0 , … , k j=0,\ldots,k j = 0 , … , k . Then y j = f 01 ( x p j ) y_j=f^{01}(x_{p_j}) y j = f 01 ( x p j ) , j = 0 , … , k j=0,\ldots,k j = 0 , … , k .
Remark 7.4. In the defining equations (199) –(201) for d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k I ; x 0 , … , x n , to get a nonzero term the y j y_j y j are determined by the x i x_i x i by equations of the form y j = f J ( x i ) y_j=f^J(x_i) y j = f J ( x i ) for some J ⊆ I J\subseteq I J ⊆ I with i 0 , i M ∈ J i_0,i_M\in J i 0 , i M ∈ J , and if the y j y_j y j are not of this form then d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}=0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 . Thus, for fixed I I I , n n n , k k k , x 0 , … , x n x_0,\ldots,x_n x 0 , … , x n , there are only finitely many lists y 0 , … , y k y_0,\ldots,y_k y 0 , … , y k such that d n , k ; y 0 , … , y k I ; x 0 , … , x n ≠ 0 d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}\ne0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 .
Below, we will give equations involving ∑ y 0 , … , y k ( ⋯ ) ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n \sum_{y_0,\ldots,y_k}(\cdots)\circ d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} ∑ y 0 , … , y k ( ⋯ ) ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n , such as (205) , (207) and (208) . Although these involve apparently infinite sums over objects y 0 , … , y k y_0,\ldots,y_k y 0 , … , y k , there are only finitely many choices for which d n , k ; y 0 , … , y k I ; x 0 , … , x n ≠ 0 d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}\ne0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 , so they are in effect finite sums, with no convergence issues.
Here is the analogue of Lemma 6.14, proved in the same way.
Lemma 7.5. d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}=0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 above unless n ≥ min ( k , k + 2 − ∣ I ∣ ) n\ge\min(k,k+2-|I|) n ≥ min ( k , k + 2 − ∣ I ∣ ) .
Here is the analogue of Theorem 6.15, proved in §10.1.
Theorem 7.6. In Definition 7.2, let k = k ′ + k ′ ′ k=k'+k'' k = k ′ + k ′′ with k ′ , k ′ ′ ≥ 1 k',k''\geq1 k ′ , k ′′ ≥ 1 . Then
(a) If M = 0 M=0 M = 0 then
d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = { ( − 1 ) k ′ ′ ( k ′ + n ) d n − k ′ ′ , k ′ ; y 0 , … , y k ′ { i 0 } ; x 0 , … , x n − k ′ ′ ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) if y i = x n − k + i , k ′ ≤ i ≤ k , 0 , otherwise + { ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ d n − k ′ , k ′ ′ ; y k ′ , … , y k { i 0 } ; x k ′ , … , x n if y i = x i , 0 ≤ i ≤ k ′ , 0 , otherwise. (204) d^{\{i_0\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\begin{cases}
(-1)^{k''(k'+n)}d^{\{i_0\};x_0,\ldots,x_{n-k''}}_{n-k'',k';y_0,\ldots,y_{k'}}\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(y_{k'},y_{k'+1})}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(y_{k-1},y_k)}
& \text{if }y_i=x_{n-k+i},\ k'\leq i\leq k,\\
0,&\text{otherwise}
\end{cases}
+
\begin{cases}
(-1)^{k'}\operatorname{id}_{\mathcal{H}^{i_0}(y_0,y_1)}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(y_{k'-1},y_{k'})}\otimes
d^{\{i_0\};x_{k'},\ldots,x_n}_{n-k',k'';y_{k'},\ldots,y_k}
& \text{if }y_i=x_i,\ 0\leq i\leq k',\\
0,&\text{otherwise.}
\end{cases} \tag*{(204)} d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = { ( − 1 ) k ′′ ( k ′ + n ) d n − k ′′ , k ′ ; y 0 , … , y k ′ { i 0 } ; x 0 , … , x n − k ′′ ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) 0 , if y i = x n − k + i , k ′ ≤ i ≤ k , otherwise + { ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ d n − k ′ , k ′′ ; y k ′ , … , y k { i 0 } ; x k ′ , … , x n 0 , if y i = x i , 0 ≤ i ≤ k ′ , otherwise. (204) (b) If M > 0 M>0 M > 0 then
d n , k ; y 0 , … , y k I ; x 0 , … , x n = ∑ n = n ′ + n ′ ′ , n ′ , n ′ ′ ≥ 0 , 0 < l ≤ M , 0 = j 0 < j 1 < ⋯ < j l = M , k 0 , … , k l ≥ 0 , k 0 = k ′ , k l = k ′ ′ , w 0 i , … , w k i i ∈ O i , 0 < i < l ( − 1 ) k ′ ′ ( k ′ + l + M ) + n ′ ′ ( k ′ ′ + l + 1 ) + ∑ a = 1 l − 1 ( j a − a ) ⋅ ( d k l − 1 , k l ; y 0 , … , y k ′ { i j l − 1 , … , i j l } ; w 0 l − 1 , … , w k l − 1 l − 1 ∘ d k l − 2 , k l − 1 ; w 0 l − 1 , … , w k l − 1 l − 1 { i j l − 2 , … , i j l − 1 } ; w 0 l − 2 , … , w k l − 2 l − 2 ∘ ⋯ ∘ d k 0 , k 1 ; w 0 1 , … , w k 1 1 { i j 0 , … , i j 1 } ; x 0 , … , x n ′ ) ⊗ d n ′ ′ , k ′ ′ ; y k ′ , y k ′ + 1 , … , y k { i j 0 , i j 1 , i j 2 , … , i j l } ; x n ′ , x n ′ + 1 , … , x n , (205) \begin{aligned}
d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
={}&
\sum_{\substack{n=n'+n'',\ n',n''\geq0,\\
0<l\leq M,\\
0=j_0<j_1<\cdots<j_l=M,\\
k_0,\ldots,k_l\geq0,\ k_0=k',\ k_l=k'',\\
w^i_0,\ldots,w^i_{k_i}\in\mathcal{O}^i,\ 0<i<l}}
(-1)^{k''(k'+l+M)+n''(k''+l+1)+\sum_{a=1}^{l-1}(j_a-a)}
\\
&\quad\cdot
\left(
d^{\{i_{j_{l-1}},\ldots,i_{j_l}\};w^{l-1}_0,\ldots,w^{l-1}_{k_{l-1}}}_{k_{l-1},k_l;y_0,\ldots,y_{k'}}
\circ
d^{\{i_{j_{l-2}},\ldots,i_{j_{l-1}}\};w^{l-2}_0,\ldots,w^{l-2}_{k_{l-2}}}_{k_{l-2},k_{l-1};w^{l-1}_0,\ldots,w^{l-1}_{k_{l-1}}}
\circ\cdots\circ
d^{\{i_{j_0},\ldots,i_{j_1}\};x_0,\ldots,x_{n'}}_{k_0,k_1;w^1_0,\ldots,w^1_{k_1}}
\right)
\\
&\quad\otimes
d^{\{i_{j_0},i_{j_1},i_{j_2},\ldots,i_{j_l}\};x_{n'},x_{n'+1},\ldots,x_n}_{n'',k'';y_{k'},y_{k'+1},\ldots,y_k}, \tag*{(205)}
\end{aligned} d n , k ; y 0 , … , y k I ; x 0 , … , x n = n = n ′ + n ′′ , n ′ , n ′′ ≥ 0 , 0 < l ≤ M , 0 = j 0 < j 1 < ⋯ < j l = M , k 0 , … , k l ≥ 0 , k 0 = k ′ , k l = k ′′ , w 0 i , … , w k i i ∈ O i , 0 < i < l ∑ ( − 1 ) k ′′ ( k ′ + l + M ) + n ′′ ( k ′′ + l + 1 ) + ∑ a = 1 l − 1 ( j a − a ) ⋅ ( d k l − 1 , k l ; y 0 , … , y k ′ { i j l − 1 , … , i j l } ; w 0 l − 1 , … , w k l − 1 l − 1 ∘ d k l − 2 , k l − 1 ; w 0 l − 1 , … , w k l − 1 l − 1 { i j l − 2 , … , i j l − 1 } ; w 0 l − 2 , … , w k l − 2 l − 2 ∘ ⋯ ∘ d k 0 , k 1 ; w 0 1 , … , w k 1 1 { i j 0 , … , i j 1 } ; x 0 , … , x n ′ ) ⊗ d n ′′ , k ′′ ; y k ′ , y k ′ + 1 , … , y k { i j 0 , i j 1 , i j 2 , … , i j l } ; x n ′ , x n ′ + 1 , … , x n , (205) where there are only finitely many nonzero terms in the sum.
The ∞ \infty ∞ -categories A ∞ Cat R u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , A ∞ Cat R , K u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ This section is the analogue of §6.3. We first define a simplicial set A ∞ Cat ^ R u n , ∞ \widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ of unital A ∞ A_{\infty} A ∞ -categories over R R R , and then define our ∞ \infty ∞ -category as a subset A ∞ Cat R u n , ∞ ⊂ A ∞ Cat ^ R u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ . Theorem 7.17 extends A ∞ Cat ^ R u n , ∞ \widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ to an ∞ \infty ∞ -bicategory ( A ∞ Cat ^ R u n , ∞ , T ) (\widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R},T) ( A ∞ Cat R un , ∞ , T ) , as in §2.8, with A ∞ Cat R u n , ∞ = Core ( A ∞ Cat ^ R u n , ∞ , T ) A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}=\operatorname{Core}(\widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R},T) A ∞ Cat R un , ∞ = Core ( A ∞ Cat R un , ∞ , T ) .
Definition 7.7. Let R R R be a commutative ring. We will define a simplicial set A ∞ Cat ^ R u n , ∞ \widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ of unital A ∞ A_{\infty} A ∞ -categories over R R R .
(a) For N ≥ 0 N\geq0 N ≥ 0 , define the N N N -simplices ( A ∞ Cat ^ R u n , ∞ ) N (\widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R})_N ( A ∞ Cat R un , ∞ ) N in A ∞ Cat ^ R u n , ∞ \widehat{A_{\infty}\operatorname{Cat}}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ to be the set of quadruples
( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } , (\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(m^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}, ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } , where:
(i) O i \mathcal{O}^i O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , thought of as the set of objects Obj A i \operatorname{Obj}\mathcal{A}^i Obj A i of a (small) A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^i A i at vertex i i i of the N N N -simplex Δ N \Delta^N Δ N .
(ii) f i j : O i → O j f^{ij}:\mathcal{O}^i\to\mathcal{O}^j f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , thought of as the map f ~ Obj i j : Obj A i → Obj A j \widetilde{f}^{ij}_{\operatorname{Obj}}:\operatorname{Obj}\mathcal{A}^i\to\operatorname{Obj}\mathcal{A}^j f Obj ij : Obj A i → Obj A j in an A ∞ A_{\infty} A ∞ -functor f ~ i j : A i → A j \widetilde{f}^{ij}:\mathcal{A}^i\to\mathcal{A}^j f ij : A i → A j living on the edge [ i , j ] [i,j] [ i , j ] of Δ N \Delta^N Δ N . We require these maps to satisfy f i i = id O i f^{ii}=\operatorname{id}_{\mathcal{O}^i} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N . We do not require that f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik .
Suppose ∅ ≠ I ⊆ { 0 , 1 , … , N } \varnothing\neq I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } with I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M . Define f I : O i 0 → O i M f^I:\mathcal{O}^{i_0}\to\mathcal{O}^{i_M} f I : O i 0 → O i M as in (7.2).
(iii) For all i = 0 , 1 , … , N i = 0, 1, \ldots, N i = 0 , 1 , … , N and x , y ∈ O i x, y \in\mathcal{O}^{i} x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded R R R -module H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d \mathcal{H}^{i}(x,y) = \bigoplus_{d\in\mathbb{Z}}\mathcal{H}^{i}(x,y)^{d} H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d , thought of as the Hom space Hom A i ( x , y ) \operatorname{Hom}_{\mathcal{A}^{i}}(x,y) Hom A i ( x , y ) in the A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i . We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^{i} = (\mathcal{H}^{i}(x,y))_{x,y\in\mathcal{O}^{i}} H i = ( H i ( x , y ) ) x , y ∈ O i .
(iv) For each nonempty subset I ⊂ { 0 , … , N } I \subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I = \{i_{0}, i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0} < i_{1} < \cdots< i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M \ge0 M ≥ 0 , and for all k ≥ 0 k \ge0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_{0},\ldots,x_{k} \in\mathcal{O}^{i_{0}} x 0 , … , x k ∈ O i 0 , thought of as objects in A i 0 \mathcal{A}^{i_{0}} A i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) (206) m_{k}^{I;x_{0},\ldots,x_{k}}:\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{i_{0}}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{i_{0}}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k})) \tag*{(206)} m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k )) (206)
graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k . Here when k = 0 k=0 k = 0 , by convention the empty tensor product on the top line of (206) is R R R , so m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 maps
m 0 I ; x 0 : R ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M . m_{0}^{I;x_{0}}:R\longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0}))^{2-M}. m 0 I ; x 0 : R ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M .
We sometimes identify m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 with m 0 I ; x 0 ( 1 ) ∈ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M m_{0}^{I;x_{0}}(1)\in\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0}))^{2-M} m 0 I ; x 0 ( 1 ) ∈ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M .
In fact when k = 0 k=0 k = 0 we only want maps m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 for ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 , so we require that m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . Alternatively, we can say that m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k in (206) should only be defined for k ≥ 1 k\ge1 k ≥ 1 when ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 and for k ≥ 0 k\ge0 k ≥ 0 when ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 , and then we should exclude terms in (208) and other equations below involving m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 for ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 .
We write m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 m_{k}^{I}=(m_{k}^{I;x_{0},\ldots,x_{k}})_{x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}}} m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 and m I = ( m k I ) k ≥ 0 m^{I}=(m_{k}^{I})_{k\ge0} m I = ( m k I ) k ≥ 0 .
We require these maps m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k to satisfy the following identities generalizing (6.18) for all ∅ ≠ I = { i 0 , i 1 , … , i M } ⊆ { 0 , 1 , … , N } \varnothing\ne I=\{i_{0},i_{1},\ldots,i_{M}\}\subseteq\{0,1,\ldots,N\} ∅ = I = { i 0 , i 1 , … , i M } ⊆ { 0 , 1 , … , N } , n ≥ 0 n\ge0 n ≥ 0 , and x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 , firstly if $M=0
0 = ∑ k = 1 , … , n , y 0 , … , y k ∈ O i 0 m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n , (207) 0=\sum_{\substack{k=1,\ldots,n,\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m_{k}^{\{i_{0}\};y_{0},\ldots,y_{k}}\circ d_{n,k;y_{0},\ldots,y_{k}}^{\{i_{0}\};x_{0},\ldots,x_{n}}, \tag*{(207)} 0 = k = 1 , … , n , y 0 , … , y k ∈ O i 0 ∑ m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n , (207)
and secondly if $M>0
0 = ∑ k = 1 , … , n y 0 , … , y k ∈ O i 0 m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M ∑ k = 1 , … , n + M − 1 y 0 , … , y k ∈ O i M m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (208) \begin{aligned}
0={}&\sum_{\substack{k=1,\ldots,n\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m_{k}^{I;y_{0},\ldots,y_{k}}\circ d_{n,k;y_{0},\ldots,y_{k}}^{\{i_{0}\};x_{0},\ldots,x_{n}}\\
&+(-1)^{M}\sum_{\substack{k=1,\ldots,n+M-1\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}}}}m_{k}^{\{i_{M}\};y_{0},\ldots,y_{k}}\circ d_{n,k;y_{0},\ldots,y_{k}}^{I;x_{0},\ldots,x_{n}}. \tag*{(208)}
\end{aligned} 0 = k = 1 , … , n y 0 , … , y k ∈ O i 0 ∑ m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M k = 1 , … , n + M − 1 y 0 , … , y k ∈ O i M ∑ m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (208)
Here d n , k ; y 0 , … , y k I ; x 0 , … , x n d_{n,k;y_{0},\ldots,y_{k}}^{I;x_{0},\ldots,x_{n}} d n , k ; y 0 , … , y k I ; x 0 , … , x n is defined using the m l J ; w 0 , … , w l m_{l}^{J;w_{0},\ldots,w_{l}} m l J ; w 0 , … , w l as in Definition 7.2. In both (7.15) and (7.16) there are only finitely many nonzero terms, and all nonzero terms map to H i M ( f I ( x 0 ) , f i 0 i M ( x n ) ) \mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{n})) H i M ( f I ( x 0 ) , f i 0 i M ( x n )) , so (7.15)–(7.16) should be regarded as equations in this space.
(v) There should exist elements 1 x i ∈ H i ( x , x ) 0 1_x^i \in\mathcal{H}^i(x,x)^0 1 x i ∈ H i ( x , x ) 0 for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N and x ∈ O i x\in\mathcal{O}^i x ∈ O i , which are uniquely determined by (209) , satisfying
m 2 { i } ; x 0 , x 0 , x 1 ( 1 x 0 i ⊗ a ) = m 2 { i } ; x 0 , x 1 , x 1 ( a ⊗ 1 x 1 i ) = a (209) m_2^{\{i\};x_0,x_0,x_1}(1^i_{x_0}\otimes a)=m_2^{\{i\};x_0,x_1,x_1}(a\otimes1^i_{x_1})=a \tag*{(209)} m 2 { i } ; x 0 , x 0 , x 1 ( 1 x 0 i ⊗ a ) = m 2 { i } ; x 0 , x 1 , x 1 ( a ⊗ 1 x 1 i ) = a (209)
for all i = 0 , … , N i=0,\ldots,N i = 0 , … , N , x 0 , x 1 ∈ O i x_0,x_1\in\mathcal{O}^i x 0 , x 1 ∈ O i and a ∈ H i ( x 0 , x 1 ) a\in\mathcal{H}^i(x_0,x_1) a ∈ H i ( x 0 , x 1 ) .
m 1 { i , j } ; x , x ( 1 x i ) = 1 f i j ( x ) i j (210) m_1^{\{i,j\};x,x}(1_x^i)=1^{ij}_{f^{ij}(x)} \tag*{(210)} m 1 { i , j } ; x , x ( 1 x i ) = 1 f ij ( x ) ij (210)
for all 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N and x ∈ O i x\in\mathcal{O}^i x ∈ O i ,
m k + 1 I ; x 0 , … , x j , x j , … , x k ( a 1 ⊗ ⋯ ⊗ a j ⊗ 1 x j i 0 ⊗ a j + 1 ⊗ ⋯ ⊗ a k ) = 0 (211) m_{k+1}^{I;x_0,\ldots,x_j,x_j,\ldots,x_k}(a_1\otimes\cdots\otimes a_j\otimes1^{i_0}_{x_j}\otimes a_{j+1}\otimes\cdots\otimes a_k)=0 \tag*{(211)} m k + 1 I ; x 0 , … , x j , x j , … , x k ( a 1 ⊗ ⋯ ⊗ a j ⊗ 1 x j i 0 ⊗ a j + 1 ⊗ ⋯ ⊗ a k ) = 0 (211)
for all I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_0<i_1<\cdots<i_M\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } , 0 ≤ j ≤ k 0\leq j\leq k 0 ≤ j ≤ k , x 0 , … , x k ∈ O i 0 x_0,\ldots,x_k\in\mathcal{O}^{i_0} x 0 , … , x k ∈ O i 0 , and a l ∈ H i 0 ( x l − 1 , x l ) a_l\in\mathcal{H}^{i_0}(x_{l-1},x_l) a l ∈ H i 0 ( x l − 1 , x l ) , l = 1 , … , k l=1,\ldots,k l = 1 , … , k , provided ( M , k ) ≠ ( 0 , 1 ) , ( 1 , 0 ) (M,k)\ne(0,1),(1,0) ( M , k ) = ( 0 , 1 ) , ( 1 , 0 ) .
The exclusion ( M , k ) ≠ ( 0 , 1 ) , ( 1 , 0 ) (M,k)\ne(0,1),(1,0) ( M , k ) = ( 0 , 1 ) , ( 1 , 0 ) in (211) is because (209) prescribes the case ( M , k ) = ( 0 , 1 ) (M,k)=(0,1) ( M , k ) = ( 0 , 1 ) , and (210) the case ( M , k ) = ( 1 , 0 ) (M,k)=(1,0) ( M , k ) = ( 1 , 0 ) .
(b) If α : [ N ˉ ] → [ N ] \alpha:[\bar N]\to[N] α : [ N ˉ ] → [ N ] is a morphism in Δ \Delta Δ , that is, α : { 0 , … , N ˉ } → { 0 , … , N } \alpha:\{0,\ldots,\bar N\}\to\{0,\ldots,N\} α : { 0 , … , N ˉ } → { 0 , … , N } is nondecreasing, let A ^ ∞ Cat R u n , ∞ ( α ) : ( A ^ ∞ Cat R u n , ∞ ) N → A ^ ∞ Cat R , N ˉ u n , ∞ \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\alpha):(\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R)^N\to\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,\bar N} A ∞ Cat R un , ∞ ( α ) : ( A ∞ Cat R un , ∞ ) N → A ∞ Cat R , N ˉ un , ∞ map
( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } )
to
( ( O ˉ i ) i = 0 N ˉ , ( f ˉ i j ) 0 ≤ i ≤ j ≤ N ˉ , ( H ˉ i ) i = 0 N ˉ , ( m ˉ I ˉ ) ∅ ≠ I ˉ ⊆ { 0 , … , N ˉ } ) , ((\bar{\mathcal{O}}^i)_{i=0}^{\bar N},(\bar f^{ij})_{0\leq i\leq j\leq\bar N},(\bar{\mathcal{H}}^i)_{i=0}^{\bar N},(\bar m^{\bar I})_{\varnothing\ne\bar I\subseteq\{0,\ldots,\bar N\}}), (( O ˉ i ) i = 0 N ˉ , ( f ˉ ij ) 0 ≤ i ≤ j ≤ N ˉ , ( H ˉ i ) i = 0 N ˉ , ( m ˉ I ˉ ) ∅ = I ˉ ⊆ { 0 , … , N ˉ } ) ,
where O ˉ i = O α ( i ) \bar{\mathcal{O}}^i=\mathcal{O}^{\alpha(i)} O ˉ i = O α ( i ) for i = 0 , … , N ˉ i=0,\ldots,\bar N i = 0 , … , N ˉ , f ˉ i j = f α ( i ) α ( j ) \bar f^{ij}=f^{\alpha(i)\alpha(j)} f ˉ ij = f α ( i ) α ( j ) for 0 ≤ i ≤ j ≤ N ˉ 0\leq i\leq j\leq\bar N 0 ≤ i ≤ j ≤ N ˉ , and H ˉ i = H α ( i ) \bar{\mathcal{H}}^i=\mathcal{H}^{\alpha(i)} H ˉ i = H α ( i ) for i = 0 , … , N ˉ i=0,\ldots,\bar N i = 0 , … , N ˉ . To define m ˉ I ˉ \bar m^{\bar I} m ˉ I ˉ , if I ˉ = { ı ˉ 0 < ı ˉ 1 < ⋯ < ı ˉ M ˉ } ⊆ { 0 , 1 , … , N ˉ } \bar I=\{\bar\imath_0<\bar\imath_1<\cdots<\bar\imath_{\bar M}\}\subseteq\{0,1,\ldots,\bar N\} I ˉ = { ˉ 0 < ˉ 1 < ⋯ < ˉ M ˉ } ⊆ { 0 , 1 , … , N ˉ } , k ≥ 0 k\geq0 k ≥ 0 and x 0 , … , x k ∈ O ˉ ı ˉ 0 = O α ( ı ˉ 0 ) x_0,\ldots,x_k\in\bar{\mathcal{O}}^{\bar\imath_0}=\mathcal{O}^{\alpha(\bar\imath_0)} x 0 , … , x k ∈ O ˉ ˉ 0 = O α ( ˉ 0 ) , generalizing (6.22) we define
m ˉ k I ˉ ; x 0 , … , x k = { m k α ( I ˉ ) ; x 0 , … , x k , if α ∣ I ˉ is injective , id H i ( x 0 , x 1 ) , if k = 1 and I ˉ = { ı ˉ 0 , ı ˉ 1 } with ı ˉ 0 ≠ ı ˉ 1 , α ( ı ˉ 0 ) = α ( ı ˉ 1 ) = i , 1 f ı ˉ 0 ı ˉ 2 ( x 0 ) ı ˉ 2 , if k = 0 , I ˉ = { ı ˉ 0 < ı ˉ 1 < ı ˉ 2 } and α ∣ I ˉ is not injective , 0 , otherwise. (212) \bar m_k^{\bar I;x_0,\ldots,x_k}=
\begin{cases}
m_k^{\alpha(\bar I);x_0,\ldots,x_k}, & \text{if }\alpha|_{\bar I}\text{ is injective},\\
\operatorname{id}_{\mathcal{H}^i}(x_0,x_1), & \text{if }k=1\text{ and }\bar I=\{\bar\imath_0,\bar\imath_1\}\text{ with }\bar\imath_0\ne\bar\imath_1,\ \alpha(\bar\imath_0)=\alpha(\bar\imath_1)=i,\\
1^{\bar\imath_2}_{f^{\bar\imath_0\bar\imath_2}(x_0)}, & \text{if }k=0,\ \bar I=\{\bar\imath_0<\bar\imath_1<\bar\imath_2\}\text{ and }\alpha|_{\bar I}\text{ is not injective},\\
0, & \text{otherwise.}
\end{cases} \tag*{(212)} m ˉ k I ˉ ; x 0 , … , x k = ⎩ ⎨ ⎧ m k α ( I ˉ ) ; x 0 , … , x k , id H i ( x 0 , x 1 ) , 1 f ˉ 0 ˉ 2 ( x 0 ) ˉ 2 , 0 , if α ∣ I ˉ is injective , if k = 1 and I ˉ = { ˉ 0 , ˉ 1 } with ˉ 0 = ˉ 1 , α ( ˉ 0 ) = α ( ˉ 1 ) = i , if k = 0 , I ˉ = { ˉ 0 < ˉ 1 < ˉ 2 } and α ∣ I ˉ is not injective , otherwise. (212)
We will show in Proposition 7.9 that A ^ ∞ Cat R u n , ∞ ( α ) \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\alpha) A ∞ Cat R un , ∞ ( α ) maps to A ^ ∞ Cat R , N ˉ u n , ∞ \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,\bar N} A ∞ Cat R , N ˉ un , ∞ .
Clearly, if β : [ N ^ ] → [ N ˉ ] \beta:[\widehat N]\to[\bar N] β : [ N ] → [ N ˉ ] is another morphism in Δ \Delta Δ then
A ^ ∞ Cat R u n , ∞ ( α ∘ β ) = A ^ ∞ Cat R u n , ∞ ( β ) ∘ A ^ ∞ Cat R u n , ∞ ( α ) : ( A ^ ∞ Cat R u n , ∞ ) N ⟶ A ^ ∞ Cat R , N ^ u n , ∞ . \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\alpha\circ\beta)
=
\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\beta)\circ
\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\alpha):
(\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R)^N
\longrightarrow
\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,\widehat N}. A ∞ Cat R un , ∞ ( α ∘ β ) = A ∞ Cat R un , ∞ ( β ) ∘ A ∞ Cat R un , ∞ ( α ) : ( A ∞ Cat R un , ∞ ) N ⟶ A ∞ Cat R , N un , ∞ .
Also A ^ ∞ Cat R u n , ∞ ( id [ N ] ) = id ( A ^ ∞ Cat R u n , ∞ ) N \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R(\operatorname{id}_{[N]})=\operatorname{id}_{(\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R)^N} A ∞ Cat R un , ∞ ( id [ N ] ) = id ( A ∞ Cat R un , ∞ ) N . Hence A ^ ∞ Cat R u n , ∞ : Δ o p → S e t \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R:\Delta^{\mathrm{op}}\to\mathrm{Set} A ∞ Cat R un , ∞ : Δ op → Set is a functor, that is, A ^ ∞ Cat R u n , ∞ \widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is a simplicial set.
Remark 7.8. (On Set Theory foundations.) Recall from Remark 1.23 that we fix a Grothendieck universe U \mathcal{U} U , and implicitly assume all sets in our theory are U \mathcal{U} U -small. As in Remark 6.17, without this convention, the N N N -simplices ( A ^ ∞ Cat R u n , ∞ ) N (\widehat{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_R)_N ( A ∞ Cat R un , ∞ ) N above would be proper classes, rather than sets. This would cause Set Theory problems.
The following analogue of Proposition 6.18 will be proved in §10.2.
Proposition 7.9. The data ( ( O ˉ i ) i = 0 N ˉ , ( f ˉ i j ) 0 ≤ i ≤ j ≤ N ˉ , ( H ˉ i ) i = 0 N ˉ , ( m ˉ I ) ∅ ≠ I ⊂ { 0 , … , N ˉ } ) \left((\bar{\mathcal{O}}^{i})_{i=0}^{\bar{N}},(\bar{f}^{ij})_{0\leq i\leq j\leq\bar{N}},(\bar{\mathcal{H}}^{i})_{i=0}^{\bar{N}},(\bar{m}^{I})_{\varnothing\ne I\subset\{0,\ldots,\bar{N}\}}\right) ( ( O ˉ i ) i = 0 N ˉ , ( f ˉ ij ) 0 ≤ i ≤ j ≤ N ˉ , ( H ˉ i ) i = 0 N ˉ , ( m ˉ I ) ∅ = I ⊂ { 0 , … , N ˉ } ) in Definition 7.7(b) lies in ( A ^ ∞ Cat R u n , ∞ ) N ˉ (\widehat{\mathcal{A}}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty})_{\bar{N}} ( A ∞ Cat R un , ∞ ) N ˉ .
The following analogue of Theorem 6.19 will be proved in §10.3.
Theorem 7.10. Let ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subset\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) be an N N N -simplex in A ^ ∞ Cat R u n , ∞ \widehat{\mathcal{A}}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ as in Definition 7.7, and define maps d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n from the m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k as in (7.6)–(7.9). Then for each nonempty subset I ⊂ { 0 , … , N } I\subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , written I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M , and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 and y 0 , … , y k ∈ O i M y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}} y 0 , … , y k ∈ O i M , in R R R -linear maps as in (7.6) we have
∑ k ≤ l ≤ n , w 0 , … , w l ∈ O i 0 d l , k ; y 0 , … , y k { i 0 } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n = 0 if M = 0 , \sum_{\substack{k\leq l\leq n,\\ w_{0},\ldots,w_{l}\in\mathcal{O}^{i_{0}}}}d^{\{i_{0}\};w_{0},\ldots,w_{l}}_{l,k;y_{0},\ldots,y_{k}}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,l;w_{0},\ldots,w_{l}}=0
\qquad\text{if }M=0, k ≤ l ≤ n , w 0 , … , w l ∈ O i 0 ∑ d l , k ; y 0 , … , y k { i 0 } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n = 0 if M = 0 , ∑ 1 ≤ l ≤ n , w 0 , … , w l ∈ O i 0 d l , k ; y 0 , … , y k I ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n + ( − 1 ) M ∑ k ≤ l ≤ n + M − 1 , w 0 , … , w l ∈ O i M d l , k { i M } ∘ d n , l ; w 0 , … , w l I ; x 0 , … , x n = 0 if M > 0. (213) \sum_{\substack{1\leq l\leq n,\\ w_{0},\ldots,w_{l}\in\mathcal{O}^{i_{0}}}}d^{I;w_{0},\ldots,w_{l}}_{l,k;y_{0},\ldots,y_{k}}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,l;w_{0},\ldots,w_{l}}
+(-1)^{M}\sum_{\substack{k\leq l\leq n+M-1,\\ w_{0},\ldots,w_{l}\in\mathcal{O}^{i_{M}}}}d^{\{i_{M}\}}_{l,k}\circ d^{I;x_{0},\ldots,x_{n}}_{n,l;w_{0},\ldots,w_{l}}=0
\qquad\text{if }M>0. \tag*{(213)} 1 ≤ l ≤ n , w 0 , … , w l ∈ O i 0 ∑ d l , k ; y 0 , … , y k I ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n + ( − 1 ) M k ≤ l ≤ n + M − 1 , w 0 , … , w l ∈ O i M ∑ d l , k { i M } ∘ d n , l ; w 0 , … , w l I ; x 0 , … , x n = 0 if M > 0. (213) where there are only finitely many nonzero terms in the sums.
Here is the analogue of Example 6.20.
Example 7.11. (a) A 0-simplex ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) (\mathcal{O}^{0},f^{00}=\operatorname{id}_{\mathcal{O}^{0}},\mathcal{H}^{0},m^{\{0\}}) ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) of A ^ ∞ Cat R u n , ∞ \widehat{\mathcal{A}}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to a (small) unital A ∞ A_{\infty} A ∞ -category A 0 \mathcal{A}^{0} A 0 over R R R , as in Definition 4.5, with objects Obj A 0 = O 0 \operatorname{Obj}\mathcal{A}^{0}=\mathcal{O}^{0} Obj A 0 = O 0 , Hom spaces Hom A 0 ( x , y ) = H 0 ( x , y ) \operatorname{Hom}_{\mathcal{A}^{0}}(x,y)=\mathcal{H}^{0}(x,y) Hom A 0 ( x , y ) = H 0 ( x , y ) , morphisms ( m { 0 } ; x 0 , … , x k ) k ≥ 1 (m^{\{0\};x_{0},\ldots,x_{k}})_{k\geq1} ( m { 0 } ; x 0 , … , x k ) k ≥ 1 , and units 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 1_{x}^{0}\in\operatorname{Hom}_{\mathcal{A}^{0}}^{0}(x,x)=\mathcal{H}^{0}(x,x) 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) .
(b) A 1-simplex of A ^ ∞ Cat R u n , ∞ \widehat{\mathcal{A}}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to a unital A ∞ A_{\infty} A ∞ -functor F 01 : A 0 → A 1 \mathcal{F}^{01}:\mathcal{A}^{0}\to\mathcal{A}^{1} F 01 : A 0 → A 1 of unital A ∞ A_{\infty} A ∞ -categories, as in Definition 4.9, with object map F O b j 01 = f 01 \mathcal{F}^{01}_{\mathrm{Obj}}=f^{01} F Obj 01 = f 01 and morphisms F k 01 ; x 0 , … , x k = m { 0 , 1 } ; x 0 , … , x k \mathcal{F}^{01;x_{0},\ldots,x_{k}}_{k}=m^{\{0,1\};x_{0},\ldots,x_{k}} F k 01 ; x 0 , … , x k = m { 0 , 1 } ; x 0 , … , x k .
(c) A 2-simplex of A ^ ∞ Cat R u n , ∞ \widehat{\mathcal{A}}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to unital A ∞ A_{\infty} A ∞ -categories A 0 , A 1 , A 2 \mathcal{A}^{0},\mathcal{A}^{1},\mathcal{A}^{2} A 0 , A 1 , A 2 over R R R as in (a), and unital A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), and R R R -linear maps for all k ≥ 0 k\geq0 k ≥ 0 and x 0 , … , x k ∈ O 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{0} x 0 , … , x k ∈ O 0
m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 12 ∘ f 01 ( x 0 ) , f 02 ( x k ) ) m^{\{0,1,2\};x_{0},\ldots,x_{k}}_{k}:\mathcal{H}^{0}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{0}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{0}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{2}(f^{12}\circ f^{01}(x_{0}),f^{02}(x_{k})) m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 12 ∘ f 01 ( x 0 ) , f 02 ( x k ))
graded of degree − k -k − k , satisfying (7.19) and for all n ≥ 0 n \ge0 n ≥ 0 and x 0 , … , x n ∈ O 0 x_0,\ldots,x_n \in\mathcal{O}^0 x 0 , … , x n ∈ O 0
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) + ∑ 1 ≤ l ≤ k ≤ n + 1 n = n 1 + ⋯ + n k , n l ≥ 0 , n i ≥ 1 , i ≠ l ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) [ ( m n 1 { 1 , 2 } ∘ m n 1 { 0 , 1 } ) x p 0 , … , x p 1 ⊗ ⋯ ⊗ ( m n l − 1 { 1 , 2 } ∘ m n l − 1 { 0 , 1 } ) x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = 0 , (214) \begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk}m^{\{0,1,2\};x_0,x_1,\ldots,x_i,x_{i+j},x_{i+j+1},\ldots,x_n}_{i+k+1}\\
&\qquad\left(\operatorname{id}_{\mathcal{H}^0(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^0(x_{i-1},x_i)}\otimes m^{\{0\};x_i,x_{i+1},\ldots,x_{i+j}}_j\right.\\
&\qquad\qquad\left.\otimes\operatorname{id}_{\mathcal{H}^0(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^0(x_{n-1},x_n)}\right)\\
&\quad+\sum_{\substack{1\leq l\leq k\leq n+1\\n=n_1+\cdots+n_k,\ n_l\geq0,\ n_i\geq1,\ i\ne l}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)}\\
&\qquad m_k^{\{2\};f^{\{0,1,2\}}(x_{p_0}),f^{\{0,1,2\}}(x_{p_1}),\ldots,f^{\{0,1,2\}}(x_{p_{l-1}}),f^{02}(x_{p_l}),\ldots,f^{02}(x_{p_k})}\\
&\qquad\left[\left(m_{n_1}^{\{1,2\}}\circ m_{n_1}^{\{0,1\}}\right)_{x_{p_0},\ldots,x_{p_1}}\otimes\cdots\otimes\left(m_{n_{l-1}}^{\{1,2\}}\circ m_{n_{l-1}}^{\{0,1\}}\right)_{x_{p_{l-2}},\ldots,x_{p_{l-1}}}\right.\\
&\qquad\qquad\left.\otimes m_{n_l}^{\{0,1,2\};x_{p_{l-1}},\ldots,x_{p_l}}\otimes m_{n_{l+1}}^{\{0,2\};x_{p_l},\ldots,x_{p_{l+1}}}\otimes\cdots\otimes m_{n_k}^{\{0,2\};x_{p_{k-1}},\ldots,x_{p_k}}\right]=0, \tag*{(214)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) + 1 ≤ l ≤ k ≤ n + 1 n = n 1 + ⋯ + n k , n l ≥ 0 , n i ≥ 1 , i = l ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) [ ( m n 1 { 1 , 2 } ∘ m n 1 { 0 , 1 } ) x p 0 , … , x p 1 ⊗ ⋯ ⊗ ( m n l − 1 { 1 , 2 } ∘ m n l − 1 { 0 , 1 } ) x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = 0 , (214)
This is an analogue of (6.24) and (6.25) which we have simplified by writing in terms of the composition m { 1 , 2 } ∘ m { 0 , 1 } m^{\{1,2\}}\circ m^{\{0,1\}} m { 1 , 2 } ∘ m { 0 , 1 } of the data m { 0 , 1 } m^{\{0,1\}} m { 0 , 1 } , m { 1 , 2 } m^{\{1,2\}} m { 1 , 2 } , corresponding to working with the composition of A ∞ A_\infty A ∞ -functors F 12 ∘ F 01 \mathcal{F}^{12}\circ\mathcal{F}^{01} F 12 ∘ F 01 . Comparing (4.15) and (7.22) and using (7.19), we see that m { 0 , 1 , 2 } m^{\{0,1,2\}} m { 0 , 1 , 2 } is equivalent to a unital A ∞ A_\infty A ∞ -natural transformation F 12 ∘ F 01 ⇒ F 02 \mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} F 12 ∘ F 01 ⇒ F 02 , in the sense of Definition 4.22.
Here is the analogue of Definition 6.21.
Definition 7.12. Define a simplicial subset A ∞ C a t R u n , ∞ ⊂ A ∞ C a t ^ R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R\subset\widehat{A_\infty\mathrm{Cat}}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ , by an N N N -simplex ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H I ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^I)_{i=0}^N,(m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H I ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ∞ C a t ^ R u n , ∞ \widehat{A_\infty\mathrm{Cat}}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ lies in A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ if and only if whenever 0 ≤ i 0 < i 1 < i 2 ≤ N 0\leq i_0<i_1<i_2\leq N 0 ≤ i 0 < i 1 < i 2 ≤ N , any (hence all) of the following four equivalent conditions hold:
(i) For all objects x 0 ∈ O i 0 x_0\in\mathcal{O}^{i_0} x 0 ∈ O i 0 , the following morphism is an isomorphism in the cohomology category H 0 ( O i 2 ) H^0(\mathcal{O}^{i_2}) H 0 ( O i 2 ) :
[ m 0 { i 0 , i 1 , i 2 } ( 1 ) ] : f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) ⟶ f i 2 i 0 ( x 0 ) . [m^{\{i_0,i_1,i_2\}}_0(1)] : f^{i_1i_2}\circ f^{i_0i_1}(x_0)\longrightarrow f^{i_2i_0}(x_0). [ m 0 { i 0 , i 1 , i 2 } ( 1 )] : f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) ⟶ f i 2 i 0 ( x 0 ) . (ii) [ m 0 { i 0 , i 1 , i 2 } ( 1 ) ] : H 0 ( F { i 1 , i 2 } ) ∘ H 0 ( F { i 0 , i 1 } ) ⇒ H 0 ( F { i 0 , i 2 } ) [m^{\{i_0,i_1,i_2\}}_0(1)] : H^0(\mathcal{F}^{\{i_1,i_2\}})\circ H^0(\mathcal{F}^{\{i_0,i_1\}})\Rightarrow H^0(\mathcal{F}^{\{i_0,i_2\}}) [ m 0 { i 0 , i 1 , i 2 } ( 1 )] : H 0 ( F { i 1 , i 2 } ) ∘ H 0 ( F { i 0 , i 1 } ) ⇒ H 0 ( F { i 0 , i 2 } ) is a natural isomorphism of functors H 0 ( O i 0 ) → H 0 ( O i 2 ) H^0(\mathcal{O}^{i_0})\to H^0(\mathcal{O}^{i_2}) H 0 ( O i 0 ) → H 0 ( O i 2 ) .
(iii) The unital A ∞ A_\infty A ∞ -natural transformation H i 0 i 1 i 2 : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 \mathcal{H}^{i_0i_1i_2}:\mathcal{F}^{i_1i_2}\circ\mathcal{F}^{i_0i_1}\Rightarrow\mathcal{F}^{i_0i_2} H i 0 i 1 i 2 : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 in Example 7.11(c) is a unital A ∞ A_\infty A ∞ -natural isomorphism, in the sense of Definition 4.22.
(iv) [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 [m^{\{i_0,i_1,i_2\}}] : \mathcal{F}^{i_1i_2}\circ\mathcal{F}^{i_0i_1}\to\mathcal{F}^{i_0i_2} [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 is an isomorphism in the cohomology category H 0 ( Fun A ∞ u n ( O i 0 , O i 2 ) ) H^0(\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{O}^{i_0},\mathcal{O}^{i_2})) H 0 ( Fun A ∞ un ( O i 0 , O i 2 )) .
Here (i) and (ii) are equivalent by definition of natural isomorphism of ordinary functors, and (iii) and (iv) are equivalent by definition of unital A ∞ A_\infty A ∞ -natural isomorphisms in Definition 4.22, and (ii) and (iii) are equivalent by Proposition 4.24, so (i)–(iv) are equivalent. Part (i) is the analogue of the condition that [ m 0 { i 0 , i 1 , i 2 } ( 1 ) ] [m^{\{i_0,i_1,i_2\}}_0(1)] [ m 0 { i 0 , i 1 , i 2 } ( 1 )] is invertible for A ∞ A_\infty A ∞ -algebras in Definition 6.21. If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ then A ∞ C a t ^ R u n , ∞ ( α ) \widehat{A_\infty\mathrm{Cat}}^{\mathrm{un},\infty}_R(\alpha) A ∞ Cat R un , ∞ ( α ) in Definition 7.7(b) maps ( A ∞ C a t R u n , ∞ ) N → ( A ∞ C a t R u n , ∞ ) N ‾ (A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R)_N\to(A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R)_{\overline{N}} ( A ∞ Cat R un , ∞ ) N → ( A ∞ Cat R un , ∞ ) N , so A ∞ C a t R u n , ∞ A_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is a simplicial set.
The following analogue of Theorem 6.22 , one of our main results, is proved in §10.4. The proof is long and difficult.
Theorem 7.13. In Definition 7.12 , A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is an ∞ \infty ∞ -category.
Remark 7.14. In the generalization to filtered A ∞ A_{\infty} A ∞ -categories in Chapter 8, the analogues of Definition 7.12 (i),(ii) do not make sense as we can have ( m 1 { i 2 } ) 2 ≠ 0 (m_{1}^{\{i_{2}\}})^{2}\ne0 ( m 1 { i 2 } ) 2 = 0 , so the cohomology category H 0 ( A i 2 ) H^{0}(\mathcal{A}^{i_{2}}) H 0 ( A i 2 ) is not defined. However, the analogues of Definition 7.12 (iii),(iv) do make sense. We have written the proof of Theorem 7.13 using Definition 7.12 (iv), so that it extends to the filtered case.
Definition 7.15. Let ( ( O i ) , ( f i j ) , ( H i ) , ( m I ) ) ((\mathcal{O}^{i}), (f^{ij}), (\mathcal{H}^{i}), (m^{I})) (( O i ) , ( f ij ) , ( H i ) , ( m I )) be an N N N -simplex in A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ as in Definitions 7.7 and 7.12 , for N ≥ 2 N\ge2 N ≥ 2 . If N > 2 N>2 N > 2 , we call ( ( O i ) , … , ( m I ) ) ((\mathcal{O}^{i}),\ldots,(m^{I})) (( O i ) , … , ( m I )) special if m { 0 , … , N } = 0 m^{\{0,\ldots,N\}}=0 m { 0 , … , N } = 0 . If N = 2 N=2 N = 2 we call ( ( O i ) , … , ( m I ) ) ((\mathcal{O}^{i}),\ldots,(m^{I})) (( O i ) , … , ( m I )) special if in the notation of Example 7.11 (c), we have F 02 = F 21 ∘ F 01 \mathcal{F}^{02}=\mathcal{F}^{21}\circ\mathcal{F}^{01} F 02 = F 21 ∘ F 01 and H 012 = id F 02 \mathcal{H}^{012}=\operatorname{id}_{\mathcal{F}^{02}} H 012 = id F 02 .
In the proof of Theorem 7.13 in §10.4, to prove A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is an ∞ \infty ∞ -category, as in Definition 2.9 , for every 0 < i < N 0<i<N 0 < i < N and simplicial set map f : Λ i N → A ∞ C a t R u n , ∞ f:\Lambda^{N}_{i}\to\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} f : Λ i N → A ∞ Cat R un , ∞ , we show that f f f can be extended to an N N N -simplex f ~ : Δ N → A ∞ C a t R u n , ∞ \widetilde{f}:\Delta^{N}\to\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} f : Δ N → A ∞ Cat R un , ∞ as in (2.5). The proof also yields:
Proposition 7.16. Let f : Λ i N → A ∞ C a t R u n , ∞ f:\Lambda^{N}_{i}\to\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} f : Λ i N → A ∞ Cat R un , ∞ be a map of simplicial sets 0 < i < N 0<i<N 0 < i < N . Then we can extend f f f to a unique map f ~ : Δ N → A ∞ C a t R u n , ∞ \widetilde{f}:\Delta^{N}\to\mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} f : Δ N → A ∞ Cat R un , ∞ as in (2.5), such that f ~ \widetilde{f} f is a special N N N -simplex in the sense of Definition 7.15 .
In § 2.8 , following Lurie [62, §4], we defined ∞ \infty ∞ -bicategories, a kind of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category. The next theorem will be proved in §10.5.
Theorem 7.17. In the simplicial set A ^ ∞ C a t R u n , ∞ \widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ in Definition 7.7, define a 2-simplex to be thin if the unital A ∞ A_{\infty} A ∞ -natural transformation H 012 : F 21 ∘ F 01 ⇒ F 02 \mathcal{H}^{012}:\mathcal{F}^{21}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} H 012 : F 21 ∘ F 01 ⇒ F 02 in Example 7.11 (c) is an A ∞ A_{\infty} A ∞ -natural isomorphism, that is, if it is a 2-simplex in A ∞ C a t R u n , ∞ ⊂ A ^ ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\subset\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ . Write T ⊂ A ^ ∞ C a t R u n , ∞ ( [ 2 ] ) \mathcal{T}\subset\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}([2]) T ⊂ A ∞ Cat R un , ∞ ([ 2 ]) for the set of thin 2-simplices. Then ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is an ∞ \infty ∞ -bicategory in the sense of § 2.8 . We have A ∞ C a t R u n , ∞ = Core ( A ^ ∞ C a t R u n , ∞ , T ) \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}=\operatorname{Core}(\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) A ∞ Cat R un , ∞ = Core ( A ∞ Cat R un , ∞ , T ) , as in Definition 2.41 .
Remark 7.18. (a) Theorem 7.13 follows trivially from Theorem 7.17, as the core of an ∞ \infty ∞ -bicategory is an ∞ \infty ∞ -category. But we use the proof of Theorem 7.13 in that of Theorem 7.17, so we prove Theorem 7.13 first.
(b) Other than proving that ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is an ∞ \infty ∞ -bicategory, we will not take these ideas much further in this book; our focus will be on ∞ \infty ∞ -categories, not ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -categories. But maybe someone will find applications for this ∞ \infty ∞ -bicategory in future. At the least, the author thinks ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is interesting as an example of an ∞ \infty ∞ -bicategory which appears ‘in nature’.
(c) The most basic difference between A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ and ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is that A ∞ C a t R u n , ∞ \mathcal{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is an ∞ \infty ∞ -category ( ( ∞ , 1 ) ((\infty,1) (( ∞ , 1 ) -category) in which 2-morphisms are A ∞ A_{\infty} A ∞ -natural isomorphisms, and ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is an ∞ \infty ∞ -bicategory ( ( ∞ , 2 ) ((\infty,2) (( ∞ , 2 ) -category) in which 2-morphisms are unital A ∞ A_{\infty} A ∞ -natural transformations.
(d) In §4.11 we discussed Faonte’s proposal [24] for defining an ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category of unital A ∞ A_\infty A ∞ -categories over R R R , as a fibrant Segal category in the model category ( S S e t , J o y ) (\mathrm{SSet},\mathrm{Joy}) ( SSet , Joy ) , at least when R R R is a field of characteristic zero. Although we will not prove this, we expect that under the equivalences between different models of ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category in Gagna–Harpaz–Lanari [34], our ∞ \infty ∞ -bicategory ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R,\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) should be equivalent to Faonte’s ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category. As evidence for this, Theorem 9.1 (a) below shows left mapping spaces in ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R,\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) are given by
Hom A ^ ∞ C a t R u n , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ u n ( A , B ) ) . \operatorname{Hom}^{L}_{\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R}(\mathcal{A},\mathcal{B}) \cong\mathcal{N}_{A_\infty}\bigl(\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A},\mathcal{B})\bigr). Hom A ∞ Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ un ( A , B ) ) .
where the right hand side is the mapping space in Faonte’s ( ∞ , 2 ) (\infty,2) ( ∞ , 2 ) -category.
(e) We will use Theorem 7.17 in the proof of Theorem 9.12 in §11.5.
(f) In Definition 2.43 we defined strong ∞ \infty ∞ -bicategories , which are just called ∞ \infty ∞ -bicategories in Lurie [63, §5.4]. The author expects that ( A ^ ∞ C a t R u n , ∞ , T ) (\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R,\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is not a strong ∞ \infty ∞ -bicategory. That is, there may be 2-simplices t : Δ 2 → A ^ ∞ C a t R u n , ∞ t:\Delta^2\to\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R t : Δ 2 → A ∞ Cat R un , ∞ which satisfy Definition 2.40 (a), but for which the unital A ∞ A_\infty A ∞ -natural transformation H 012 : F 12 ∘ F 01 ⇒ F 02 \mathcal{H}^{012}:\mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} H 012 : F 12 ∘ F 01 ⇒ F 02 in Example 7.11 (c) is not an A ∞ A_\infty A ∞ -natural isomorphism. Using Definition 2.40 (a), it is not difficult to show that H 012 \mathcal{H}^{012} H 012 has a left inverse in the cohomology category H 0 ( Fun A ∞ u n ( A 0 , A 2 ) ) H^0(\operatorname{Fun}^{\mathrm{un}}_{A_\infty}(\mathcal{A}^0,\mathcal{A}^2)) H 0 ( Fun A ∞ un ( A 0 , A 2 )) , but it does not seem possible to show that H 012 \mathcal{H}^{012} H 012 has a right inverse unless F 01 \mathcal{F}^{01} F 01 or F 12 \mathcal{F}^{12} F 12 is a homotopy equivalence. This is our reason for choosing the ∞ \infty ∞ -bicategory theory from [62, §4] rather than from [63, §5.4].
Here is the analogue of Definition 6.25.
Definition 7.19. Let K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Define simplicial subsets
A ∞ C a t R , K u n , ∞ ⊂ A ∞ C a t R u n , ∞ ⊂ A ^ ∞ C a t R u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K}\subset\mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R\subset\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R , K un , ∞ ⊂ A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ such that an N N N -simplex ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H I ) ∅ = I ⊆ { 0 , … , N } ) in A ^ ∞ C a t R u n , ∞ \widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ in Definition 7.7(a) lies in A ∞ C a t R , K u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ if f i k = f j k ∘ f i j f^{ik}=f^{jk}\circ f^{ij} f ik = f j k ∘ f ij for all 0 ≤ i ≤ j ≤ k ≤ N 0\leq i\leq j\leq k\leq N 0 ≤ i ≤ j ≤ k ≤ N , and for all 0 ≤ i 0 < i 1 < i 2 ≤ N 0\leq i_0<i_1<i_2\leq N 0 ≤ i 0 < i 1 < i 2 ≤ N and x 0 ∈ O i 0 x_0\in\mathcal{O}^{i_0} x 0 ∈ O i 0 we have m 0 { i 0 , i 1 , i 2 } ; x 0 : λ ↦ λ 1 i 0 i 2 ( x 0 ) \mathfrak{m}^{\{i_0,i_1,i_2\};x_0}_0:\lambda\mapsto\lambda^{i_0i_2}_1(x_0) m 0 { i 0 , i 1 , i 2 } ; x 0 : λ ↦ λ 1 i 0 i 2 ( x 0 ) , and m 0 I ; x 0 = 0 \mathfrak{m}^{I;x_0}_0=0 m 0 I ; x 0 = 0 for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\neq I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } with ∣ I ∣ ≤ K |I|\leq K ∣ I ∣ ≤ K and ∣ I ∣ ≠ 3 |I|\neq3 ∣ I ∣ = 3 and x 0 ∈ O min I x_0\in\mathcal{O}^{\min I} x 0 ∈ O m i n I .
These conditions are preserved by the maps in Definition 7.7(b), and so define a simplicial subset A ∞ C a t R , K u n , ∞ ⊂ A ^ ∞ C a t R u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K}\subset\widehat{\mathcal{A}}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R , K un , ∞ ⊂ A ∞ Cat R un , ∞ . As [ m 1 { i 0 , i 1 , i 2 } ; x 0 ( 1 ) ] = [ f 1 i 0 i 2 ( x 0 ) ] [\mathfrak{m}^{\{i_0,i_1,i_2\};x_0}_1(1)]=[f^{i_0i_2}_1(x_0)] [ m 1 { i 0 , i 1 , i 2 } ; x 0 ( 1 )] = [ f 1 i 0 i 2 ( x 0 )] is invertible in H 0 ( A i 2 ) H^0(\mathcal{A}^{i_2}) H 0 ( A i 2 ) for all x 0 ∈ O i 0 x_0\in\mathcal{O}^{i_0} x 0 ∈ O i 0 we see that A ∞ C a t R , K u n , ∞ ⊂ A ∞ C a t R u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K}\subset\mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R , K un , ∞ ⊂ A ∞ Cat R un , ∞ by Definition 7.12 (i). Clearly A ∞ C a t R , K u n , ∞ ⊂ A ∞ C a t R , L u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K}\subset\mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,L} A ∞ Cat R , K un , ∞ ⊂ A ∞ Cat R , L un , ∞ if L ≤ K L\leq K L ≤ K in { 3 , 4 , 5 , … , ∞ } \{3,4,5,\ldots,\infty\} { 3 , 4 , 5 , … , ∞ } , giving the inclusion relations in the first column of (7.1). Theorem 7.29 will show that A ∞ C a t R , K u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ is an ∞ \infty ∞ -category. We say that A ∞ C a t R , K u n , ∞ \mathcal{A}_\infty\mathrm{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ has type K K K .
need this as m 0 { i 0 , i 1 , i 2 } ; x 0 m^{\{i_0,i_1,i_2\};x_0}_0 m 0 { i 0 , i 1 , i 2 } ; x 0 maps R → H i 2 ( f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) , f i 0 i 2 ( x 0 ) ) R \to\mathcal{H}^{i_2}(f^{i_1i_2}\circ f^{i_0i_1}(x_0),f^{i_0i_2}(x_0)) R → H i 2 ( f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) , f i 0 i 2 ( x 0 )) , so defining m 0 { i 0 , i 1 , i 2 } ; x 0 : λ ↦ λ 1 f i 0 i 2 ( x 0 ) m^{\{i_0,i_1,i_2\};x_0}_0:\lambda\mapsto\lambda1_{f^{i_0i_2}(x_0)} m 0 { i 0 , i 1 , i 2 } ; x 0 : λ ↦ λ 1 f i 0 i 2 ( x 0 ) only makes sense if f i 0 i 2 ( x 0 ) = f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) f^{i_0i_2}(x_0)=f^{i_1i_2}\circ f^{i_0i_1}(x_0) f i 0 i 2 ( x 0 ) = f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) .
This condition is trivial for A ∞ A_\infty A ∞ -algebras, when we take f i j = id { ∗ } f^{ij}=\operatorname{id}_{\{\ast\}} f ij = id { ∗ } as in Definition 7.21, but for A ∞ A_\infty A ∞ -categories, it is very restrictive . It means that we do not have an interesting notion of natural isomorphism of functors in A ∞ Cat R , K u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ for K ≥ 3 K\geq3 K ≥ 3 . In Example 7.11(c), a 2-simplex in A ∞ Cat R , K u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ∞ Cat R , K un , ∞ defines unital A ∞ A_\infty A ∞ -categories A 0 \mathcal{A}^0 A 0 , A 1 \mathcal{A}^1 A 1 , A 2 \mathcal{A}^2 A 2 over R R R as in (a), and unital A ∞ A_\infty A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^i\to\mathcal{A}^j F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) , such that the cohomology functors H 0 ( F 12 ∘ F 01 ) , H 0 ( F 02 ) : H 0 ( A 0 ) → H 0 ( A 2 ) H^0(\mathcal{F}^{12}\circ\mathcal{F}^{01}),H^0(\mathcal{F}^{02}):H^0(\mathcal{A}^0)\to H^0(\mathcal{A}^2) H 0 ( F 12 ∘ F 01 ) , H 0 ( F 02 ) : H 0 ( A 0 ) → H 0 ( A 2 ) are not just naturally isomorphic, as in Definition 7.12(ii), but actually equal.
(b) In [16], with Guillem Cazassus and Alexander Ritter, the author is planning an application in Symplectic Geometry of A ∞ Cat R , ∞ u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,\infty} A ∞ Cat R , ∞ un , ∞ in §7.3 below, which is a non-unital version of A ∞ Cat R , ∞ u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,\infty} A ∞ Cat R , ∞ un , ∞ . The aim is to give a new construction of the Fukaya A ∞ A_\infty A ∞ -category of (say exact) Lagrangians L L L in a symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) . To do this, we must make some arbitrary choices, such as an almost complex structure J J J on X X X compatible with ω \omega ω . These arbitrary choices are organized into a contractible Kan complex K \mathcal{K} K , and our definition of Fukaya category yields a map of simplicial sets K → A ∞ Cat R , ∞ ∞ \mathcal{K}\to A_\infty\operatorname{Cat}^{\infty}_{R,\infty} K → A ∞ Cat R , ∞ ∞ .
The point we want to make here is that though different choices κ \kappa κ in K \mathcal{K} K yield different A ∞ A_\infty A ∞ -categories A κ \mathcal{A}_\kappa A κ , the set of objects in A κ \mathcal{A}_\kappa A κ , which is a set O \mathcal{O} O of suitable Lagrangians L L L in ( X , ω ) (X,\omega) ( X , ω ) , is independent of κ \kappa κ . So it is natural in this case to require A ∞ A_\infty A ∞ -functors A κ 1 → A κ 2 \mathcal{A}_{\kappa_1}\to\mathcal{A}_{\kappa_2} A κ 1 → A κ 2 to induce the same map id O \operatorname{id}_{\mathcal{O}} id O on objects.
(c) The condition f i k = f j k ∘ f i j f^{ik}=f^{jk}\circ f^{ij} f ik = f j k ∘ f ij simplifies parts of Definition 7.7: the maps f I : O i 0 → O i M f^I:\mathcal{O}^{i_0}\to\mathcal{O}^{i_M} f I : O i 0 → O i M in Definition 7.7(a)(ii) satisfy f I = f i 0 i M f^I=f^{i_0i_M} f I = f i 0 i M .
Definition 7.21. For A ∞ Alg R u n , ∞ A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ as in §6.3, define an ∞ \infty ∞ -functor I A ∞ Cat A ∞ Alg u n : A ∞ Alg R u n , ∞ → A ∞ Cat R u n , ∞ I^{A_\infty\operatorname{Cat}^{\mathrm{un}}_{A_\infty\operatorname{Alg}}}:A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_R\to A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R I A ∞ Cat A ∞ Alg un : A ∞ Alg R un , ∞ → A ∞ Cat R un , ∞ to act on N N N -simplices by
I A ∞ Cat A ∞ Alg u n ( [ N ] ) : ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) ⟼ ( ( { ∗ } ) i = 0 N , ( id { ∗ } ) 0 ≤ i ≤ j ≤ N , ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) . I^{A_\infty\operatorname{Cat}^{\mathrm{un}}_{A_\infty\operatorname{Alg}}}([N]):\bigl((A^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subset\{0,\ldots,N\}}\bigr)
\longmapsto\bigl((\{\ast\})_{i=0}^N,(\operatorname{id}_{\{\ast\}})_{0\leq i\leq j\leq N},(A^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subset\{0,\ldots,N\}}\bigr). I A ∞ Cat A ∞ Alg un ([ N ]) : ( ( A i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) ⟼ ( ({ ∗ } ) i = 0 N , ( id { ∗ } ) 0 ≤ i ≤ j ≤ N , ( A i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) .
That is, we take O i \mathcal{O}^i O i to be the one point set { ∗ } \{\ast\} { ∗ } , and f i j = id { ∗ } f^{ij}=\operatorname{id}_{\{\ast\}} f ij = id { ∗ } , and H i ( ∗ , ∗ ) = A i \mathcal{H}^i(\ast,\ast)=A^i H i ( ∗ , ∗ ) = A i , and m k I ; ∗ , … , ∗ = m k I m^{I;\ast,\ldots,\ast}_k=m^I_k m k I ; ∗ , … , ∗ = m k I . This just turns A ∞ A_\infty A ∞ -algebras into A ∞ A_\infty A ∞ -categories with one object ∗ \ast ∗ . As the definition of A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ extends that of A ∞ Alg R u n , ∞ A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ from one object to many, this is a strict isomorphism with the full ∞ \infty ∞ -subcategory of A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ with objects unital A ∞ A_\infty A ∞ -categories A \mathcal{A} A with one object ∗ \ast ∗ .
This restricts to an ∞ \infty ∞ -functor I A ∞ Cat A ∞ Alg u n : A ∞ Alg R , K u n , ∞ → A ∞ Cat R , K u n , ∞ I^{A_\infty\operatorname{Cat}^{\mathrm{un}}_{A_\infty\operatorname{Alg}}}:A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_{R,K}\to A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} I A ∞ Cat A ∞ Alg un : A ∞ Alg R , K un , ∞ → A ∞ Cat R , K un , ∞ for all K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } .
Results for A ∞ Alg R u n , ∞ A_\infty\operatorname{Alg}^{\mathrm{un},\infty}_R A ∞ Alg R un , ∞ such as Proposition 6.18 and Theorems 6.19 and 6.22 can be deduced from their analogues for A ∞ Cat R u n , ∞ A_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ by restricting to the image of I A ∞ Cat A ∞ Alg u n I^{A_\infty\operatorname{Cat}^{\mathrm{un}}_{A_\infty\operatorname{Alg}}} I A ∞ Cat A ∞ Alg un , and changing notation appropriately.
Other ∞ \infty ∞ -categories of A ∞ A_\infty A ∞ -categories
This section is the analogue of §6.4. We first define ∞ \infty ∞ -categories A ˙ ∞ Cat R , K u n , ∞ \dot{A}_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ isomorphic to A ∞ Cat R , K u n , ∞ A_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ∞ Cat R , K un , ∞ for K = 3 , 4 , … , ∞ K=3,4,\ldots,\infty K = 3 , 4 , … , ∞ , extending Definition 6.27.
Let ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) \left((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subset\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) be an N N N -simplex in A ∞ Cat R , K u n , ∞ A_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ∞ Cat R , K un , ∞ in Definition 7.19. This will correspond to an N N N -simplex ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ ≠ I ⊂ { 0 , … , N } ) \left((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(\dot{m}^I)_{\varnothing\ne I\subset\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ = I ⊂ { 0 , … , N } ) in A ˙ ∞ Cat R , K u n , ∞ \dot{A}_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ , where we define
m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { m 0 { i 0 , i 1 , i 2 } ; x 0 − 1 f i 0 i 2 ( x 0 ) i 2 , k = 0 and M = 2 , m k { i 0 , … , i M } ; x 0 , … , x k , otherwise . (215) \dot{m}^{\{i_0,\ldots,i_M\};x_0,\ldots,x_k}_k =
\begin{cases}
m^{\{i_0,i_1,i_2\};x_0}_0-1^{i_2}_{f^{i_0i_2}(x_0)}, & k=0\text{ and }M=2,\\
m^{\{i_0,\ldots,i_M\};x_0,\ldots,x_k}_k, & \text{otherwise}.
\end{cases} \tag*{(215)} m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { m 0 { i 0 , i 1 , i 2 } ; x 0 − 1 f i 0 i 2 ( x 0 ) i 2 , m k { i 0 , … , i M } ; x 0 , … , x k , k = 0 and M = 2 , otherwise . (215)
extending (6.27). This only makes sense if f i 0 i 2 ( x 0 ) = f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) f^{i_0i_2}(x_0)=f^{i_1i_2}\circ f^{i_0i_1}(x_0) f i 0 i 2 ( x 0 ) = f i 1 i 2 ∘ f i 0 i 1 ( x 0 ) . So we only do it for the A ∞ Cat R , K u n , ∞ A_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ∞ Cat R , K un , ∞ for K = 3 , 4 , … , ∞ K=3,4,\ldots,\infty K = 3 , 4 , … , ∞ , which include the condition f i k = f j k ∘ f i j f^{ik}=f^{jk}\circ f^{ij} f ik = f j k ∘ f ij for all $0\leq i\leq j\leq k\leq N as in Definitions 7.19 and Remark 7.20. The construction of this section does not work for A ∞ Cat R u n , ∞ = A ∞ Cat R , 2 u n , ∞ A_\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty}=A_\infty\operatorname{Cat}_{R,2}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ = A ∞ Cat R , 2 un , ∞ in Definitions 7.7 and 7.12, which allow f i k ≠ f j k ∘ f i j f^{ik}\ne f^{jk}\circ f^{ij} f ik = f j k ∘ f ij .
Definition 7.22. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . We will define a simplicial set A ˙ ∞ Cat R , K u n , ∞ \dot{A}_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ of unital A ∞ A_\infty A ∞ -categories over R R R .
(a) For N ≥ 0 N\geq0 N ≥ 0 , define the N N N -simplices ( A ˙ ∞ Cat R , K u n , ∞ ) N (\dot{A}_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty})_N ( A ˙ ∞ Cat R , K un , ∞ ) N in A ˙ ∞ Cat R , K u n , ∞ \dot{A}_\infty\operatorname{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ to be the set of quadruples ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) \left((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subset\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) , where:
(i) O i \mathcal{O}^i O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , thought of as the set of objects Obj A i \operatorname{Obj}\mathcal{A}^i Obj A i of a (unital) A ∞ A_\infty A ∞ -category A i \mathcal{A}^i A i at vertex i i i of the N N N -simplex Δ N \Delta^N Δ N .
(ii) f i j : O i → O j f^{ij}:\mathcal{O}^i\to\mathcal{O}^j f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , thought of as the map F ~ O b j i j : Obj A i → Obj A j \widetilde{F}^{ij}_{\mathrm{Obj}}:\operatorname{Obj}\mathcal{A}^i\to\operatorname{Obj}\mathcal{A}^j F Obj ij : Obj A i → Obj A j in a (unital) A ∞ A_\infty A ∞ -functor F ~ i j : A i → A j \widetilde{F}^{ij}:\mathcal{A}^i\to\mathcal{A}^j F ij : A i → A j living on the edge [ i , j ] [i,j] [ i , j ] of Δ N \Delta^N Δ N . We require these maps to satisfy f i i = id O i f^{ii}=\operatorname{id}_{\mathcal{O}^i} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N and f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik for 0 ≤ i ≤ j ≤ k ≤ N 0\leq i\leq j\leq k\leq N 0 ≤ i ≤ j ≤ k ≤ N .
(iii) For all i = 0 , 1 , … , N i=0,1,\ldots,N i = 0 , 1 , … , N and x , y ∈ O i x,y\in\mathcal{O}^i x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded R R R -module H i ( x , y ) = ⨁ d ∈ Z H d i ( x , y ) \mathcal{H}^i(x,y)=\bigoplus_{d\in\mathbb{Z}}\mathcal{H}^i_d(x,y) H i ( x , y ) = ⨁ d ∈ Z H d i ( x , y ) , thought of as the Hom space Hom A i ( x , y ) \operatorname{Hom}_{\mathcal{A}^i}(x,y) Hom A i ( x , y ) in the A ∞ A_\infty A ∞ -category A i \mathcal{A}^i A i . We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^i=(\mathcal{H}^i(x,y))_{x,y\in\mathcal{O}^i} H i = ( H i ( x , y ) ) x , y ∈ O i .
(iv) For each nonempty subset I ⊂ { 0 , … , N } I\subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M with M ≥ 0 M\geq0 M ≥ 0 , and for all k ≥ 0 k\geq0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_0,\ldots,x_k\in\mathcal{O}^{i_0} x 0 , … , x k ∈ O i 0 , thought of as objects in A i 0 \mathcal{A}^{i_0} A i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k ) ) (216) m^{I;x_0,\ldots,x_k}_k:\mathcal{H}^{i_0}(x_0,x_1)\otimes_R\mathcal{H}^{i_0}(x_1,x_2)\otimes_R\cdots\otimes_R\mathcal{H}^{i_0}(x_{k-1},x_k)\longrightarrow\mathcal{H}^{i_M}\left(f^{i_0i_M}(x_0),f^{i_0i_M}(x_k)\right) \tag*{(216)} m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k ) ) (216)
graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k . Here when k = 0 k=0 k = 0 , by convention the empty tensor product on the top line of (7.24) is R R R , so m 0 I ; x 0 m^{I;x_0}_0 m 0 I ; x 0 maps
m 0 I ; x 0 : R ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M . m^{I;x_0}_0:R\longrightarrow\mathcal{H}^{i_M}\left(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0)\right)^{2-M}. m 0 I ; x 0 : R ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M .
When k = 0 k=0 k = 0 we require that m 0 I ; x 0 = 0 m^{I;x_0}_0=0 m 0 I ; x 0 = 0 for all ∅ ≠ I ⊂ { 0 , … , N } \varnothing\ne I\subset\{0,\ldots,N\} ∅ = I ⊂ { 0 , … , N } with ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K and x 0 ∈ O min I x_0\in\mathcal O^{\min I} x 0 ∈ O m i n I .
We write m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 m^I_k=(m^{I;x_0,\ldots,x_k}_k)_{x_0,\ldots,x_k\in\mathcal O^{i_0}} m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 and m I = ( m k I ) k ≥ 0 m^I=(m^I_k)_{k\ge0} m I = ( m k I ) k ≥ 0 .
We require these maps m k I ; x 0 , … , x k m^{I;x_0,\ldots,x_k}_k m k I ; x 0 , … , x k to satisfy the following identity generalizing (6.28) for all I = { i 0 < ⋯ < i M } ⊂ { 0 , … , N } I=\{i_0<\cdots<i_M\}\subset\{0,\ldots,N\} I = { i 0 < ⋯ < i M } ⊂ { 0 , … , N } , n ≥ 0 n\ge0 n ≥ 0 and objects x 0 , … , x n ∈ O i 0 x_0,\ldots,x_n\in\mathcal O^{i_0} x 0 , … , x n ∈ O i 0
∑ 0 ≤ j ≤ M , 0 ≤ k ≤ max ( n , n + j − 1 ) y 0 , … , y k ∈ O i j ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = ∑ j = 1 , … , M − 1 ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n , (217) \sum_{\substack{0\le j\le M,\ 0\le k\le\max(n,n+j-1)\\ y_0,\ldots,y_k\in\mathcal O^{i_j}}}(-1)^j m^{\{i_j,\ldots,i_M\};y_0,\ldots,y_k}_k\circ d^{\{i_0,\ldots,i_j\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{j=1,\ldots,M-1}(-1)^j m^{I\setminus\{i_j\};x_0,\ldots,x_n}_n, \tag*{(217)} 0 ≤ j ≤ M , 0 ≤ k ≤ m a x ( n , n + j − 1 ) y 0 , … , y k ∈ O i j ∑ ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = j = 1 , … , M − 1 ∑ ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n , (217)
where d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n d^{\{i_0,\ldots,i_j\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n is defined from the m k J ; w 0 , … , w k m^{J;w_0,\ldots,w_k}_k m k J ; w 0 , … , w k as in (7.7)–(7.9). In (7.25) there are only finitely many nonzero terms, which all map to H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x n ) ) \mathcal H^{i_M}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_n)) H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x n )) , so (7.25) is an equation in this space.
(v) Definition 7.7(a)(v) holds.
(b) If α : [ N ‾ ] → [ N ] \alpha:[\overline N]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , that is, α : { 0 , … , N ‾ } → { 0 , … , N } \alpha:\{0,\ldots,\overline N\}\to\{0,\ldots,N\} α : { 0 , … , N } → { 0 , … , N } is nondecreasing, let A ˙ ∞ Cat R , K u n , ∞ ( α ) : ( A ˙ ∞ Cat R , K u n , ∞ ) N → ( A ˙ ∞ Cat R , K u n , ∞ ) N ‾ \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K}(\alpha):(\dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K})^N\to(\dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K})^{\overline N} A ˙ ∞ Cat R , K un , ∞ ( α ) : ( A ˙ ∞ Cat R , K un , ∞ ) N → ( A ˙ ∞ Cat R , K un , ∞ ) N map ( ( O i ) , ( f i j ) , ( H i ) , ( m I ) ) ((\mathcal O^i),(f^{ij}),(\mathcal H^i),(m^I)) (( O i ) , ( f ij ) , ( H i ) , ( m I )) to ( ( O ‾ i ) , ( f ‾ i j ) , ( H ‾ i ) , ( m ‾ I ‾ ) ) ((\overline{\mathcal O}^i),(\overline f^{ij}),(\overline{\mathcal H}^i),(\overline m^{\overline I})) (( O i ) , ( f ij ) , ( H i ) , ( m I )) , where O ‾ i = O α ( i ) \overline{\mathcal O}^i=\mathcal O^{\alpha(i)} O i = O α ( i ) for i = 0 , … , N ‾ i=0,\ldots,\overline N i = 0 , … , N , and f ‾ i j = f α ( i ) α ( j ) \overline f^{ij}=f^{\alpha(i)\alpha(j)} f ij = f α ( i ) α ( j ) for 0 ≤ i ≤ j ≤ N ‾ 0\le i\le j\le\overline N 0 ≤ i ≤ j ≤ N , and H ‾ i = H α ( i ) \overline{\mathcal H}^i=\mathcal H^{\alpha(i)} H i = H α ( i ) for i = 0 , … , N ‾ i=0,\ldots,\overline N i = 0 , … , N . To define m ‾ \overline m m , if I ‾ = { i ‾ 0 < i ‾ 1 < ⋯ < i ‾ M ‾ } ⊂ { 0 , 1 , … , N ‾ } \overline I=\{\overline i_0<\overline i_1<\cdots<\overline i_{\overline M}\}\subset\{0,1,\ldots,\overline N\} I = { i 0 < i 1 < ⋯ < i M } ⊂ { 0 , 1 , … , N } , k ≥ 0 k\ge0 k ≥ 0 and x 0 , … , x k ∈ O ‾ i ‾ 0 = O α ( i ‾ 0 ) x_0,\ldots,x_k\in\overline{\mathcal O}^{\overline i_0}=\mathcal O^{\alpha(\overline i_0)} x 0 , … , x k ∈ O i 0 = O α ( i 0 ) , generalizing (6.22) we define
m ‾ k I ‾ ; x 0 , … , x k = { m k α ( I ‾ ) ; x 0 , … , x k if α ∣ I ‾ is injective , id H i ( x 0 , x 1 ) if k = 1 and I ‾ = { i ‾ 0 , i ‾ 1 } with i ‾ 0 ≠ i ‾ 1 , α ( i ‾ 0 ) = α ( i ‾ 1 ) = i , 0 otherwise. (218) \overline m^{\overline I;x_0,\ldots,x_k}_k=
\begin{cases}
m^{\alpha(\overline I);x_0,\ldots,x_k}_k & \text{if }\alpha|_{\overline I}\text{ is injective},\\
\operatorname{id}_{\mathcal H^i}(x_0,x_1) & \text{if }k=1\text{ and }\overline I=\{\overline i_0,\overline i_1\}\text{ with }\overline i_0\ne\overline i_1,\ \alpha(\overline i_0)=\alpha(\overline i_1)=i,\\
0 & \text{otherwise.}
\end{cases} \tag*{(218)} m k I ; x 0 , … , x k = ⎩ ⎨ ⎧ m k α ( I ) ; x 0 , … , x k id H i ( x 0 , x 1 ) 0 if α ∣ I is injective , if k = 1 and I = { i 0 , i 1 } with i 0 = i 1 , α ( i 0 ) = α ( i 1 ) = i , otherwise. (218)
Note that in contrast to (7.20), there is no special case when k = 0 k=0 k = 0 , ∣ I ‾ ∣ = 3 |\overline I|=3 ∣ I ∣ = 3 and α ∣ I ‾ \alpha|_{\overline I} α ∣ I is not injective, and m ‾ 0 { i ‾ 0 , i ‾ 1 , i ‾ 2 } ; x 0 = 0 \overline m^{\{\overline i_0,\overline i_1,\overline i_2\};x_0}_0=0 m 0 { i 0 , i 1 , i 2 } ; x 0 = 0 in this case.
Proposition 7.23 shows A ˙ ∞ Cat R , K u n , ∞ ( α ) \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K}(\alpha) A ˙ ∞ Cat R , K un , ∞ ( α ) does map to ( A ˙ ∞ Cat R , K u n , ∞ ) N ‾ (\dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K})^{\overline N} ( A ˙ ∞ Cat R , K un , ∞ ) N , and thus A ˙ ∞ Cat R , K u n , ∞ \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Cat R , K un , ∞ is a simplicial set as in Definitions 6.27 and 7.7. Theorem 7.29 will show that A ˙ ∞ Cat R , K u n , ∞ \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Cat R , K un , ∞ is an ∞ \infty ∞ -category. We say that A ˙ ∞ Cat R , K u n , ∞ \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Cat R , K un , ∞ has type K K K . Clearly A ˙ ∞ Cat R , K u n , ∞ ⊂ A ˙ ∞ Cat R , L u n , ∞ \dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K}\subset\dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,L} A ˙ ∞ Cat R , K un , ∞ ⊂ A ˙ ∞ Cat R , L un , ∞ if L ≤ K L\le K L ≤ K in { 3 , 4 , 5 , … , ∞ } \{3,4,5,\ldots,\infty\} { 3 , 4 , 5 , … , ∞ } , giving the inclusion relations in the second column of (7.1).
Here is the analogue of Propositions 6.28 and 7.9, proved in §10.6.
Proposition 7.23. The data ( ( O ‾ i ) i = 0 N ‾ , ( f ‾ i j ) 0 ≤ i ≤ j ≤ N ‾ , ( H ‾ i ) i = 0 N ‾ , ( m ‾ I ‾ ) ∅ ≠ I ‾ ⊂ { 0 , … , N ‾ } ) ((\overline{\mathcal O}^i)_{i=0}^{\overline N},(\overline f^{ij})_{0\le i\le j\le\overline N},(\overline{\mathcal H}^i)_{i=0}^{\overline N},(\overline m^{\overline I})_{\varnothing\ne\overline I\subset\{0,\ldots,\overline N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) in Definition 7.22(b) lies in ( A ˙ ∞ Cat R , K u n , ∞ ) N ‾ (\dot A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K})^{\overline N} ( A ˙ ∞ Cat R , K un , ∞ ) N .
The next theorem, an analogue of Theorems 6.29 and 7.10, will be proved in §10.7 using equation (7.25).
Theorem 7.24. Let ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) be an N N N -simplex in A ˙ ∞ C a t R , K u n , ∞ \dot{\mathop{\mathrm{A}}\nolimits}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ as in Definition 7.22, and define maps d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n from the m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k as in (7.6)–(7.9). Then for each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , written I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M , and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 and y 0 , … , y k ∈ O i M y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}} y 0 , … , y k ∈ O i M , in R R R -linear maps as in (7.6) we have
∑ j = 0 M ∑ 0 ≤ l ≤ max ( n , n + j − 1 ) , w 0 , … , w l ∈ O i j ( − 1 ) j d l , k ; y 0 , … , y k { i j , … , i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 , … , i j } ; x 0 , … , x n = ∑ j = 1 M − 1 ( − 1 ) j d n , k ; y 0 , … , y k I ∖ { i j } ; x 0 , … , x n . (219) \begin{aligned}
\sum_{j=0}^{M}\ \sum_{\substack{0\leq l\leq\max(n,n+j-1),\\ w_{0},\ldots,w_{l}\in\mathcal{O}^{i_{j}}}}(-1)^{j}d^{\{i_{j},\ldots,i_{M}\};w_{0},\ldots,w_{l}}_{l,k;y_{0},\ldots,y_{k}}\circ d^{\{i_{0},\ldots,i_{j}\};x_{0},\ldots,x_{n}}_{n,l;w_{0},\ldots,w_{l}}\\
=\sum_{j=1}^{M-1}(-1)^{j}d^{I\setminus\{i_{j}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} . \tag*{(219)}
\end{aligned} j = 0 ∑ M 0 ≤ l ≤ m a x ( n , n + j − 1 ) , w 0 , … , w l ∈ O i j ∑ ( − 1 ) j d l , k ; y 0 , … , y k { i j , … , i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 , … , i j } ; x 0 , … , x n = j = 1 ∑ M − 1 ( − 1 ) j d n , k ; y 0 , … , y k I ∖ { i j } ; x 0 , … , x n . (219) where there are only finitely many nonzero terms in the sums.
Here is the analogue of Examples 6.30 and 7.11.
Example 7.25. (a) A 0-simplex ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) (\mathcal{O}^{0},f^{00}=\operatorname{id}_{\mathcal{O}^{0}},\mathcal{H}^{0},m^{\{0\}}) ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) of A ˙ ∞ C a t R , K u n , ∞ \dot{\mathop{\mathrm{A}}\nolimits}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ is equivalent to a unital A ∞ A_{\infty} A ∞ -category A 0 \mathcal{A}^{0} A 0 over R R R , with objects Obj A 0 = O 0 \operatorname{Obj}\mathcal{A}^{0}=\mathcal{O}^{0} Obj A 0 = O 0 , Hom spaces Hom A 0 ( x , y ) = H 0 ( x , y ) \operatorname{Hom}_{\mathcal{A}^{0}}(x,y)=\mathcal{H}^{0}(x,y) Hom A 0 ( x , y ) = H 0 ( x , y ) , morphisms ( m k { 0 } ; x 0 , … , x k ) k ≥ 1 (m^{\{0\};x_{0},\ldots,x_{k}}_{k})_{k\geq1} ( m k { 0 } ; x 0 , … , x k ) k ≥ 1 , and units 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 1_{x}^{0}\in\operatorname{Hom}_{\mathcal{A}^{0}}^{0}(x,x)=\mathcal{H}^{0}(x,x)^{0} 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 . (b) A 1-simplex of A ˙ ∞ C a t R , K u n , ∞ \dot{\mathop{\mathrm{A}}\nolimits}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ is equivalent to unital A ∞ A_{\infty} A ∞ -categories A 0 , A 1 \mathcal{A}^{0},\mathcal{A}^{1} A 0 , A 1 over R R R , and a unital A ∞ A_{\infty} A ∞ -functor F 01 : A 0 → A 1 \mathcal{F}^{01}:\mathcal{A}^{0}\to\mathcal{A}^{1} F 01 : A 0 → A 1 , with F O b j 01 = f 01 \mathcal{F}^{01}_{\mathrm{Obj}}=f^{01} F Obj 01 = f 01 and morphisms F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k \mathcal{F}^{01;x_{0},\ldots,x_{k}}_{k}=m^{\{0,1\};x_{0},\ldots,x_{k}}_{k} F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k . (c) A 2-simplex of A ˙ ∞ C a t R , K u n , ∞ \dot{\mathop{\mathrm{A}}\nolimits}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{un},\infty} A ˙ ∞ Cat R , K un , ∞ is equivalent to unital A ∞ A_{\infty} A ∞ -categories A 0 , A 1 , A 2 \mathcal{A}^{0},\mathcal{A}^{1},\mathcal{A}^{2} A 0 , A 1 , A 2 over R R R as in (a), and unital A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), and R R R -linear maps for all k ≥ 1 k\geq1 k ≥ 1 and x 0 , … , x k ∈ O 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{0} x 0 , … , x k ∈ O 0
m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 02 ( x 0 ) , f 02 ( x k ) ) m^{\{0,1,2\};x_{0},\ldots,x_{k}}_{k}:\mathcal{H}^{0}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{0}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{0}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{2}(f^{02}(x_{0}),f^{02}(x_{k})) m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 02 ( x 0 ) , f 02 ( x k ))
graded of degree − k -k − k , with m 0 { 0 , 1 , 2 } ; x 0 = 0 m^{\{0,1,2\};x_{0}}_{0}=0 m 0 { 0 , 1 , 2 } ; x 0 = 0 , satisfying (7.19) and
∑ i , k ≥ 0 , j ≥ 1 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) − ( m { 1 , 2 } ∘ m { 0 , 1 } ) n x 0 , … , x n + ∑ 1 ≤ l ≤ k ≤ n + 1 n = n 1 + ⋯ + n k n i ≥ 1 , i = 1 , … , k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) ∘ [ ( m { 1 , 2 } ∘ m { 0 , 1 } ) n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ ( m { 1 , 2 } ∘ m { 0 , 1 } ) n l − 1 x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = − m n { 0 , 2 } ; x 0 , … , x n . (220) \begin{aligned}
&\sum_{\substack{i,k\geq0,\ j\geq1\\i+j+k=n}}(-1)^{i+jk}m^{\{0,1,2\};x_{0},\ldots,x_{i},x_{i+j},x_{i+j+1},\ldots,x_{n}}_{i+k+1}\\
&\qquad\circ\left(\operatorname{id}_{\mathcal{H}^{0}(x_{0},x_{1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{i-1},x_{i})}\otimes m^{\{0\};x_{i},x_{i+1},\ldots,x_{i+j}}_{j}\right.\\
&\qquad\qquad\left.\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{n-1},x_{n})}\right)\\
&\quad-\left(m^{\{1,2\}}\circ m^{\{0,1\}}\right)^{x_{0},\ldots,x_{n}}_{n}\\
&\quad+\sum_{\substack{1\leq l\leq k\leq n+1\\n=n_{1}+\cdots+n_{k}\\n_{i}\geq1,\ i=1,\ldots,k}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_{i}+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_{i}+1)(k-i)}\\
&\qquad\qquad\cdot m^{\{2\};f^{\{0,1,2\}}(x_{p_{0}}),f^{\{0,1,2\}}(x_{p_{1}}),\ldots,f^{\{0,1,2\}}(x_{p_{l-1}}),f^{02}(x_{p_{l}}),\ldots,f^{02}(x_{p_{k}})}_{k}\\
&\qquad\qquad\circ\left[\left(m^{\{1,2\}}\circ m^{\{0,1\}}\right)^{x_{p_{0}},\ldots,x_{p_{1}}}_{n_{1}}\otimes\cdots\otimes\left(m^{\{1,2\}}\circ m^{\{0,1\}}\right)^{x_{p_{l-2}},\ldots,x_{p_{l-1}}}_{n_{l-1}}\right.\\
&\qquad\qquad\qquad\left.\otimes m^{\{0,1,2\};x_{p_{l-1}},\ldots,x_{p_{l}}}_{n_{l}}\otimes m^{\{0,2\};x_{p_{l}},\ldots,x_{p_{l+1}}}_{n_{l+1}}\otimes\cdots\otimes m^{\{0,2\};x_{p_{k-1}},\ldots,x_{p_{k}}}_{n_{k}}\right]\\
&\quad=-m^{\{0,2\};x_{0},\ldots,x_{n}}_{n}. \tag*{(220)}
\end{aligned} i , k ≥ 0 , j ≥ 1 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) − ( m { 1 , 2 } ∘ m { 0 , 1 } ) n x 0 , … , x n + 1 ≤ l ≤ k ≤ n + 1 n = n 1 + ⋯ + n k n i ≥ 1 , i = 1 , … , k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ i = 1 l − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) ⋅ m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) ∘ [ ( m { 1 , 2 } ∘ m { 0 , 1 } ) n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ ( m { 1 , 2 } ∘ m { 0 , 1 } ) n l − 1 x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = − m n { 0 , 2 } ; x 0 , … , x n . (220)
By equations (4.27), (4.28), (7.19), and (220) , m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 1 m^{\{0,1,2\}}=(m^{\{0,1,2\}}_{k})_{k\geq1} m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 1 is a unital A ∞ A_{\infty} A ∞ -homotopy F { 1 , 2 } ∘ F { 0 , 1 } ⇒ F { 0 , 2 } \mathcal{F}^{\{1,2\}}\circ\mathcal{F}^{\{0,1\}}\Rightarrow\mathcal{F}^{\{0,2\}} F { 1 , 2 } ∘ F { 0 , 1 } ⇒ F { 0 , 2 } . Thus, a 2-simplex in A ˙ ∞ C a t R , K u n , ∞ \dot{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K} A ˙ ∞ Cat R , K un , ∞ is a 2-commutative diagram of unital A ∞ A_{\infty} A ∞ -categories, unital A ∞ A_{\infty} A ∞ -functors, and a unital A ∞ A_{\infty} A ∞ -homotopy
A 0 → F { 0 , 1 } A 1 F { 0 , 2 } ↓ ↓ m { 0 , 1 , 2 } A 2 \begin{CD}
\mathcal{A}^{0} @>{\mathcal{F}^{\{0,1\}}}>> \mathcal{A}^{1}\\
@V{\mathcal{F}^{\{0,2\}}}VV @VV{m^{\{0,1,2\}}}V\\
@. \mathcal{A}^{2}
\end{CD} A 0 F { 0 , 2 } ↓ ⏐ F { 0 , 1 } A 1 ↓ ⏐ m { 0 , 1 , 2 } A 2
Note as in §4.4 that A ∞ A_{\infty} A ∞ -homotopies are less general than A ∞ A_{\infty} A ∞ -natural transformations and A ∞ A_{\infty} A ∞ -natural isomorphisms, which we met in Example 7.11(c) and Definition 7.12(iii).
The following analogue of Theorem 6.31 is proved in §10.8.
Theorem 7.26. There is a strict isomorphism of ∞ \infty ∞ -categories
I R , K u n : A ∞ C a t R , K u n , ∞ ⟶ A ˙ ∞ C a t R , K u n , ∞ , K ∈ { 3 , 4 , 5 , … , ∞ } , (221) I^{\mathrm{un}}_{R,K}:A_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K}\longrightarrow\dot{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K},\qquad K\in\{3,4,5,\ldots,\infty\}, \tag*{(221)} I R , K un : A ∞ Cat R , K un , ∞ ⟶ A ˙ ∞ Cat R , K un , ∞ , K ∈ { 3 , 4 , 5 , … , ∞ } , (221) from Definitions 7.19 and 7.22, which acts on N N N -simplices by
I R , K u n : ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) ⟼ I^{\mathrm{un}}_{R,K}:\left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subset\{0,\ldots,N\}}\right)\longmapsto I R , K un : ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) ⟼ ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ ≠ I ⊂ { 0 , … , N } ) , \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(\dot{m}^{I})_{\varnothing\neq I\subset\{0,\ldots,N\}}\right), ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ = I ⊂ { 0 , … , N } ) , where m ˙ k { i 0 , … , i M } ; x 0 , … , x k \dot{m}^{\{i_{0},\ldots,i_{M}\};x_{0},\ldots,x_{k}}_{k} m ˙ k { i 0 , … , i M } ; x 0 , … , x k is defined from m k { i 0 , … , i M } ; x 0 , … , x k m^{\{i_{0},\ldots,i_{M}\};x_{0},\ldots,x_{k}}_{k} m k { i 0 , … , i M } ; x 0 , … , x k by
m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { 0 , k = 0 and M = 2 , m k { i 0 , … , i M } ; x 0 , … , x k , otherwise. \dot{m}^{\{i_{0},\ldots,i_{M}\};x_{0},\ldots,x_{k}}_{k}=
\begin{cases}
0, & k=0\text{ and }M=2,\\
m^{\{i_{0},\ldots,i_{M}\};x_{0},\ldots,x_{k}}_{k}, & \text{otherwise.}
\end{cases} m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { 0 , m k { i 0 , … , i M } ; x 0 , … , x k , k = 0 and M = 2 , otherwise.
Here are the analogues of Definitions 6.33 and 6.34.
Definition 7.27. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K \in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Generalizing Definition 7.22, define a simplicial set A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ of cohomologically unital A ∞ A_{\infty} A ∞ -categories over R R R such that the N N N -simplices ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) for N ≥ 0 N\geq0 N ≥ 0 satisfy Definition 7.22(a)(i)–(iv), but instead of Definition 7.22(a)(v), we require that
(v)′ As in Example 7.25(a), Definition 7.22(a)(i)–(iv) imply that A i = ( O i , H i , m { i } ) \mathcal{A}^{i}=(\mathcal{O}^{i},\mathcal{H}^{i},m^{\{i\}}) A i = ( O i , H i , m { i } ) is an A ∞ A_{\infty} A ∞ -category for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N . We require this A ∞ A_{\infty} A ∞ -category to be cohomologically unital , as in Definition 4.7.
(vi)′ As in Example 7.25(b), Definition 7.22(a)(i)–(iv) mean F i j = ( f i j , m { i , j } ) : A i → A j \mathcal{F}^{ij}=(f^{ij},m^{\{i,j\}}):\mathcal{A}^{i}\to\mathcal{A}^{j} F ij = ( f ij , m { i , j } ) : A i → A j is an A ∞ A_{\infty} A ∞ -functor for all 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N . We require this to be a cohomologically unital A ∞ A_{\infty} A ∞ -functor, as in Definition 4.9.
If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , define A ∞ C a t R , K c u , ∞ ( α ) : ( A ∞ C a t R , K c u , ∞ ) N → ( A ∞ C a t R , K c u , ∞ ) N ‾ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty}(\alpha):(\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty})_{N}\to(\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty})_{\overline{N}} A ∞ Cat R , K cu , ∞ ( α ) : ( A ∞ Cat R , K cu , ∞ ) N → ( A ∞ Cat R , K cu , ∞ ) N as in Definition 7.22(b). The proof of Proposition 7.23 also works in the cohomologically unital case, and shows A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ is a simplicial set. It contains A ∞ C a t R , K u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{un},\infty} A ∞ Cat R , K un , ∞ in Definition 7.22 as a simplicial subset.
Theorem 7.24 also holds in A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ , as the proof does not use unitality.
Theorem 7.29 will show that A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ is an ∞ \infty ∞ -category. We say that A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ has type K K K . We also write A ∞ C a t R c u , ∞ = A ∞ C a t R , 3 c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{cu},\infty}=\mathrm{A}_{\infty}\mathrm{Cat}_{R,3}^{\mathrm{cu},\infty} A ∞ Cat R cu , ∞ = A ∞ Cat R , 3 cu , ∞ .
Following Definition 7.21 we can define ∞ \infty ∞ -functors A ∞ A l g R , K c u , ∞ → A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty}\to\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ → A ∞ Cat R , K cu , ∞ for all K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } , for A ∞ A l g R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\mathrm{cu},\infty} A ∞ Alg R , K cu , ∞ as in Definition 6.33, which are strict isomorphisms with the full ∞ \infty ∞ -subcategory of A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ with objects cohomologically unital A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A with one object ∗ * ∗ .
Definition 7.28. Let R R R be a commutative ring and K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } . Generalizing Definition 7.22, define a simplicial set A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ of (possibly non-unital) A ∞ A_{\infty} A ∞ -categories over R R R whose N N N -simplices ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) for N ≥ 0 N\geq0 N ≥ 0 satisfy Definition 7.22(a)(i)–(iv), but need not satisfy (a)(v). If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , define A ∞ C a t R , K ∞ ( α ) : ( A ∞ C a t R , K ∞ ) N → ( A ∞ C a t R , K ∞ ) N ‾ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty}(\alpha):(\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty})_{N}\to(\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty})_{\overline{N}} A ∞ Cat R , K ∞ ( α ) : ( A ∞ Cat R , K ∞ ) N → ( A ∞ Cat R , K ∞ ) N as in Definition 7.22(b). The proof of Proposition 7.23 also works in the non-unital case, and shows A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ is a simplicial set. It contains A ∞ C a t R , K c u , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\mathrm{cu},\infty} A ∞ Cat R , K cu , ∞ in Definition 7.27 as a simplicial subset.
Theorem 7.24 also holds in A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ , as the proof does not use unitality.
Theorem 7.29 will show that A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ is an ∞ \infty ∞ -category. We say that A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ has type K K K . We also write A ∞ C a t R ∞ = A ∞ C a t R , 3 ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\infty}=\mathrm{A}_{\infty}\mathrm{Cat}_{R,3}^{\infty} A ∞ Cat R ∞ = A ∞ Cat R , 3 ∞ .
Following Definition 7.21 we can define ∞ \infty ∞ -functors A ∞ A l g R , K ∞ → A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\infty}\to\mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Alg R , K ∞ → A ∞ Cat R , K ∞ for all K ∈ { 3 , 4 , 5 , … , ∞ } K\in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } , for A ∞ A l g R , K ∞ \mathrm{A}_{\infty}\mathrm{Alg}_{R,K}^{\infty} A ∞ Alg R , K ∞ as in Definition 6.34, which are strict isomorphisms with the full ∞ \infty ∞ -subcategory of A ∞ C a t R , K ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R,K}^{\infty} A ∞ Cat R , K ∞ with objects A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A with one object ∗ * ∗ .
The following generalization of Theorem 6.35, one of our main results, is proved in §10.9. The proof is easier than that of Theorem 7.13.
Theorem 7.29. The following simplicial sets defined in Definitions 7.19, 7.22, 7.27, and 7.28, for all K ∈ { 3 , 4 , 5 , … , ∞ } K \in\{3,4,5,\ldots,\infty\} K ∈ { 3 , 4 , 5 , … , ∞ } , are all ∞ \infty ∞ -categories:
A ∞ C a t R , K u n , ∞ , A ˙ ∞ C a t R , K u n , ∞ , A ∞ C a t R , K c u , ∞ , A ∞ C a t R , K ∞ . (222) A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,K}^{\mathrm{un},\infty},\ \dot{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,K}^{\mathrm{un},\infty},\ A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,K}^{\mathrm{cu},\infty},\ A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,K}^{\infty}. \tag*{(222)} A ∞ Cat R , K un , ∞ , A ˙ ∞ Cat R , K un , ∞ , A ∞ Cat R , K cu , ∞ , A ∞ Cat R , K ∞ . (222)
Here is the analogue of Definition 6.36. It completes our diagram (7.1) of inclusions of ∞ \infty ∞ -categories of A ∞ A_{\infty} A ∞ -categories.
Definition 7.30. Let A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un be the category from Definition 4.9, and N ( A ∞ C a t R u n ) \mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}}) N ( A ∞ Cat R un ) be its nerve from Example 2.11, which is an ∞ \infty ∞ -category. Then N N N -simplices in N ( A ∞ C a t R u n ) \mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}}) N ( A ∞ Cat R un ) are of the form ( ( A i ) i = 0 N , ( F i ) i = 1 N ) ((\mathcal{A}^{i})_{i=0}^{N},(\mathcal{F}^{i})_{i=1}^{N}) (( A i ) i = 0 N , ( F i ) i = 1 N ) , where A i \mathcal{A}^{i} A i is a unital A ∞ A_{\infty} A ∞ -category over R R R for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and F i : A i − 1 → A i \mathcal{F}^{i}:\mathcal{A}^{i-1}\to\mathcal{A}^{i} F i : A i − 1 → A i is a unital A ∞ A_{\infty} A ∞ -functor for i = 1 , … , N i=1,\ldots,N i = 1 , … , N .
Define an ∞ \infty ∞ -functor inc R u n : N ( A ∞ C a t R u n ) → A ˙ ∞ C a t R , ∞ u n , ∞ \operatorname{inc}_{R}^{\mathrm{un}}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\to\dot{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} inc R un : N ( A ∞ Cat R un ) → A ˙ ∞ Cat R , ∞ un , ∞ to act on N N N -simplices for N ≥ 0 N\geq0 N ≥ 0 mapping
( ( A i ) i = 0 N , ( F i ) i = 1 N ) ⟼ ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) , ((\mathcal{A}^{i})_{i=0}^{N},(\mathcal{F}^{i})_{i=1}^{N})\longmapsto((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}), (( A i ) i = 0 N , ( F i ) i = 1 N ) ⟼ (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) ,
where we set O i = Obj A i \mathcal{O}^{i}=\operatorname{Obj}\mathcal{A}^{i} O i = Obj A i , and writing F i i = id A i : A i → A i \mathcal{F}^{ii}=\operatorname{id}_{\mathcal{A}^{i}}:\mathcal{A}^{i}\to\mathcal{A}^{i} F ii = id A i : A i → A i and F i j = F j ∘ ⋯ ∘ F i + 1 : A i → A j \mathcal{F}^{ij}=\mathcal{F}^{j}\circ\cdots\circ\mathcal{F}^{i+1}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij = F j ∘ ⋯ ∘ F i + 1 : A i → A j for 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N , we define f i j = F O b j i j : O i → O j f^{ij}=\mathcal{F}^{ij}_{\mathrm{Obj}}:\mathcal{O}^{i}\to\mathcal{O}^{j} f ij = F Obj ij : O i → O j for 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N . We set H i ( x , y ) = Hom A i ( x , y ) \mathcal{H}^{i}(x,y)=\operatorname{Hom}_{\mathcal{A}^{i}}(x,y) H i ( x , y ) = Hom A i ( x , y ) for x , y ∈ O i = Obj A i x,y\in\mathcal{O}^{i}=\operatorname{Obj}\mathcal{A}^{i} x , y ∈ O i = Obj A i . We take m { i } ; x 0 , … , x k k m_{\{i\};x_{0},\ldots,x_{k}}^{k} m { i } ; x 0 , … , x k k to be the map m x 0 , … , x k k m_{x_{0},\ldots,x_{k}}^{k} m x 0 , … , x k k in A i \mathcal{A}^{i} A i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and m { i , j } ; x 0 , … , x k k m_{\{i,j\};x_{0},\ldots,x_{k}}^{k} m { i , j } ; x 0 , … , x k k to be the map F i j ; x 0 , … , x k k \mathcal{F}_{ij;x_{0},\ldots,x_{k}}^{k} F ij ; x 0 , … , x k k in F i j \mathcal{F}^{ij} F ij for 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N , and m I ; x 0 , … , x k k = 0 m_{I;x_{0},\ldots,x_{k}}^{k}=0 m I ; x 0 , … , x k k = 0 if ∣ I ∣ ≥ 3 \lvert I\rvert\geq3 ∣ I ∣ ≥ 3 and k ≥ 0 k\geq0 k ≥ 0 .
We claim ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) is an N N N -simplex of A ˙ ∞ C a t R , ∞ u n , ∞ \dot{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} A ˙ ∞ Cat R , ∞ un , ∞ . We can verify (7.25) for all ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } , n ≥ 0 n\geq0 n ≥ 0 and x 0 , … , x n ∈ O min I x_{0},\ldots,x_{n}\in\mathcal{O}^{\min I} x 0 , … , x n ∈ O m i n I in a very similar way to Definition 6.36(i)–(iv). Definition 7.22(a)(v) follows as the A i \mathcal{A}^{i} A i , F i j \mathcal{F}^{ij} F ij are unital.
It is now easy to see that inc R u n \operatorname{inc}_{R}^{\mathrm{un}} inc R un is a morphism of simplicial sets, that is, an ∞ \infty ∞ -functor. It is a strict isomorphism with the ∞ \infty ∞ -subcategory of A ˙ ∞ C a t R , ∞ u n , ∞ \dot{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} A ˙ ∞ Cat R , ∞ un , ∞ consisting of N N N -simplices ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) with m I ; x 0 , … , x k k = 0 m_{I;x_{0},\ldots,x_{k}}^{k}=0 m I ; x 0 , … , x k k = 0 if ∣ I ∣ ≥ 3 \lvert I\rvert\geq3 ∣ I ∣ ≥ 3 and k ≥ 0 k\geq0 k ≥ 0 . Similarly we define ∞ \infty ∞ -functors inc R c u : N ( A ∞ C a t R c u ) ↪ A ∞ C a t R , ∞ c u , ∞ \operatorname{inc}_{R}^{\mathrm{cu}}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{cu}})\hookrightarrow A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{cu},\infty} inc R cu : N ( A ∞ Cat R cu ) ↪ A ∞ Cat R , ∞ cu , ∞ for cohomologically unital A ∞ A_{\infty} A ∞ -categories, and inc R : N ( A ∞ C a t R ) ↪ A ∞ C a t R , ∞ ∞ \operatorname{inc}_{R}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R})\hookrightarrow A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\infty} inc R : N ( A ∞ Cat R ) ↪ A ∞ Cat R , ∞ ∞ for general A ∞ A_{\infty} A ∞ -categories.
We can also define an ∞ \infty ∞ -functor i n c ˘ R u n : N ( A ∞ C a t R u n ) → A ∞ C a t R , ∞ u n , ∞ \operatorname{\breve{inc}}_{R}^{\mathrm{un}}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\to A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} inc ˘ R un : N ( A ∞ Cat R un ) → A ∞ Cat R , ∞ un , ∞ which is identified with inc R u n \operatorname{inc}_{R}^{\mathrm{un}} inc R un under the isomorphism I R , ∞ u n I_{R,\infty}^{\mathrm{un}} I R , ∞ un in (7.29). The difference is that as in (7.23), when we define ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) above we set m { i 0 , i 1 , i 2 } ; x 0 0 = 1 i 0 i 2 ( x 0 ) m_{\{i_{0},i_{1},i_{2}\};x_{0}}^{0}=\mathbb{1}_{i_{0}i_{2}}(x_{0}) m { i 0 , i 1 , i 2 } ; x 0 0 = 1 i 0 i 2 ( x 0 ) , not m { i 0 , i 1 , i 2 } ; x 0 0 = 0 m_{\{i_{0},i_{1},i_{2}\};x_{0}}^{0}=0 m { i 0 , i 1 , i 2 } ; x 0 0 = 0 .
Similarly we define an ∞ \infty ∞ -functor i n c ˘ R u n : N ( A ∞ C a t R u n ) → A ∞ C a t R , ∞ u n , ∞ \operatorname{\breve{inc}}_{R}^{\mathrm{un}}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\to A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} inc ˘ R un : N ( A ∞ Cat R un ) → A ∞ Cat R , ∞ un , ∞ .
Remark 7.31. Composing i n c ˘ R u n \operatorname{\breve{inc}}_{R}^{\mathrm{un}} inc ˘ R un with A ∞ C a t R , ∞ u n , ∞ ↪ A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty}\hookrightarrow A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R , ∞ un , ∞ ↪ A ∞ Cat R un gives an ∞ \infty ∞ -functor i n c ˘ R u n : N ( A ∞ C a t R u n ) → A ∞ C a t R , ∞ u n , ∞ \operatorname{\breve{inc}}_{R}^{\mathrm{un}}:\mathcal{N}(A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\to A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R,\infty}^{\mathrm{un},\infty} inc ˘ R un : N ( A ∞ Cat R un ) → A ∞ Cat R , ∞ un , ∞ , which is a strict isomorphism with an ∞ \infty ∞ -subcategory of A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un .
It follows from Example 7.11(c) and Definition 7.12 that if morphisms F , G : A → B \mathcal{F},\mathcal{G}:\mathcal{A}\to\mathcal{B} F , G : A → B in A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un are unitaly A ∞ A_{\infty} A ∞ -naturally isomorphic, then their images under inc R u n \operatorname{inc}_{R}^{\mathrm{un}} inc R un are homotopic in A ∞ C a t R u n , ∞ A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , that is [ inc R u n ( F ) ] = [ inc R u n ( G ) ] [\operatorname{inc}_{R}^{\mathrm{un}}(\mathcal{F})]=[\operatorname{inc}_{R}^{\mathrm{un}}(\mathcal{G})] [ inc R un ( F )] = [ inc R un ( G )] in Ho ( A ∞ Cat R u n , ∞ ) \operatorname{Ho}(\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty}) Ho ( A ∞ Cat R un , ∞ ) . Therefore inc R u n \operatorname{inc}_{R}^{\mathrm{un}} inc R un maps unital homotopy equivalences in A ∞ Cat R u n \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un , as in Definition 4.27, to equivalences in A ∞ Cat R u n , ∞ \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ .
Thus by the universal property of ∞ \infty ∞ -category localizations in §2.6, there exists an ∞ \infty ∞ -functor Φ \Phi Φ in the 2-commutative diagram
N ( A ∞ Cat R u n ) → Π l o c ∞ A ∞ Cat R u n [ H E u n − 1 ] ∞ inc R u n ↓ ↑ Φ A ∞ Cat R u n , ∞ = A ∞ Cat R u n , ∞ \begin{CD}
\mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}) @>{\Pi_{\mathrm{loc}}^{\infty}}>> \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}}[\mathcal{HE}_{\mathrm{un}}^{-1}]^{\infty} \\
@V{\operatorname{inc}_{R}^{\mathrm{un}}}VV @AA{\Phi}A \\
\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} @= \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ Π loc ∞ A ∞ Cat R un [ HE un − 1 ] ∞ ⏐ ↑ Φ A ∞ Cat R un , ∞
where H E u n \mathcal{HE}_{\mathrm{un}} HE un is the class of unital homotopy equivalences in A ∞ Cat R u n \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un . Theorem 9.7 will show that Φ \Phi Φ is an equivalence of ∞ \infty ∞ -categories.
Definition 7.32. Define ∞ \infty ∞ -subcategories
A ∞ Cat R u n , s u , ∞ ⊂ A ∞ Cat R u n , ∞ , A ∞ Cat R u n , h s , ∞ ⊂ A ∞ Cat R u n , ∞ , \mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\mathrm{su},\infty}\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty},\qquad
\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\mathrm{hs},\infty}\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty}, A ∞ Cat R un , su , ∞ ⊂ A ∞ Cat R un , ∞ , A ∞ Cat R un , hs , ∞ ⊂ A ∞ Cat R un , ∞ , to be the full ∞ \infty ∞ -subcategories with objects that are unital A ∞ A_{\infty} A ∞ -categories with split units for superscript ‘su’, as in §4.6, and objects that are h h h -projective unital A ∞ A_{\infty} A ∞ -categories with split units for superscript ‘hs’, as in §4.6–§4.7.
Definition 7.33. Define ∞ \infty ∞ -subcategories
A ∞ △ Cat R u n , ∞ ⊂ A ∞ Cat R u n , ∞ , A ∞ △ Cat R , K u n , ∞ ⊂ A ∞ Cat R , K u n , ∞ , A ∞ △ Cat R c u , ∞ ⊂ A ∞ Cat R c u , ∞ , A ∞ △ Cat R , K c u , ∞ ⊂ A ∞ Cat R , K c u , ∞ , A ∞ △ Cat R ∞ ⊂ A ∞ Cat R ∞ , A ∞ △ Cat R , K ∞ ⊂ A ∞ Cat R , K ∞ , \begin{aligned}
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R}^{\mathrm{un},\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty},&
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R,K}^{\mathrm{un},\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R,K}^{\mathrm{un},\infty},\\
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R}^{\mathrm{cu},\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{cu},\infty},&
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R,K}^{\mathrm{cu},\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R,K}^{\mathrm{cu},\infty},\\
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R}^{\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\infty},&
\mathcal{A}_{\infty}^{\triangle}\operatorname{Cat}_{R,K}^{\infty}&\subset\mathcal{A}_{\infty}\operatorname{Cat}_{R,K}^{\infty},
\end{aligned} A ∞ △ Cat R un , ∞ A ∞ △ Cat R cu , ∞ A ∞ △ Cat R ∞ ⊂ A ∞ Cat R un , ∞ , ⊂ A ∞ Cat R cu , ∞ , ⊂ A ∞ Cat R ∞ , A ∞ △ Cat R , K un , ∞ A ∞ △ Cat R , K cu , ∞ A ∞ △ Cat R , K ∞ ⊂ A ∞ Cat R , K un , ∞ , ⊂ A ∞ Cat R , K cu , ∞ , ⊂ A ∞ Cat R , K ∞ , for K ∈ { 3 , 4 , … , ∞ } K\in\{3,4,\ldots,\infty\} K ∈ { 3 , 4 , … , ∞ } to be the full ∞ \infty ∞ -subcategories whose objects are pretriangulated A ∞ A_{\infty} A ∞ -categories over R R R , as in Definition 4.60.
Definition 7.34. Define ∞ \infty ∞ -categories DGCat R ∗ , ∞ \operatorname{DGCat}_{R}^{*,\infty} DGCat R ∗ , ∞ in a commutative diagram of inclusions of ∞ \infty ∞ -subcategories:
DGCat R h s , ∞ → ≄ a l l R DGCat R A ∞ , h s , ∞ → ≃ A ∞ Cat R u n , h s , ∞ ↓ ≄ ↓ ≄ ↓ ≄ DGCat R s u , ∞ → ≄ a l l R DGCat R A ∞ , s u , ∞ → ≃ A ∞ Cat R u n , s u , ∞ ↓ ≄ ↓ ≄ ↓ ≄ DGCat R ∞ → ≄ a l l R DGCat R A ∞ , ∞ ⟶ A ∞ Cat R u n , ∞ , (223) \begin{aligned}
\operatorname{DGCat}_{R}^{\mathrm{hs},\infty}&\xrightarrow{\not\simeq\ \mathrm{all}\ R}\operatorname{DGCat}_{R}^{A_{\infty},\mathrm{hs},\infty}\xrightarrow{\simeq}\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\mathrm{hs},\infty}\\
\downarrow{\not\simeq}&&\downarrow{\not\simeq}&&\downarrow{\not\simeq}\\
\operatorname{DGCat}_{R}^{\mathrm{su},\infty}&\xrightarrow{\not\simeq\ \mathrm{all}\ R}\operatorname{DGCat}_{R}^{A_{\infty},\mathrm{su},\infty}\xrightarrow{\simeq}\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\mathrm{su},\infty}\\
\downarrow{\not\simeq}&&\downarrow{\not\simeq}&&\downarrow{\not\simeq}\\
\operatorname{DGCat}_{R}^{\infty}&\xrightarrow{\not\simeq\ \mathrm{all}\ R}\operatorname{DGCat}_{R}^{A_{\infty},\infty}\longrightarrow\mathcal{A}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty}, \tag*{(223)}
\end{aligned} DGCat R hs , ∞ ↓ ≃ DGCat R su , ∞ ↓ ≃ DGCat R ∞ ≃ all R DGCat R A ∞ , hs , ∞ ≃ A ∞ Cat R un , hs , ∞ ≃ all R DGCat R A ∞ , su , ∞ ≃ A ∞ Cat R un , su , ∞ ≃ all R DGCat R A ∞ , ∞ ⟶ A ∞ Cat R un , ∞ , ↓ ≃ ↓ ≃ ↓ ≃ ↓ ≃ (223) such that:
(i) Objects of DGCat R ∗ , ∞ \operatorname{DGCat}_{R}^{*,\infty} DGCat R ∗ , ∞ are dg-categories over R R R , which are h h h -projective with split units in the top row of (7.31) with superscript ‘hs’, and have split units in the middle row of (7.31) with superscript ‘su’.
(ii) In the left hand column of (7.31), 1-morphisms in D G C a t R ∗ , ∞ \mathbf{DGCat}_{R}^{*,\infty} DGCat R ∗ , ∞ are dg-functors A → B \mathcal{A} \to\mathcal{B} A → B between dg-categories. In the middle column of (7.31), with superscript A ∞ A_\infty A ∞ , 1-morphisms in D G C a t R A ∞ , ∗ , ∞ \mathbf{DGCat}_{R}^{A_\infty,*,\infty} DGCat R A ∞ , ∗ , ∞ are unital A ∞ A_\infty A ∞ -functors A → B \mathcal{A} \to\mathcal{B} A → B between dg-categories.
(iii) For N ≥ 2 N \ge2 N ≥ 2 , the N N N -simplices in each D G C a t R ∗ , ∞ \mathbf{DGCat}_{R}^{*,\infty} DGCat R ∗ , ∞ are all N N N -simplices in A ∞ C a t R u n , ∞ \mathbf{A_\infty Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ which have 0- and 1-simplices as prescribed in (i),(ii).
Since the classes of 1-morphisms in (ii) are closed under composition, these do define ∞ \infty ∞ -subcategories. See Theorem 7.35 for the meaning of the labels ‘≃ \simeq ≃ ’, ‘≄ \not\simeq ≃ ’ on the inclusions in (7.31).
Theorem 7.35. In equation (7.31), the inclusions marked ‘≃ \simeq ≃ ’ are Joyal equivalences for all commutative rings R R R . For the inclusions marked ‘≄ \not\simeq ≃ ’, there exist commutative rings R R R (e.g. R = Z R = \mathbb{Z} R = Z ) for which they are not Joyal equivalences. For those marked ‘≄ \not\simeq ≃ all R R R ’, they are not Joyal equivalences for all R R R .
If R R R is a field, the upper vertical morphisms in (7.31) are identity morphisms, so they are Joyal equivalences in this case, even though marked ‘≄ \not\simeq ≃ ’.
Proof. By Theorem 2.20 , an inclusion in (7.31) is a Joyal equivalence if it is essentially surjective, as in Definition 2.12 , and fully faithful, as in Definition 2.17 . The two inclusions D G C a t R A ∞ , h s , ∞ ↪ A ∞ C a t R u n , h s , ∞ \mathbf{DGCat}_{R}^{A_\infty,\mathrm{hs},\infty} \hookrightarrow\mathbf{A_\infty Cat}_{R}^{\mathrm{un},\mathrm{hs},\infty} DGCat R A ∞ , hs , ∞ ↪ A ∞ Cat R un , hs , ∞ and D G C a t R A ∞ , s u , ∞ ↪ A ∞ C a t R u n , s u , ∞ \mathbf{DGCat}_{R}^{A_\infty,\mathrm{su},\infty} \hookrightarrow\mathbf{A_\infty Cat}_{R}^{\mathrm{un},\mathrm{su},\infty} DGCat R A ∞ , su , ∞ ↪ A ∞ Cat R un , su , ∞ are essentially surjective by Theorem 4.46 and Proposition 4.51 . They are trivially fully faithful as they are inclusions of full ∞ \infty ∞ -subcategories.
For the inclusions marked ‘≄ \not\simeq ≃ ’ in (7.31), Example 7.36 shows they are not Joyal equivalences when R = Z R = \mathbb{Z} R = Z . The last part holds as in §4.6–§4.7.
Example 7.36. (a) Take R = Z R = \mathbb{Z} R = Z and let A \mathcal{A} A be the dg-category over R R R with one object ∗ * ∗ and Hom A ( ∗ , ∗ ) = Z 2 ⋅ id ∗ \operatorname{Hom}_{\mathcal{A}}(*,*) = \mathbb{Z}_{2} \cdot\operatorname{id}_{*} Hom A ( ∗ , ∗ ) = Z 2 ⋅ id ∗ . Then A \mathcal{A} A does not have split units, so does not lie in the essential image of the vertical inclusions between the second and third rows of (7.31) by Proposition 4.40 . Thus these are not Joyal equivalences.
(b) Take R = Z R = \mathbb{Z} R = Z and let A \mathcal{A} A be the dg-category over R R R with one object ∗ * ∗ and Hom A ( ∗ , ∗ ) = Z ⋅ id ∗ ⊕ Z 2 ⋅ ϵ \operatorname{Hom}_{\mathcal{A}}(*,*) = \mathbb{Z} \cdot\operatorname{id}_{*} \oplus\mathbb{Z}_{2} \cdot\epsilon Hom A ( ∗ , ∗ ) = Z ⋅ id ∗ ⊕ Z 2 ⋅ ϵ with ϵ 2 = 0 \epsilon^{2} = 0 ϵ 2 = 0 . Then A \mathcal{A} A is not h-projective, so does not lie in the essential image of the vertical inclusions between the first and second rows of (7.31) by Proposition 4.51 . Hence these are not Joyal equivalences.
(c) Let R R R be any ring. Take A \mathcal{A} A to be the dg-category over R R R with one object ∗ * ∗ and Hom A ( ∗ , ∗ ) = R ⋅ id ∗ ⊕ R ⋅ x \operatorname{Hom}_{\mathcal{A}}(*,*) = R \cdot\operatorname{id}_{*} \oplus R \cdot x Hom A ( ∗ , ∗ ) = R ⋅ id ∗ ⊕ R ⋅ x with deg x = 0 \deg x = 0 deg x = 0 and d x = 0 dx = 0 d x = 0 . Take B \mathcal{B} B to be the dg-category over R R R with one object ∗ * ∗ and Hom B ( ∗ , ∗ ) = R ⋅ id ∗ ⊕ R ⋅ y \operatorname{Hom}_{\mathcal{B}}(*,*) = R \cdot\operatorname{id}_{*} \oplus R \cdot y Hom B ( ∗ , ∗ ) = R ⋅ id ∗ ⊕ R ⋅ y with deg y = − 1 \deg y = -1 deg y = − 1 and d y = 0 dy = 0 d y = 0 . Define F : A → B \mathcal{F} \colon\mathcal{A} \to\mathcal{B} F : A → B to be the unital A ∞ A_\infty A ∞ -functor with F o b j ( ∗ ) = ∗ \mathcal{F}_{\mathrm{obj}}(*) = * F obj ( ∗ ) = ∗ , F 1 ( id ∗ ) = id ∗ \mathcal{F}_{1}(\operatorname{id}_{*}) = \operatorname{id}_{*} F 1 ( id ∗ ) = id ∗ , F 1 ( x ) = 0 \mathcal{F}_{1}(x) = 0 F 1 ( x ) = 0 and F 2 ( x , x ) = y \mathcal{F}_{2}(x,x) = y F 2 ( x , x ) = y . Then F \mathcal{F} F is not A ∞ A_\infty A ∞ -naturally isomorphic to any dg-functor F : A → B \mathcal{F} \colon\mathcal{A} \to\mathcal{B} F : A → B . This means that all three horizontal inclusions between the first and second columns of (7.31) are not fully faithful, and so are not Joyal equivalences.
∞-categories of filtered A ∞ A_{\infty} A ∞ -algebras and filtered A ∞ A_{\infty} A ∞ -categories
Filtered A ∞ A_{\infty} A ∞ -categories were discussed in Chapter 5. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, and R R R a complete Γ \Gamma Γ -filtered ring. We will define many ∞ \infty ∞ -categories of filtered A ∞ A_{\infty} A ∞ -categories over R R R , which fit into a diagram of inclusions of ∞ \infty ∞ -categories:
A ∞ f i Cat R u n , ∞ = A ∞ f i Cat R , 0 , 2 u n , ∞ A ∞ f i Cat R , 0 , 3 u n , ∞ ⟶ A ˙ ∞ f i Cat R , 0 , 3 u n , ∞ ⟶ A ∞ f i Cat R , 0 , 3 ∞ A ∞ f i Cat R , K ′ , L ′ u n , ∞ ⟶ A ˙ ∞ f i Cat R , K ′ , L ′ u n , ∞ ⟶ A ∞ f i Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) A ∞ f i Cat R , K , L u n , ∞ ⟶ A ˙ ∞ f i Cat R , K , L u n , ∞ ⟶ A ∞ f i Cat R , K , L ∞ ( K ≥ K ′ , L ≥ L ′ ) ( 0 ≤ K ≤ L , L ≥ 3 ) A ∞ f i Cat R , K , ∞ u n , ∞ ⟶ A ˙ ∞ f i Cat R , K , ∞ u n , ∞ ⟶ A ∞ f i Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) . N ( A ∞ f i Cat R , K u n ) ⟶ N ( A ˙ ∞ f i Cat R , K u n ) ⟶ N ( A ∞ f i Cat R , K ) (224) \begin{aligned}
A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R}^{\mathrm{un},\infty}
&=A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,0,2}^{\mathrm{un},\infty}\\
A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,0,3}^{\mathrm{un},\infty}
&\longrightarrow\dot{A}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,0,3}^{\mathrm{un},\infty}
\longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,0,3}^{\infty}\\
A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K',L'}^{\mathrm{un},\infty}
&\longrightarrow\dot{A}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K',L'}^{\mathrm{un},\infty}
\longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K',L'}^{\infty}
\qquad(0\leq K'\leq L',\ L'\geq3)\\
A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}
&\longrightarrow\dot{A}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}
\longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\infty}
\qquad(K\geq K',\ L\geq L')\\
&&\hspace{-4em}(0\leq K\leq L,\ L\geq3)\\
A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,\infty}^{\mathrm{un},\infty}
&\longrightarrow\dot{A}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,\infty}^{\mathrm{un},\infty}
\longrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,\infty}^{\infty}
\qquad(K=0,1,\infty).\\
N\left(A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K}^{\mathrm{un}}\right)
&\longrightarrow N\left(\dot{A}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K}^{\mathrm{un}}\right)
\longrightarrow N\left(A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K}\right) \tag*{(224)}
\end{aligned} A ∞ fi Cat R un , ∞ A ∞ fi Cat R , 0 , 3 un , ∞ A ∞ fi Cat R , K ′ , L ′ un , ∞ A ∞ fi Cat R , K , L un , ∞ A ∞ fi Cat R , K , ∞ un , ∞ N ( A ∞ fi Cat R , K un ) = A ∞ fi Cat R , 0 , 2 un , ∞ ⟶ A ˙ ∞ fi Cat R , 0 , 3 un , ∞ ⟶ A ∞ fi Cat R , 0 , 3 ∞ ⟶ A ˙ ∞ fi Cat R , K ′ , L ′ un , ∞ ⟶ A ∞ fi Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) ⟶ A ˙ ∞ fi Cat R , K , L un , ∞ ⟶ A ∞ fi Cat R , K , L ∞ ( K ≥ K ′ , L ≥ L ′ ) ⟶ A ˙ ∞ fi Cat R , K , ∞ un , ∞ ⟶ A ∞ fi Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) . ⟶ N ( A ˙ ∞ fi Cat R , K un ) ⟶ N ( A ∞ fi Cat R , K ) ( 0 ≤ K ≤ L , L ≥ 3 ) (224)
We then define ∞ \infty ∞ -categories of filtered A ∞ A_{\infty} A ∞ -algebras by restricting to the full ∞ \infty ∞ -subcategories of filtered A ∞ A_{\infty} A ∞ -categories with one object ∗ \ast ∗ .
We comment on these:
(i) The objects of the ∞ \infty ∞ -categories in the first two columns, with superscript ‘un’, are unital filtered A ∞ A_{\infty} A ∞ -categories.
(ii) The objects in the third column are all filtered A ∞ A_{\infty} A ∞ -categories.
(In contrast to Chapter 7 we do not define ∞ \infty ∞ -categories of cohomologically unital filtered A ∞ A_{\infty} A ∞ -categories, as we do not have a good definition of these.)
(iii) The ∞ \infty ∞ -categories in the first two columns, linked by ‘=’, are strictly isomorphic, but with definitions written in a different way.
(iv) We have subscripts R , K , L R,K,L R , K , L for K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\leq L K ≤ L , and L ≥ 2 L\geq2 L ≥ 2 for the first column and L ≥ 3 L\geq3 L ≥ 3 for the second and third. We call ( K , L ) (K,L) ( K , L ) the type of the ∞ \infty ∞ -category, generalizing type K K K in Chapters 6–7. We have inclusions A ∞ f i Cat R , K , L ∗ , ∞ ↪ A ∞ f i Cat R , K ′ , L ′ ∗ , ∞ A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\ast,\infty}\hookrightarrow A_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K',L'}^{\ast,\infty} A ∞ fi Cat R , K , L ∗ , ∞ ↪ A ∞ fi Cat R , K ′ , L ′ ∗ , ∞ if K ≥ K ′ K\geq K' K ≥ K ′ and L ≥ L ′ L\geq L' L ≥ L ′ .
As in Chapters 6–7, the N N N -simplices in the ∞ \infty ∞ -categories in (224) all involve families of maps m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k for ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } , k ≥ 0 k\geq0 k ≥ 0 and x 0 , … , x k x_{0},\ldots,x_{k} x 0 , … , x k in Obj A min I \operatorname{Obj}A^{\min I} Obj A m i n I , where m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k is filtration-preserving.
For type ( K , L ) (K,L) ( K , L ) and ∣ I ∣ ≠ 3 |I| \ne3 ∣ I ∣ = 3 , the rule is that m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) ∈ F γ > 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{\gamma>0}\operatorname{Hom}_{\mathcal{A}_{i_{M}}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F γ > 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) if K < ∣ I ∣ ≤ L K<|I|\le L K < ∣ I ∣ ≤ L , and m 0 I ; x 0 ( 1 ) ∈ F 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{0}\operatorname{Hom}_{\mathcal{A}_{i_{M}}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) if L < ∣ I ∣ L<|I| L < ∣ I ∣ .
(v) N ( A ∞ f i Cat R , K u n ) , N ( A ∞ f i Cat R , K ) \mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K}),\mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R,K}) N ( A ∞ fi Cat R , K un ) , N ( A ∞ fi Cat R , K ) are the nerves of the ordinary categories A ∞ f i Cat R , K u n , A ∞ f i Cat R , K \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K},\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R,K} A ∞ fi Cat R , K un , A ∞ fi Cat R , K from Definition 5.18, for K = 0 , 1 , ∞ K=0,1,\infty K = 0 , 1 , ∞ .
Now 2-morphisms in A ∞ f i Cat R u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ are filtered A ∞ A_{\infty} A ∞ -natural isomorphisms in the sense of Definition 5.22(c). We define a second diagram of ∞ \infty ∞ -categories:
A ∞ w f Cat R u n , ∞ = A ∞ w f Cat R , 0 , 2 u n , ∞ A ∞ w f Cat R , 0 , 3 u n , ∞ = A ˙ ∞ w f Cat R , 0 , 3 u n , ∞ ⟶ A ∞ w f Cat R , 0 , 3 ∞ A ∞ w f Cat R , K ′ , L ′ u n , ∞ = A ˙ ∞ w f Cat R , K ′ , L ′ u n , ∞ ⟶ A ∞ w f Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) A ∞ w f Cat R , K , L u n , ∞ = A ˙ ∞ w f Cat R , K , L u n , ∞ ⟶ A ∞ w f Cat R , K , L ∞ ( 0 ≤ K ≤ L , L ≥ 3 ) A ∞ w f Cat R , K , ∞ u n , ∞ = A ˙ ∞ w f Cat R , K , ∞ u n , ∞ ⟶ A ∞ w f Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) N ( A ∞ f i Cat R , K u n ) = N ( A ∞ f i Cat R , K u n ) ⟶ N ( A ∞ f i Cat R , K ) (225) \begin{aligned}
\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}
&=\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,2}\\
\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,3}
&=\dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,0,3}
\longrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,0,3}\\
\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K',L'}
&=\dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K',L'}
\longrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K',L'}
\qquad(0\le K'\le L',\ L'\ge3)\\
\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L}
&=\dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L}
\longrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}
\qquad(0\le K\le L,\ L\ge3)\\
\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty}
&=\dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty}
\longrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,\infty}
\qquad(K=0,1,\infty)\\
\mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K})
&=\mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K})
\longrightarrow\mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R,K}) \tag*{(225)}
\end{aligned} A ∞ wf Cat R un , ∞ A ∞ wf Cat R , 0 , 3 un , ∞ A ∞ wf Cat R , K ′ , L ′ un , ∞ A ∞ wf Cat R , K , L un , ∞ A ∞ wf Cat R , K , ∞ un , ∞ N ( A ∞ fi Cat R , K un ) = A ∞ wf Cat R , 0 , 2 un , ∞ = A ˙ ∞ wf Cat R , 0 , 3 un , ∞ ⟶ A ∞ wf Cat R , 0 , 3 ∞ = A ˙ ∞ wf Cat R , K ′ , L ′ un , ∞ ⟶ A ∞ wf Cat R , K ′ , L ′ ∞ ( 0 ≤ K ′ ≤ L ′ , L ′ ≥ 3 ) = A ˙ ∞ wf Cat R , K , L un , ∞ ⟶ A ∞ wf Cat R , K , L ∞ ( 0 ≤ K ≤ L , L ≥ 3 ) = A ˙ ∞ wf Cat R , K , ∞ un , ∞ ⟶ A ∞ wf Cat R , K , ∞ ∞ ( K = 0 , 1 , ∞ ) = N ( A ∞ fi Cat R , K un ) ⟶ N ( A ∞ fi Cat R , K ) (225)
such that 2-morphisms in A ∞ w f Cat R u n , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ are weakly filtered A ∞ A_{\infty} A ∞ -natural isomorphisms in the sense of Definition 5.22(d). There are inclusions of ∞ \infty ∞ -categories A ∞ f i Cat R , K , L ∗ , ∞ ↪ A ∞ w f Cat R , K , L ∗ , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K,L}\hookrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K,L} A ∞ fi Cat R , K , L ∗ , ∞ ↪ A ∞ wf Cat R , K , L ∗ , ∞ . The difference with the ∞ \infty ∞ -categories in (224) is that (iv) above should be replaced by:
(iv)′ ^{\prime} ′ We have subscripts ‘R , K , L R,K,L R , K , L ’ for K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\le L K ≤ L , and L ≥ 2 L\ge2 L ≥ 2 for the first column and L ≥ 3 L\ge3 L ≥ 3 for the second and third. We call ( K , L ) (K,L) ( K , L ) the type of the ∞ \infty ∞ -category, generalizing type K K K in Chapters 6–7. We have inclusions A ∞ w f Cat R , K , L ∗ , ∞ ↪ A ∞ w f Cat R , K ′ , L ′ ∗ , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K,L}\hookrightarrow\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{*,\infty}_{R,K',L'} A ∞ wf Cat R , K , L ∗ , ∞ ↪ A ∞ wf Cat R , K ′ , L ′ ∗ , ∞ if K ≥ K ′ K\ge K' K ≥ K ′ and L ≥ L ′ L\ge L' L ≥ L ′ .
The N N N -simplices in the ∞ \infty ∞ -categories in (225) all involve families of maps m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k for ∅ ≠ I ⊆ { 0 , … , N } \varnothing\ne I\subseteq\{0,\ldots,N\} ∅ = I ⊆ { 0 , … , N } , k ≥ 0 k\ge0 k ≥ 0 and x 0 , … , x k x_{0},\ldots,x_{k} x 0 , … , x k in Obj A min I \operatorname{Obj}\mathcal{A}_{\min I} Obj A m i n I , where m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k is filtration-preserving if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 , and m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k decreases filtrations by a bounded amount ϵ ∈ Γ \epsilon\in\Gamma ϵ ∈ Γ if ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 .
For type ( K , L ) (K,L) ( K , L ) and ∣ I ∣ ≠ 3 |I|\ne3 ∣ I ∣ = 3 , the rule is that m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) ∈ F γ > 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{\gamma>0}\operatorname{Hom}_{\mathcal{A}_{i_{M}}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F γ > 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) if K < ∣ I ∣ ≤ L K<|I|\le L K < ∣ I ∣ ≤ L , and m 0 I ; x 0 ( 1 ) ∈ Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in\operatorname{Hom}_{\mathcal{A}_{i_{M}}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) if L < ∣ I ∣ L<|I| L < ∣ I ∣ .
The objects (0-simplices) and morphisms (1-simplices) in (224) and (225) are the same, the two differ in N N N -simplices for N ≥ 2 N\ge2 N ≥ 2 .
In brief, in (8.1) all maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k must be filtration-preserving, but in (8.2) the m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k can decrease filtrations for ∣ I ∣ > L ≥ 2 |I| > L \ge2 ∣ I ∣ > L ≥ 2 . See Remark 5.23(a) for an application in Symplectic Geometry which needs weakly filtered A ∞ A_{\infty} A ∞ -natural isomorphisms, and so should be carried out in the ∞ \infty ∞ -category A ∞ w f Cat R u n , ∞ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ .
The ∞ \infty ∞ -categories A ∞ f i Cat R u n , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , A ∞ f i Cat R , K , L u n , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ , A ∞ w f Cat R u n , ∞ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ , A ∞ w f Cat R , K , L u n , ∞ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ wf Cat R , K , L un , ∞
This section is the analogue of §7.2. Here is a change in the filtered case.
Remark 8.1. In § 6.2 and § 7.1 we defined balanced { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled trees Γ \Gamma Γ in Definition 6.7, and then in Definitions 6.11 and 7.2, given collections of R R R -linear maps m k I m^{I}_{k} m k I or m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k , we defined operators d n , k I d^{I}_{n,k} d n , k I or d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n as sums over balanced Γ 1 , … , Γ k \Gamma_{1},\ldots,\Gamma_{k} Γ 1 , … , Γ k of multiple compositions of m l J m^{J}_{l} m l J or m l J ; x j 0 , … , x j l m^{J;x_{j_{0}},\ldots,x_{j_{l}}}_{l} m l J ; x j 0 , … , x j l .
Definition 6.7(a),(b) require that for Γ \Gamma Γ to be balanced, every internal vertex v v v in Γ \Gamma Γ should have ∣ I v ∣ ≥ 2 |I_{v}| \ge2 ∣ I v ∣ ≥ 2 , but every bud b b b in Γ \Gamma Γ should have ∣ I b ∣ ≥ 3 |I_{b}| \ge3 ∣ I b ∣ ≥ 3 . Now a bud is just an internal vertex with no incoming edges, corresponding to an operator m 0 I v m^{I_{v}}_{0} m 0 I v or m 0 I v ; x m^{I_{v};x}_{0} m 0 I v ; x . In Chapters 6 and 7 we required that m 0 I v = m 0 I v ; x = 0 m^{I_{v}}_{0}=m^{I_{v};x}_{0}=0 m 0 I v = m 0 I v ; x = 0 if ∣ I v ∣ = 1 |I_{v}|=1 ∣ I v ∣ = 1 or 2 2 2 . This is the reason for the condition that ∣ I b ∣ ≥ 3 |I_{b}|\ge3 ∣ I b ∣ ≥ 3 in Definition 6.7(b), as if ∣ I b ∣ = 2 |I_{b}|=2 ∣ I b ∣ = 2 then m 0 I b = m 0 I b ; x = 0 m^{I_{b}}_{0}=m^{I_{b};x}_{0}=0 m 0 I b = m 0 I b ; x = 0 .
In the filtered case we no longer require m 0 I v = m 0 I v ; x = 0 m^{I_{v}}_{0}=m^{I_{v};x}_{0}=0 m 0 I v = m 0 I v ; x = 0 if ∣ I v ∣ = 1 |I_{v}|=1 ∣ I v ∣ = 1 or 2 2 2 , but only that m 0 I v ∈ F ≻ 0 A max I v m^{I_{v}}_{0}\in F^{\succ0}A^{\max I_{v}} m 0 I v ∈ F ≻ 0 A m a x I v or m 0 I v ; x ∈ F ≻ 0 H max I v ( f I v ( x ) , f min I v max I v ( x ) ) m^{I_{v};x}_{0}\in F^{\succ0}\mathcal{H}^{\max I_{v}}(f^{I_{v}}(x),f^{\min I_{v}\max I_{v}}(x)) m 0 I v ; x ∈ F ≻ 0 H m a x I v ( f I v ( x ) , f m i n I v m a x I v ( x )) . Therefore we must relax the definition of balanced, to allow ∣ I b ∣ ≥ 2 |I_{b}|\ge2 ∣ I b ∣ ≥ 2 for buds b b b instead of ∣ I b ∣ ≥ 3 |I_{b}|\ge3 ∣ I b ∣ ≥ 3 . We call such { 0 , … , N } \{0,\ldots,N\} { 0 , … , N } -labelled trees filtered balanced .
In the filtered case, we should define operators d n , k I d^{I}_{n,k} d n , k I or d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n as in Definitions 6.11 and 7.2, but using filtered balanced labelled trees Γ 1 , … , Γ k \Gamma_{1},\ldots,\Gamma_{k} Γ 1 , … , Γ k .
The proofs of Lemmas 6.14 and 7.5 use the condition that ∣ I b ∣ ≥ 3 |I_{b}|\ge3 ∣ I b ∣ ≥ 3 for buds b b b in Γ \Gamma Γ , so these lemmas no longer hold in the filtered case. Instead, we should replace them with Lemma 8.2 below.
Because of the failure of Lemmas 6.14 and 7.5, some upper limits in sums such as k = n + M − 1 k=n+M-1 k = n + M − 1 in (6.18), (6.23), (7.16), (7.21) and k = max ( n , n + j − 1 ) k=\max(n,n+j-1) k = max ( n , n + j − 1 ) in (6.28), (6.30), (7.25), (7.27) are no longer correct in the filtered case. Instead, we should sum these up to k = ∞ k=\infty k = ∞ , to give an infinite sum which is convergent in the filtered sense, as explained in Definition 8.3. Also limits k = n k=n k = n in (6.18), (7.15), (7.16) should be replaced by k = n + 1 k=n+1 k = n + 1 , to allow for m 0 i 0 ≠ 0 m^{i_{0}}_{0}\ne0 m 0 i 0 = 0 or m 0 i 0 ; x 0 ≠ 0 m^{i_{0};x_{0}}_{0}\ne0 m 0 i 0 ; x 0 = 0 .
Lemma 8.2. (a) Suppose we define operators d n , k I d^{I}_{n,k} d n , k I using maps m l J m^{J}_{l} m l J as in Definition 6.11, but no longer requiring m 0 I = 0 m^{I}_{0}=0 m 0 I = 0 for ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 , and using filtered balanced Γ i \Gamma_{i} Γ i rather than balanced Γ i \Gamma_{i} Γ i in (6.11). If $|I|=1 then d n , k I = 0 d^{I}_{n,k}=0 d n , k I = 0 unless k ≤ n + 1 k\le n+1 k ≤ n + 1 . If ∣ I ∣ ≥ 2 |I|\ge2 ∣ I ∣ ≥ 2 and k > n + ∣ I ∣ − 2 k>n+|I|-2 k > n + ∣ I ∣ − 2 then every term in (6.11) involves at least k − n − ∣ I ∣ + 2 k-n-|I|+2 k − n − ∣ I ∣ + 2 factors m 0 I b m^{I_{b}}_{0} m 0 I b with $|I_{b}|=2.
(b) The analogue of (a) holds for operators d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n as in Definition 7.2, with at least k − n − ∣ I ∣ + 2 k-n-|I|+2 k − n − ∣ I ∣ + 2 factors m 0 I b ; x m^{I_{b};x}_{0} m 0 I b ; x with $|I_{b}|=2.
Here is the filtered analogue of Definition 7.7, used to define (8.1).
Definition 8.3. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring, as in Definitions 5.4 and 5.6. We will define a simplicial set A ^ ∞ f i Cat R u n , ∞ \widehat{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ of unital Γ \Gamma Γ -filtered A ∞ A_{\infty} A ∞ -categories over R R R .
(a) For N ≥ 0 N \ge0 N ≥ 0 , define the N N N -simplices ( A ^ ∞ f i Cat R u n , ∞ ) N (\widehat{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R})_{N} ( A ∞ fi Cat R un , ∞ ) N in A ^ ∞ f i Cat R u n , ∞ \widehat{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ to be the set of quadruples ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , where:
(i) O i \mathcal{O}^{i} O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , thought of as the set of objects Obj A i \operatorname{Obj}\mathcal{A}^{i} Obj A i of a filtered A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i at vertex i i i of the N N N -simplex Δ N \Delta^{N} Δ N .
(ii) f i j : O i → O j f^{ij}\colon\mathcal{O}^{i}\to\mathcal{O}^{j} f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\le i\le j\le N 0 ≤ i ≤ j ≤ N , thought of as the map f Obj i j : Obj A i → Obj A j f^{ij}_{\operatorname{Obj}}\colon\operatorname{Obj}\mathcal{A}^{i}\to\operatorname{Obj}\mathcal{A}^{j} f Obj ij : Obj A i → Obj A j in a filtered A ∞ A_{\infty} A ∞ -functor F i j : A i → A j \mathcal{F}^{ij}\colon\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j living on the edge [ i , j ] [i,j] [ i , j ] of Δ N \Delta^{N} Δ N . We require these maps to satisfy f i i = id O i f^{ii}=\operatorname{id}_{\mathcal{O}^{i}} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N . We do not require that f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik .
Suppose ∅ ≠ I ⊆ { 0 , 1 , … , N } \varnothing\ne I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } with I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M . Define f I : O i 0 → O i M f^{I}\colon\mathcal{O}^{i_{0}}\to\mathcal{O}^{i_{M}} f I : O i 0 → O i M as in (194) .
(iii) For all i = 0 , 1 , … , N i=0,1,\ldots,N i = 0 , 1 , … , N and x , y ∈ O i x,y\in\mathcal{O}^{i} x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded Γ \Gamma Γ -filtered R R R -module
H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d , \mathcal{H}^{i}(x,y)=\bigoplus_{d\in\mathbb{Z}}\mathcal{H}^{i}(x,y)^{d}, H i ( x , y ) = d ∈ Z ⨁ H i ( x , y ) d , thought of as the Hom space Hom A i ( x , y ) \operatorname{Hom}_{\mathcal{A}^{i}}(x,y) Hom A i ( x , y ) in the A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i .
We require that H i ( x , y ) d \mathcal{H}^{i}(x,y)^{d} H i ( x , y ) d is a complete filtered R R R -module for each d ∈ Z d\in\mathbb{Z} d ∈ Z (this is called degree-wise complete ), but we do not require H i ( x , y ) \mathcal{H}^{i}(x,y) H i ( x , y ) to be complete. We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^{i}=(\mathcal{H}^{i}(x,y))_{x,y\in\mathcal{O}^{i}} H i = ( H i ( x , y ) ) x , y ∈ O i .
(iv) For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M\ge0 M ≥ 0 , and for all k ≥ 0 k\ge0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}} x 0 , … , x k ∈ O i 0 , thought of as objects in A i 0 \mathcal{A}^{i_{0}} A i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) (226) m_{k}^{I;x_{0},\ldots,x_{k}}\colon\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{i_{0}}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{i_{0}}(x_{k-1},x_{k})
\longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k})) \tag*{(226)} m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x k )) (226) graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k . Here when k = 0 k=0 k = 0 , by convention the empty tensor product on the top line of (226) is R R R , so m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 maps
m 0 I ; x 0 : R ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M . m_{0}^{I;x_{0}}\colon R\longrightarrow\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0}))^{2-M}. m 0 I ; x 0 : R ⟶ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M . We sometimes identify m 0 I ; x 0 m_{0}^{I;x_{0}} m 0 I ; x 0 with m 0 I ; x 0 ( 1 ) ∈ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M m_{0}^{I;x_{0}}(1)\in\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0}))^{2-M} m 0 I ; x 0 ( 1 ) ∈ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) 2 − M .
We require m k I ; x 0 , … , x k m_{k}^{I;x_{0},\ldots,x_{k}} m k I ; x 0 , … , x k to be filtration-preserving for all k ≥ 1 k\ge1 k ≥ 1 and x 0 , … , x k x_{0},\ldots,x_{k} x 0 , … , x k in O i 0 \mathcal{O}^{i_{0}} O i 0 , that is, for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ we have
m k x 0 , … , x k ( F γ ( H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x k − 1 , x k ) ) ) ⊆ F γ ( H i M ( f I ( x 0 ) , f i 0 i M ( x k ) ) ) . m_{k}^{x_{0},\ldots,x_{k}}\left(F^{\gamma}\left(\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes\cdots\otimes\mathcal{H}^{i_{0}}(x_{k-1},x_{k})\right)\right)
\subseteq F^{\gamma}\left(\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{k}))\right). m k x 0 , … , x k ( F γ ( H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x k − 1 , x k ) ) ) ⊆ F γ ( H i M ( f I ( x 0 ) , f i 0 i M ( x k )) ) . We require that m 0 I ; x 0 ( 1 ) ∈ F > 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{>0}\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F > 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) for all x 0 ∈ O i 0 x_{0}\in\mathcal{O}^{i_{0}} x 0 ∈ O i 0 if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 , and m 0 I ; x 0 ( 1 ) ∈ F 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{0}\mathcal{H}^{i_{M}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) for all x 0 ∈ O i 0 x_{0}\in\mathcal{O}^{i_{0}} x 0 ∈ O i 0 if ∣ I ∣ ≥ 3 |I|\ge3 ∣ I ∣ ≥ 3 .
We write m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 m^{I}_{k}=(m^{I;x_{0},\ldots,x_{k}}_{k})_{x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}}} m k I = ( m k I ; x 0 , … , x k ) x 0 , … , x k ∈ O i 0 and m I = ( m k I ) k ≥ 0 m^{I}=(m^{I}_{k})_{k\geq0} m I = ( m k I ) k ≥ 0 .
We require these maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k to satisfy the following identities generalizing (207) –(208) for all I = { i 0 < i 1 < … < i M } ⊆ { 0 , 1 , … , N } I=\{i_{0}<i_{1}<\ldots<i_{M}\}\subseteq\{0,1,\ldots,N\} I = { i 0 < i 1 < … < i M } ⊆ { 0 , 1 , … , N } , n ≥ 0 n\geq0 n ≥ 0 , and x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 , firstly if M = 0 M=0 M = 0
0 = ∑ k = 1 , … , n + 1 y 0 , … , y k ∈ O i 0 m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n . (227) 0=\sum_{\substack{k=1,\ldots,n+1\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m^{\{i_{0}\};y_{0},\ldots,y_{k}}_{k}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}. \tag*{(227)} 0 = k = 1 , … , n + 1 y 0 , … , y k ∈ O i 0 ∑ m k { i 0 } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n . (227)
and secondly if M > 0 M>0 M > 0
0 = ∑ k = 1 , … , n + 1 y 0 , … , y k ∈ O i 0 m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M ∑ k ≥ 0 , y 0 , … , y k ∈ O i M m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (228) \begin{aligned}
0={}&\sum_{\substack{k=1,\ldots,n+1\\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{0}}}}m^{I;y_{0},\ldots,y_{k}}_{k}\circ d^{\{i_{0}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}\\
&+(-1)^{M}\sum_{\substack{k\geq0,\ y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}}}}m^{\{i_{M}\};y_{0},\ldots,y_{k}}_{k}\circ d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}. \tag*{(228)}
\end{aligned} 0 = k = 1 , … , n + 1 y 0 , … , y k ∈ O i 0 ∑ m k I ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n + ( − 1 ) M k ≥ 0 , y 0 , … , y k ∈ O i M ∑ m k { i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k I ; x 0 , … , x n . (228)
Here d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n is defined using the m l J ; w 0 , … , w l m^{J;w_{0},\ldots,w_{l}}_{l} m l J ; w 0 , … , w l as in Definition 7.2, but modified as in Remark 8.1 to sum over filtered balanced trees Γ \Gamma Γ in (201) .
The second line of (228) is an infinite sum: for each value of k k k , there are only finitely many possible y 0 , … , y k y_{0},\ldots,y_{k} y 0 , … , y k with d n , k ; y 0 , … , y k I ; x 0 , … , x n ≠ 0 d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}\ne0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 , but we can have d n , k ; y 0 , … , y k I ; x 0 , … , x n ≠ 0 d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}\ne0 d n , k ; y 0 , … , y k I ; x 0 , … , x n = 0 for infinitely many k ≫ 0 k\gg0 k ≫ 0 . We claim that it converges in the filtered sense.
To justify this, note that by Lemma 8.2(b), if k > n + M − 1 k>n+M-1 k > n + M − 1 then d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n contains at least k − n − M + 1 k-n-M+1 k − n − M + 1 factors m 0 I b ; x m^{I_{b};x}_{0} m 0 I b ; x with ∣ I b ∣ = 2 \lvert I_{b}\rvert=2 ∣ I b ∣ = 2 . Here I b = { i p , i q } I_{b}=\{i_{p},i_{q}\} I b = { i p , i q } for 0 ≤ p < q ≤ M 0\leq p<q\leq M 0 ≤ p < q ≤ M and x = f P ( x j ) x=f^{P}(x_{j}) x = f P ( x j ) for j = 0 , … , n j=0,\ldots,n j = 0 , … , n and P ⊆ { i 0 , … , i p } P\subseteq\{i_{0},\ldots,i_{p}\} P ⊆ { i 0 , … , i p } is the prefix, with i 0 , i p ∈ P i_{0},i_{p}\in P i 0 , i p ∈ P . Thus there are only finitely many possibilities for m 0 I b ; x m^{I_{b};x}_{0} m 0 I b ; x , and each lies in F ≽ 0 H i b ( f I b ( x ) , f I b ( x ) ) F^{\succcurlyeq0}\mathcal{H}^{i_{b}}(f^{I_{b}}(x),f^{I_{b}}(x)) F ≽ 0 H i b ( f I b ( x ) , f I b ( x )) , so there exists δ ≫ 0 \delta\gg0 δ ≫ 0 in Γ \Gamma Γ with m 0 I b ; x ∈ F δ H i b ( f I b ( x ) , f I b ( x ) ) m^{I_{b};x}_{0}\in F^{\delta}\mathcal{H}^{i_{b}}(f^{I_{b}}(x),f^{I_{b}}(x)) m 0 I b ; x ∈ F δ H i b ( f I b ( x ) , f I b ( x )) .
By Definition 5.4(b) we can take the greatest lower bound ϵ ∈ Γ \epsilon\in\Gamma ϵ ∈ Γ of these δ ∈ Γ \delta\in\Gamma δ ∈ Γ over all possibilities for m 0 I b ; x m^{I_{b};x}_{0} m 0 I b ; x , and by Definition 5.4(d) we have ϵ ≫ 0 \epsilon\gg0 ϵ ≫ 0 . That is, there exists a single ϵ ≫ 0 \epsilon\gg0 ϵ ≫ 0 in Γ \Gamma Γ such that for all choices of m 0 I b ; x m^{I_{b};x}_{0} m 0 I b ; x we have m 0 I b ; x ∈ F ϵ H i b ( f I b ( x ) , f I b ( x ) ) m^{I_{b};x}_{0}\in F^{\epsilon}\mathcal{H}^{i_{b}}(f^{I_{b}}(x),f^{I_{b}}(x)) m 0 I b ; x ∈ F ϵ H i b ( f I b ( x ) , f I b ( x )) . Therefore, if k > n + M − 1 k>n+M-1 k > n + M − 1 then d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n in (228) increases filtrations by at least ( k − n − M + 1 ) ϵ (k-n-M+1)\epsilon ( k − n − M + 1 ) ϵ , that is, it maps F γ ( ⋯ ) → F γ + ( k − n − M + 1 ) ϵ ( ⋯ ) F^{\gamma}(\cdots)\to F^{\gamma+(k-n-M+1)\epsilon}(\cdots) F γ ( ⋯ ) → F γ + ( k − n − M + 1 ) ϵ ( ⋯ ) . Since γ + ( k − n − M + 1 ) ϵ → ∞ \gamma+(k-n-M+1)\epsilon\to\infty γ + ( k − n − M + 1 ) ϵ → ∞ in Γ \Gamma Γ as k → ∞ k\to\infty k → ∞ , it follows that when applied to any element of H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x n − 1 , x n ) \mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes\cdots\otimes\mathcal{H}^{i_{0}}(x_{n-1},x_{n}) H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x n − 1 , x n ) , equation (228) is a convergent sum.
(v) There should exist elements 1 x i ∈ F 0 H i ( x , x ) 1^{i}_{x}\in F^{0}\mathcal{H}^{i}(x,x) 1 x i ∈ F 0 H i ( x , x ) for each i = 0 , … , N i=0,\ldots,N i = 0 , … , N and x ∈ O i x\in\mathcal{O}^{i} x ∈ O i satisfying (209) –(211) . They are uniquely determined by (209) .
(b) If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , we define
A ^ ∞ f i C a t R ∞ ( α ) : ( A ^ ∞ f i C a t R ∞ ) N ⟶ A ^ ∞ f i C a t R , N ‾ ∞ \widehat{\mathcal{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\infty}_{R}(\alpha):\left(\widehat{\mathcal{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\infty}_{R}\right)_{N}\longrightarrow\widehat{\mathcal{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\infty}_{R,\overline{N}} A ∞ fi Cat R ∞ ( α ) : ( A ∞ fi Cat R ∞ ) N ⟶ A ∞ fi Cat R , N ∞
as in Definition 7.7(b). Essentially the same proof as Proposition 7.9 shows that A ^ ∞ f i Cat R u n , ∞ ( α ) \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}(\alpha) A ∞ fi Cat R un , ∞ ( α ) maps to A ^ ∞ f i Cat R , N ‾ u n , ∞ \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,\overline{N}} A ∞ fi Cat R , N un , ∞ . As in Definition 7.7(b), A ^ ∞ f i Cat R u n , ∞ : Δ o p → S e t \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\colon\Delta^{\mathrm{op}}\to\mathrm{Set} A ∞ fi Cat R un , ∞ : Δ op → Set is a functor, that is, A ^ ∞ f i Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ is a simplicial set.
We also give an analogue with weakly filtered N N N -morphisms for N ≥ 2 N\geq2 N ≥ 2 , used to define (8.2).
Definition 8.4. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring, as in Definitions 5.4 and 5.6. We will define a simplicial set A ^ ∞ w f Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ of unital Γ \Gamma Γ -filtered A ∞ A_{\infty} A ∞ -categories over R R R .
For N ≥ 0 N\geq0 N ≥ 0 , the N N N -simplices ( A ^ ∞ w f Cat R u n , ∞ ) N (\widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R})_N ( A ∞ wf Cat R un , ∞ ) N in A ^ ∞ w f Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ are defined as for ( A ^ ∞ f i Cat R u n , ∞ ) N (\widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R})_N ( A ∞ fi Cat R un , ∞ ) N in Definition 8.3(a), but with the following changes to (a)(iv): there should exist δ ⪯ 0 \delta\preceq0 δ ⪯ 0 in Γ \Gamma Γ independent of I I I , k k k , x 0 , … , x k x_0,\ldots,x_k x 0 , … , x k such that the following holds.
The R R R -linear maps m k I ; x 0 , … , x k m^{I;x_0,\ldots,x_k}_k m k I ; x 0 , … , x k in (8.3) should be filtration-preserving for ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 and k ≥ 1 k\geq1 k ≥ 1 , but for ∣ I ∣ ≥ 3 |I|\geq3 ∣ I ∣ ≥ 3 and k ≥ 1 k\geq1 k ≥ 1 they need not preserve filtrations, but should instead map F γ ( ⋯ ) → F γ + δ ( ⋯ ) F^\gamma(\cdots)\to F^{\gamma+\delta}(\cdots) F γ ( ⋯ ) → F γ + δ ( ⋯ ) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
We require that m 1 I ; x 0 ( 1 ) ∈ F 0 H i M ( f i ( x 0 ) , f i 0 i M ( x 0 ) ) m^{I;x_0}_1(1)\in F^0\mathcal{H}^{i_M}(f^i(x_0),f^{i_0i_M}(x_0)) m 1 I ; x 0 ( 1 ) ∈ F 0 H i M ( f i ( x 0 ) , f i 0 i M ( x 0 )) for all x 0 ∈ O i 0 x_0\in\mathcal{O}^{i_0} x 0 ∈ O i 0 if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 , and m 1 I ; x 0 ( 1 ) ∈ F δ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m^{I;x_0}_1(1)\in F^\delta\mathcal{H}^{i_M}(f^I(x_0),f^{i_0i_M}(x_0)) m 1 I ; x 0 ( 1 ) ∈ F δ H i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) for all x 0 ∈ O i 0 x_0\in\mathcal{O}^{i_0} x 0 ∈ O i 0 if ∣ I ∣ ≥ 3 |I|\geq3 ∣ I ∣ ≥ 3 (this replaces m 1 I ; x 0 ( 1 ) ∈ F 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m^{I;x_0}_1(1)\in F^0\mathcal{H}^{i_M}(f^I(x_0),f^{i_0i_M}(x_0)) m 1 I ; x 0 ( 1 ) ∈ F 0 H i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) ).
Each term in d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k I ; x 0 , … , x n contains at most M − 1 M-1 M − 1 factors m l J ; x 0 ′ , … , x l ′ m^{J;x'_0,\ldots,x'_l}_l m l J ; x 0 ′ , … , x l ′ with ∣ J ∣ ≥ 3 |J|\geq3 ∣ J ∣ ≥ 3 . Thus the argument in Definition 8.3(a)(iv) that (8.5) converges in the filtered sense works here, but we should include an extra shift in filtrations F γ ( ⋯ ) ↦ F γ + ( M − 1 ) δ ( ⋯ ) F^\gamma(\cdots)\mapsto F^{\gamma+(M-1)\delta}(\cdots) F γ ( ⋯ ) ↦ F γ + ( M − 1 ) δ ( ⋯ ) when M ≥ 2 M\geq2 M ≥ 2 , and conclude that the infinite sums converge because γ + ( M − 1 ) δ + ( k − n − M + 1 ) ϵ → ∞ \gamma+(M-1)\delta+(k-n-M+1)\epsilon\to\infty γ + ( M − 1 ) δ + ( k − n − M + 1 ) ϵ → ∞ in Γ \Gamma Γ as k → ∞ k\to\infty k → ∞ .
The rest of the definition of A ^ ∞ w f Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ is as in Definition 8.3. Note that A ^ ∞ f i Cat R u n , ∞ ⊂ A ^ ∞ w f Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ ⊂ A ∞ wf Cat R un , ∞ .
The following analogue of Theorem 7.10 will be proved in §10.3.
Theorem 8.5. Let ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \bigl((\mathcal{O}^i)_{i=0}^N,(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^i)_{i=0}^N,(m^I)_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\bigr) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) be an N N N -simplex in A ^ ∞ f i Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ or A ^ ∞ w f Cat R u n , ∞ \widehat{\mathsf{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ as in Definitions 8.3 and 8.4, and define maps d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k} d n , k ; y 0 , … , y k I ; x 0 , … , x n from the m k I ; x 0 , … , x k m^{I;x_0,\ldots,x_k}_k m k I ; x 0 , … , x k as in Remark 8.1. Then for each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , written I = { i 0 , i 1 , … , i M } I=\{i_0,i_1,\ldots,i_M\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_0<i_1<\cdots<i_M i 0 < i 1 < ⋯ < i M , and all k , n ≥ 0 k,n\geq0 k , n ≥ 0 , x 0 , … , x n ∈ O i 0 x_0,\ldots,x_n\in\mathcal{O}^{i_0} x 0 , … , x n ∈ O i 0 and y 0 , … , y k ∈ O i M y_0,\ldots,y_k\in\mathcal{O}^{i_M} y 0 , … , y k ∈ O i M , in R R R -linear maps as in (7.6), generalizing (7.21) we have
∑ k − 1 ≤ l ≤ n + 1 w 0 , … , w l ∈ O i 0 d l , k ; y 0 , … , y k { i 0 } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n = 0 if M = 0 , \sum_{\substack{k-1\leq l\leq n+1\\w_0,\ldots,w_l\in\mathcal{O}^{i_0}}}
d^{\{i_0\};w_0,\ldots,w_l}_{l,k;y_0,\ldots,y_k}\circ d^{\{i_0\};x_0,\ldots,x_n}_{n,l;w_0,\ldots,w_l}=0
\qquad\text{if }M=0, k − 1 ≤ l ≤ n + 1 w 0 , … , w l ∈ O i 0 ∑ d l , k ; y 0 , … , y k { i 0 } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n = 0 if M = 0 , ∑ 1 ≤ l ≤ n + 1 w 0 , … , w l ∈ O i 0 d l , k ; y 0 , … , y k I ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n + ( − 1 ) M ∑ l ≥ 0 w 0 , … , w l ∈ O i M d l , k ; y 0 , … , y k { i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l I ; x 0 , … , x n = 0 if M > 0. (229) \sum_{\substack{1\leq l\leq n+1\\w_0,\ldots,w_l\in\mathcal{O}^{i_0}}}
d^{I;w_0,\ldots,w_l}_{l,k;y_0,\ldots,y_k}\circ d^{\{i_0\};x_0,\ldots,x_n}_{n,l;w_0,\ldots,w_l}
+(-1)^M\sum_{\substack{l\geq0\\w_0,\ldots,w_l\in\mathcal{O}^{i_M}}}
d^{\{i_M\};w_0,\ldots,w_l}_{l,k;y_0,\ldots,y_k}\circ d^{I;x_0,\ldots,x_n}_{n,l;w_0,\ldots,w_l}=0
\qquad\text{if }M>0. \tag*{(229)} 1 ≤ l ≤ n + 1 w 0 , … , w l ∈ O i 0 ∑ d l , k ; y 0 , … , y k I ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 } ; x 0 , … , x n + ( − 1 ) M l ≥ 0 w 0 , … , w l ∈ O i M ∑ d l , k ; y 0 , … , y k { i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l I ; x 0 , … , x n = 0 if M > 0. (229)
where the final sum may have infinitely many nonzero terms, but converges in the filtered sense.
Here is the analogue of Example 7.11.
Example 8.6. (a) A 0-simplex ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) (\mathcal{O}^{0}, f^{00}=\operatorname{id}_{\mathcal{O}^{0}}, \mathcal{H}^{0},m^{\{0\}}) ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) of A ^ ∞ C a t R u n , ∞ \widehat{\mathsf{A}}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to a unital filtered A ∞ A_{\infty} A ∞ -category A 0 \mathcal{A}^{0} A 0 over R R R , as in Definition 5.17, with objects Obj A 0 = O 0 \operatorname{Obj}\mathcal{A}^{0}=\mathcal{O}^{0} Obj A 0 = O 0 , Hom spaces Hom A 0 ( x , y ) = H 0 ( x , y ) \operatorname{Hom}_{\mathcal{A}^{0}}(x,y)=\mathcal{H}^{0}(x,y) Hom A 0 ( x , y ) = H 0 ( x , y ) , morphisms ( m k { 0 } ; x 0 , … , x k ) k ≥ 0 (m_{k}^{\{0\};x_{0},\ldots,x_{k}})_{k\geq0} ( m k { 0 } ; x 0 , … , x k ) k ≥ 0 , and units 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 1_{x}^{0}\in\operatorname{Hom}_{\mathcal{A}^{0}}^{0}(x,x)=\mathcal{H}^{0}(x,x)^{0} 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 .
(b) A 1-simplex of A ^ ∞ C a t R u n , ∞ \widehat{\mathsf{A}}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to a unital filtered A ∞ A_{\infty} A ∞ -functor F 01 : A 0 → A 1 \mathcal{F}^{01}:\mathcal{A}^{0}\to\mathcal{A}^{1} F 01 : A 0 → A 1 of unital filtered A ∞ A_{\infty} A ∞ -categories, as in Definition 5.18, with object map F O b j 01 = f 01 \mathcal{F}_{\mathrm{Obj}}^{01}=f^{01} F Obj 01 = f 01 and morphisms F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k \mathcal{F}_{k}^{01;x_{0},\ldots,x_{k}}=m_{k}^{\{0,1\};x_{0},\ldots,x_{k}} F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k .
(c) A 2-simplex of A ^ ∞ C a t R u n , ∞ \widehat{\mathsf{A}}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ is equivalent to unital filtered A ∞ A_{\infty} A ∞ -categories A 0 , A 1 , A 2 \mathcal{A}^{0},\mathcal{A}^{1},\mathcal{A}^{2} A 0 , A 1 , A 2 over R R R as in (a), and unital filtered A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), and R R R -linear maps for all k ≥ 0 k\geq0 k ≥ 0 and x 0 , … , x k ∈ O 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{0} x 0 , … , x k ∈ O 0
m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 12 ∘ f 01 ( x 0 ) , f 02 ( x k ) ) , m_{k}^{\{0,1,2\};x_{0},\ldots,x_{k}}:\mathcal{H}^{0}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{0}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{0}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{2}(f^{12}\circ f^{01}(x_{0}),f^{02}(x_{k})), m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 12 ∘ f 01 ( x 0 ) , f 02 ( x k )) , graded of degree − k -k − k , which are filtration-preserving, satisfying (7.19) and for all n ≥ 0 n\geq0 n ≥ 0 and x 0 , … , x n ∈ O 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{0} x 0 , … , x n ∈ O 0 , a generalization of (7.22)
\begin{aligned}
&\sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{\{0,1,2\};x_{0},x_{1},\ldots,x_{i},x_{i+j},x_{i+j+1},\ldots,x_{n}}\\
&\qquad\circ\left(\operatorname{id}_{\mathcal{H}^{0}(x_{0},x_{1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{i-1},x_{i})}\otimes m_{j}^{\{0\};x_{i},x_{i+1},\ldots,x_{i+j}}\right.\\
&\qquad\qquad\left.\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal{H}^{0}(x_{n-1},x_{n})}\right)\\
&\quad+\sum_{\substack{n=n_{1}+\cdots+n_{k}\\1\leq l\leq k,\ n_{i}\geq0}}(-1)^{k+1}\cdot(-1)^{\sum_{i=1}^{l-1}(n_{i}+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_{i}+1)(k-i)}\\
&\qquad\circ\left[m_{n_{1}}^{\{1,2\};f^{\{0,1,2\}}(x_{p_{0}}),f^{\{0,1,2\}}(x_{p_{1}}),\ldots,f^{\{0,1,2\}}(x_{p_{l-1}}),f^{02}(x_{p_{l}}),\ldots,f^{02}(x_{p_{k}})\right.\\
&\qquad\qquad\left.\circ\left((m_{n_{1}}^{\{1,2\}}\circ m_{n_{1}}^{\{0,1\}})^{x_{p_{0}},\ldots,x_{p_{1}}}\otimes\cdots\otimes(m_{n_{l-1}}^{\{1,2\}}\circ m_{n_{l-1}}^{\{0,1\}})^{x_{p_{l-2}},\ldots,x_{p_{l-1}}}\right.\right.\\
&\qquad\qquad\left.\left.\otimes m_{n_{l}}^{\{0,1,2\};x_{p_{l-1}},\ldots,x_{p_{l}}}\otimes m_{n_{l+1}}^{\{0,2\};x_{p_{l}},\ldots,x_{p_{l+1}}}\otimes\cdots\otimes m_{n_{k}}^{\{0,2\};x_{p_{k-1}},\ldots,x_{p_{k}}}\right)\right]=0.
\end{aligned}} \tag*{(230)}\end{aligned}
Comparing (4.15) and (8.7), we see m { 0 , 1 , 2 } m^{\{0,1,2\}} m { 0 , 1 , 2 } is equivalent to a unital filtered A ∞ A_{\infty} A ∞ -natural transformation F 012 : F 12 ∘ F 01 ⇒ F 02 \mathcal{F}^{012}:\mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} F 012 : F 12 ∘ F 01 ⇒ F 02 , as in Definition 5.22(a).
(d) In the analogues of (a)–(c) for A ^ ∞ C a t R w f , u n , ∞ \widehat{\mathsf{A}}_{\infty}\mathsf{Cat}_{R}^{\mathrm{wf},\mathrm{un},\infty} A ∞ Cat R wf , un , ∞ , (a),(b) are unchanged, but in (c), instead of taking m k { 0 , 1 , 2 } ; x 0 , … , x k m_{k}^{\{0,1,2\};x_{0},\ldots,x_{k}} m k { 0 , 1 , 2 } ; x 0 , … , x k to be filtration-preserving, as in Definition 8.4 there should exist δ ⪯ 0 \delta\preceq0 δ ⪯ 0 in Γ \Gamma Γ independent of k , x 0 , … , x k k,x_{0},\ldots,x_{k} k , x 0 , … , x k such that m k { 0 , 1 , 2 } ; x 0 , … , x k m_{k}^{\{0,1,2\};x_{0},\ldots,x_{k}} m k { 0 , 1 , 2 } ; x 0 , … , x k maps F γ ( ⋯ ) → F γ + δ ( ⋯ ) F^{\gamma}(\cdots)\to F^{\gamma+\delta}(\cdots) F γ ( ⋯ ) → F γ + δ ( ⋯ ) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ . Hence m { 0 , 1 , 2 } m^{\{0,1,2\}} m { 0 , 1 , 2 } is equivalent to a weakly filtered A ∞ A_{\infty} A ∞ -natural transformation F 12 ∘ F 01 ⇒ F 02 \mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} F 12 ∘ F 01 ⇒ F 02 in the sense of Definition 5.22(b).
Here is the analogue of Definition 7.12.
Definition 8.7. Define a simplicial subset A ∞ f i C a t R u n , ∞ ⊂ A ^ ∞ f i C a t R u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} \subset\widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ ⊂ A ∞ fi Cat R un , ∞ , by an N N N -simplex ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ^ ∞ f i C a t R u n , ∞ \widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ lies in A ∞ f i C a t R u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ if and only if whenever 0 ≤ i 0 < i 1 < i 2 ≤ N 0\leq i_{0}<i_{1}<i_{2}\leq N 0 ≤ i 0 < i 1 < i 2 ≤ N , either (hence both) of the following two equivalent conditions hold:
(a) The unital filtered A ∞ A_{\infty} A ∞ -natural transformation m { i 0 , i 1 , i 2 } : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 m^{\{i_{0},i_{1},i_{2}\}}\colon\mathfrak{F}^{i_{1}i_{2}}\circ\mathfrak{F}^{i_{0}i_{1}}\Rightarrow\mathfrak{F}^{i_{0}i_{2}} m { i 0 , i 1 , i 2 } : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 in Example 8.6(c) is a filtered A ∞ A_{\infty} A ∞ -natural isomorphism, in the sense of Definition 5.22(c).
(b) [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 [m^{\{i_{0},i_{1},i_{2}\}}]\colon\mathfrak{F}^{i_{1}i_{2}}\circ\mathfrak{F}^{i_{0}i_{1}}\to\mathfrak{F}^{i_{0}i_{2}} [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 is an isomorphism in the cohomology category H 0 ( F 0 F u n A ∞ f i u n ( A i 0 , A i 2 ) ) H^{0}\left(F^{0}\mathrm{Fun}^{\mathrm{un}}_{\mathrm{A}_{\infty}^{\mathrm{fi}}}(\mathcal{A}^{i_{0}},\mathcal{A}^{i_{2}})\right) H 0 ( F 0 Fun A ∞ fi un ( A i 0 , A i 2 ) ) .
Here (a),(b) are equivalent by Definition 5.22(c). If α : [ N ‾ ] → [ N ] \alpha\colon[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ then A ^ ∞ f i C a t R u n , ∞ ( α ) \widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}(\alpha) A ∞ fi Cat R un , ∞ ( α ) maps ( A ∞ f i C a t R u n , ∞ ) N → ( A ∞ f i C a t R u n , ∞ ) N ‾ \left(\mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\right)_{N}\to\left(\mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\right)_{\overline{N}} ( A ∞ fi Cat R un , ∞ ) N → ( A ∞ fi Cat R un , ∞ ) N , so A ∞ f i C a t R u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ is a simplicial set.
Similarly, define a simplicial subset A ∞ w f C a t R u n , ∞ ⊂ A ^ ∞ w f C a t R u n , ∞ \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ ⊂ A ∞ wf Cat R un , ∞ , by an N N N -simplex ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ^ ∞ w f C a t R u n , ∞ \widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ lies in A ∞ w f C a t R u n , ∞ \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ if and only if whenever 0 ≤ i 0 < i 1 < i 2 ≤ N 0\leq i_{0}<i_{1}<i_{2}\leq N 0 ≤ i 0 < i 1 < i 2 ≤ N , either (hence both) of the following two equivalent conditions hold:
(c) The unital weakly filtered A ∞ A_{\infty} A ∞ -natural transformation m { i 0 , i 1 , i 2 } : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 m^{\{i_{0},i_{1},i_{2}\}}\colon\mathfrak{F}^{i_{1}i_{2}}\circ\mathfrak{F}^{i_{0}i_{1}}\Rightarrow\mathfrak{F}^{i_{0}i_{2}} m { i 0 , i 1 , i 2 } : F i 1 i 2 ∘ F i 0 i 1 ⇒ F i 0 i 2 in Example 8.6(d) is a weakly filtered A ∞ A_{\infty} A ∞ -natural isomorphism, in the sense of Definition 5.22(d).
(d) [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 [m^{\{i_{0},i_{1},i_{2}\}}]\colon\mathfrak{F}^{i_{1}i_{2}}\circ\mathfrak{F}^{i_{0}i_{1}}\to\mathfrak{F}^{i_{0}i_{2}} [ m { i 0 , i 1 , i 2 } ] : F i 1 i 2 ∘ F i 0 i 1 → F i 0 i 2 is an isomorphism in the cohomology category H 0 ( F u n A ∞ f i u n ( A i 0 , A i 2 ) ) H^{0}\left(\mathrm{Fun}^{\mathrm{un}}_{\mathrm{A}_{\infty}^{\mathrm{fi}}}(\mathcal{A}^{i_{0}},\mathcal{A}^{i_{2}})\right) H 0 ( Fun A ∞ fi un ( A i 0 , A i 2 ) ) .
The following analogue of Theorem 7.13, one of our main results, is proved in §10.4. The proof is long and difficult.
Theorem 8.9. A ∞ f i C a t R u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ and A ∞ w f C a t R u n , ∞ \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ above are ∞ \infty ∞ -categories.
The analogues of Definition 7.15 and Proposition 7.16 also work in the (weakly) filtered cases.
Here is the analogue of Theorem 7.17. It will be proved in §10.5.
Theorem 8.10. In the simplicial sets A ^ ∞ f i C a t R u n , ∞ \widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ and A ^ ∞ w f C a t R u n , ∞ \widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ in Definitions 8.3 and 8.4, define a 2-simplex to be thin if the unital (weakly) filtered A ∞ A_{\infty} A ∞ -natural transformation H 012 : F 01 ∘ F 12 ⇒ F 02 \mathcal{H}^{012}\colon\mathfrak{F}^{01}\circ\mathfrak{F}^{12}\Rightarrow\mathfrak{F}^{02} H 012 : F 01 ∘ F 12 ⇒ F 02 in Example 8.6(c) is a (weakly) filtered A ∞ A_{\infty} A ∞ -natural isomorphism, that is, if it is a 2-simplex in A ∞ f i C a t R u n , ∞ ⊂ A ^ ∞ f i C a t R u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ ⊂ A ∞ fi Cat R un , ∞ or A ∞ w f C a t R u n , ∞ ⊂ A ^ ∞ w f C a t R u n , ∞ \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ ⊂ A ∞ wf Cat R un , ∞ .
Write T f i ⊂ A ^ ∞ f i C a t R u n , ∞ ( [ 2 ] ) \mathcal{T}^{\mathrm{fi}}\subset\widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}([2]) T fi ⊂ A ∞ fi Cat R un , ∞ ([ 2 ]) and T w f ⊂ A ^ ∞ w f C a t R u n , ∞ ( [ 2 ] ) \mathcal{T}^{\mathrm{wf}}\subset\widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}([2]) T wf ⊂ A ∞ wf Cat R un , ∞ ([ 2 ]) for the sets of thin 2-simplices. Then ( A ^ ∞ f i C a t R u n , ∞ , T f i ) \left(\widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R},\mathcal{T}^{\mathrm{fi}}\right) ( A ∞ fi Cat R un , ∞ , T fi ) and ( A ^ ∞ w f C a t R u n , ∞ , T w f ) \left(\widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R},\mathcal{T}^{\mathrm{wf}}\right) ( A ∞ wf Cat R un , ∞ , T wf ) are ∞ \infty ∞ -bicategories, as in §2.8. We have A ∞ f i C a t R u n , ∞ = Core ( A ^ ∞ f i C a t R u n , ∞ , T f i ) \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}=\operatorname{Core}\left(\widehat{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R},\mathcal{T}^{\mathrm{fi}}\right) A ∞ fi Cat R un , ∞ = Core ( A ∞ fi Cat R un , ∞ , T fi ) and A ∞ w f C a t R u n , ∞ = Core ( A ^ ∞ w f C a t R u n , ∞ , T w f ) \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}=\operatorname{Core}\left(\widehat{\mathrm{A}}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R},\mathcal{T}^{\mathrm{wf}}\right) A ∞ wf Cat R un , ∞ = Core ( A ∞ wf Cat R un , ∞ , T wf ) , as in Definition 2.41.
Here is the analogue of Definition 7.19.
Definition 8.11. Let K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\le L K ≤ L and L ≥ 2 L\ge2 L ≥ 2 . Define a simplicial subset A ∞ f i C a t R , K , L u n , ∞ ⊂ A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L}\subset\mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R , K , L un , ∞ ⊂ A ∞ fi Cat R un , ∞ such that an N N N -simplex
( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ in Definition 8.3(a) lies in A ∞ f i C a t R , K , L u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ if:
(i) If I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_{0}<i_{1}<\cdots<i_{M}\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } with ∣ I ∣ ≠ 3 |I|\ne3 ∣ I ∣ = 3 and x 0 ∈ O i 0 x_{0}\in\mathcal{O}^{i_{0}} x 0 ∈ O i 0 , we have m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\le K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) ∈ F ▹ 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 ) ) m_{0}^{I;x_{0}}(1)\in F^{\triangleright0}\operatorname{Hom}_{\mathcal{A}^{i_{M}}}(f^{I}(x_{0}),f^{i_{0}i_{M}}(x_{0})) m 0 I ; x 0 ( 1 ) ∈ F ▹ 0 Hom A i M ( f I ( x 0 ) , f i 0 i M ( x 0 )) if K < ∣ I ∣ ≤ L K<|I|\le L K < ∣ I ∣ ≤ L .
(ii) If L ≥ 3 L\ge3 L ≥ 3 then f i k = f j k ∘ f i j f^{ik}=f^{jk}\circ f^{ij} f ik = f j k ∘ f ij for all 0 ≤ i ≤ j ≤ k ≤ N 0\le i\le j\le k\le N 0 ≤ i ≤ j ≤ k ≤ N ,
(iii) If L ≥ 3 L\ge3 L ≥ 3 , I = { i 0 < i 1 < i 2 } ⊆ { 0 , … , N } I=\{i_{0}<i_{1}<i_{2}\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < i 2 } ⊆ { 0 , … , N } and x 0 ∈ O i 0 x_{0}\in\mathcal{O}^{i_{0}} x 0 ∈ O i 0 then m 0 I ; x 0 ( 1 ) = 1 f i 0 i 2 ( x 0 ) i 2 m_{0}^{I;x_{0}}(1)=1_{f^{i_{0}i_{2}}(x_{0})}^{i_{2}} m 0 I ; x 0 ( 1 ) = 1 f i 0 i 2 ( x 0 ) i 2 if K ≥ 3 K\ge3 K ≥ 3 (this makes sense by (ii) as f I ( x 0 ) = f i 0 i 2 ( x 0 ) f^{I}(x_{0})=f^{i_{0}i_{2}}(x_{0}) f I ( x 0 ) = f i 0 i 2 ( x 0 ) ), and m 0 I ; x 0 ( 1 ) − 1 f i 0 i 2 ( x 0 ) i 2 ∈ F ▹ 0 Hom A i 2 ( f i 0 i 2 ( x 0 ) , f i 0 i 2 ( x 0 ) ) m_{0}^{I;x_{0}}(1)-1_{f^{i_{0}i_{2}}(x_{0})}^{i_{2}}\in F^{\triangleright0}\operatorname{Hom}_{\mathcal{A}^{i_{2}}}(f^{i_{0}i_{2}}(x_{0}),f^{i_{0}i_{2}}(x_{0})) m 0 I ; x 0 ( 1 ) − 1 f i 0 i 2 ( x 0 ) i 2 ∈ F ▹ 0 Hom A i 2 ( f i 0 i 2 ( x 0 ) , f i 0 i 2 ( x 0 )) if K ≤ 2 K\le2 K ≤ 2 .
These conditions are preserved by the maps in Definition 8.3(b), and so define a simplicial subset A ∞ f i C a t R , K , L u n , ∞ ⊂ A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L}\subset\mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R , K , L un , ∞ ⊂ A ∞ fi Cat R un , ∞ . When K = 0 K=0 K = 0 and L = 2 L=2 L = 2 part (i) is included in Definition 8.3(a)(iv) and (ii),(iii) are trivial, so A ∞ f i C a t R , 0 , 2 u n , ∞ = A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,0,2}=\mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R , 0 , 2 un , ∞ = A ∞ fi Cat R un , ∞ . Clearly A ∞ f i C a t R , K , L u n , ∞ ⊆ A ∞ f i C a t R , K ′ , L ′ u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L}\subseteq\mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K',L'} A ∞ fi Cat R , K , L un , ∞ ⊆ A ∞ fi Cat R , K ′ , L ′ un , ∞ if K ≥ K ′ K\ge K' K ≥ K ′ and L ≥ L ′ L\ge L' L ≥ L ′ , as in (8.1). Theorem 8.20 will show that A ∞ f i C a t R , K , L u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ is an ∞ \infty ∞ -category. We say that A ∞ f i C a t R , K , L u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ has type ( K , L ) (K,L) ( K , L ) .
We also define ∞ \infty ∞ -subcategories A ∞ w f C a t R , K , L u n , ∞ ⊂ A ∞ w f C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L}\subset\mathop{\mathrm{A}}{}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R , K , L un , ∞ ⊂ A ∞ wf Cat R un , ∞ as in (8.2) exactly as above, by the same (i)–(iii), but applied to N N N -simplices in A ∞ w f C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ , plus the extra condition which will be needed in the proof of Theorem 8.20:
(iv) If L ≥ 3 L\ge3 L ≥ 3 , I = { i 0 < i 1 < i 2 } ⊆ { 0 , … , N } I=\{i_{0}<i_{1}<i_{2}\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < i 2 } ⊆ { 0 , … , N } , n ≥ 1 n\ge1 n ≥ 1 , and x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 , then m n I ; x 0 , … , x n m_{n}^{I;x_{0},\ldots,x_{n}} m n I ; x 0 , … , x n is filtration-preserving.
Example 8.12. If K ≤ 2 K\le2 K ≤ 2 and L ≥ 3 L\ge3 L ≥ 3 then modifying Example 8.6(c), a 2-simplex in A ∞ f i C a t R , K , L u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ consists of unital filtered A ∞ A_{\infty} A ∞ -categories A 0 , A 1 , A 2 \mathcal{A}^{0},\mathcal{A}^{1},\mathcal{A}^{2} A 0 , A 1 , A 2 over R R R (which are strict if K > 0 K>0 K > 0 ) and unital filtered A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (which are strict if K > 1 K>1 K > 1 ), and a filtered A ∞ A_{\infty} A ∞ -natural isomorphism m { 0 , 1 , 2 } : F 12 ∘ F 01 ⇒ F 02 m^{\{0,1,2\}}:\mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} m { 0 , 1 , 2 } : F 12 ∘ F 01 ⇒ F 02 , such that m 0 { 0 , 1 , 2 } ; x 0 ( 1 ) − 1 f 02 ( x 0 ) 2 m_{0}^{\{0,1,2\};x_{0}}(1)-1_{f^{02}(x_{0})}^{2} m 0 { 0 , 1 , 2 } ; x 0 ( 1 ) − 1 f 02 ( x 0 ) 2 lies in F ▹ 0 Hom A 2 ( f 02 ( x 0 ) , f 02 ( x 0 ) ) F^{\triangleright0}\operatorname{Hom}_{\mathcal{A}^{2}}(f^{02}(x_{0}),f^{02}(x_{0})) F ▹ 0 Hom A 2 ( f 02 ( x 0 ) , f 02 ( x 0 )) for all x 0 ∈ O 0 x_{0}\in\mathcal{O}^{0} x 0 ∈ O 0 .
By Proposition 5.24, such m { 0 , 1 , 2 } m^{\{0,1,2\}} m { 0 , 1 , 2 } are in 1-1 correspondence with unital filtered A ∞ A_{\infty} A ∞ -homotopies H : F 12 ∘ F 01 ⇒ F 02 \mathcal{H}:\mathcal{F}^{12}\circ\mathcal{F}^{01}\Rightarrow\mathcal{F}^{02} H : F 12 ∘ F 01 ⇒ F 02 , as in Definition 5.21. Thus, 2-morphisms in A ∞ f i C a t R , 0 , 3 u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,0,3} A ∞ fi Cat R , 0 , 3 un , ∞ correspond to unital filtered A ∞ A_{\infty} A ∞ -homotopies.
Definition 8.13. Apart from the extra data of filtrations, the important difference between the definitions of A ∞ C a t R u n , ∞ \mathop{\mathrm{A}}{}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ in Definitions 7.7 and 7.12 and A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ in Definitions 8.3 and 8.7 is that in N N N -simplices in A ∞ C a t R u n , ∞ \mathop{\mathrm{A}}{}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ we have m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 , but in N N N -simplices in A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ we allow 0 ≠ m 0 I ; x 0 ∈ F ▹ 0 ( ⋯ ) 0\ne m_{0}^{I;x_{0}}\in F^{\triangleright0}(\cdots) 0 = m 0 I ; x 0 ∈ F ▹ 0 ( ⋯ ) if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . However, in the ∞ \infty ∞ -subcategory A ∞ f i C a t R , 2 , 2 u n , ∞ ⊂ A ∞ f i C a t R u n , ∞ \mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2}\subset\mathop{\mathrm{A}}{}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R , 2 , 2 un , ∞ ⊂ A ∞ fi Cat R un , ∞ we require that m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 if ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 .
Thus, N N N -simplices of A ∞ f i C a t R , 2 , 2 u n , ∞ \mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2} A ∞ fi Cat R , 2 , 2 un , ∞ are N N N -simplices of A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ with extra filtration data, and similarly for A ∞ w f C a t R , 2 , 2 u n , ∞ \mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2} A ∞ wf Cat R , 2 , 2 un , ∞ . Hence forgetting filtrations gives ∞ \infty ∞ -functors
Π A ∞ A ∞ f i : A ∞ f i C a t R , 2 , 2 u n , ∞ ⟶ A ∞ C a t R u n , ∞ . (231) \Pi^{\mathrm{A}^{\mathrm{fi}}_{\infty}}_{\mathrm{A}_{\infty}}:\mathrm{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2}\longrightarrow\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}. \tag*{(231)} Π A ∞ A ∞ fi : A ∞ fi Cat R , 2 , 2 un , ∞ ⟶ A ∞ Cat R un , ∞ . (231)
Π A ∞ A ∞ w f : A ∞ w f C a t R , 2 , 2 u n , ∞ ⟶ A ∞ C a t R u n , ∞ . (232) \Pi^{\mathrm{A}^{\mathrm{wf}}_{\infty}}_{\mathrm{A}_{\infty}}:\mathrm{A}^{\mathrm{wf}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2}\longrightarrow\mathrm{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}. \tag*{(232)} Π A ∞ A ∞ wf : A ∞ wf Cat R , 2 , 2 un , ∞ ⟶ A ∞ Cat R un , ∞ . (232)
Other ∞ \infty ∞ -categories of filtered A ∞ A_{\infty} A ∞ -categories
Here is the filtered analogue of Definition 7.22.
Definition 8.14. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\leq L K ≤ L and L ≥ 3 L\geq3 L ≥ 3 . We will define a simplicial set A ˙ ∞ f i C a t R , K , L u n , ∞ \dot{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ of unital filtered A ∞ A_{\infty} A ∞ -categories over R R R .
(a) For N ≥ 0 N\geq0 N ≥ 0 , define the N N N -simplices ( A ˙ ∞ f i C a t R , K , L u n , ∞ ) N (\dot{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L})_N ( A ˙ ∞ fi Cat R , K , L un , ∞ ) N in A ˙ ∞ f i C a t R , K , L u n , ∞ \dot{\mathrm{A}}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ to be the set of quadruples ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m k I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I}_{k})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m k I ) ∅ = I ⊆ { 0 , … , N } ) , where
(i) O i \mathcal{O}^{i} O i is a set for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , thought of as the set of objects Obj A i \operatorname{Obj}\mathcal{A}^{i} Obj A i of a filtered A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i at vertex i i i of the N N N -simplex Δ N \Delta^{N} Δ N .
(ii) f i j : O i → O j f^{ij}:\mathcal{O}^{i}\to\mathcal{O}^{j} f ij : O i → O j is a map for all 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , thought of as the map F Obj i j : Obj A i → Obj A j \mathcal{F}^{ij}_{\operatorname{Obj}}:\operatorname{Obj}\mathcal{A}^{i}\to\operatorname{Obj}\mathcal{A}^{j} F Obj ij : Obj A i → Obj A j in a filtered A ∞ A_{\infty} A ∞ -functor F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j living on the edge [ i , j ] [i,j] [ i , j ] of Δ N \Delta^{N} Δ N . We require these maps to satisfy f i i = id O i f^{ii}=\operatorname{id}_{\mathcal{O}^{i}} f ii = id O i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N and f j k ∘ f i j = f i k f^{jk}\circ f^{ij}=f^{ik} f j k ∘ f ij = f ik for 0 ≤ i ≤ j ≤ k ≤ N 0\leq i\leq j\leq k\leq N 0 ≤ i ≤ j ≤ k ≤ N .
(iii) For all i = 0 , 1 , … , N i=0,1,\ldots,N i = 0 , 1 , … , N and x , y ∈ O i x,y\in\mathcal{O}^{i} x , y ∈ O i , we are given a Z \mathbb{Z} Z -graded Γ \Gamma Γ -filtered R R R -module H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d \mathcal{H}^{i}(x,y)=\bigoplus_{d\in\mathbb{Z}}\mathcal{H}^{i}(x,y)^{d} H i ( x , y ) = ⨁ d ∈ Z H i ( x , y ) d , thought of as the Hom space Hom A i ( x , y ) \operatorname{Hom}_{\mathcal{A}^{i}}(x,y) Hom A i ( x , y ) in the A ∞ A_{\infty} A ∞ -category A i \mathcal{A}^{i} A i .
We require that H i ( x , y ) d \mathcal{H}^{i}(x,y)^{d} H i ( x , y ) d is a complete filtered R R R -module for each d ∈ Z d\in\mathbb{Z} d ∈ Z (this is called degree-wise complete ), but we do not require H i ( x , y ) \mathcal{H}^{i}(x,y) H i ( x , y ) to be complete. We write H i = ( H i ( x , y ) ) x , y ∈ O i \mathcal{H}^{i}=(\mathcal{H}^{i}(x,y))_{x,y\in\mathcal{O}^{i}} H i = ( H i ( x , y ) ) x , y ∈ O i .
(iv) For each nonempty subset I ⊆ { 0 , … , N } I\subseteq\{0,\ldots,N\} I ⊆ { 0 , … , N } , which we write as I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M with M ≥ 0 M\geq0 M ≥ 0 , and for all k ≥ 0 k\geq0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{i_{0}} x 0 , … , x k ∈ O i 0 , thought of as objects in A i 0 \mathcal{A}^{i_{0}} A i 0 , we are given an R R R -linear map
m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k ) ) (233) m^{I;x_{0},\ldots,x_{k}}_{k}:\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{i_{0}}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{i_{0}}(x_{k-1},x_{k})
\longrightarrow\mathcal{H}^{i_{M}}(f^{i_{0}i_{M}}(x_{0}),f^{i_{0}i_{M}}(x_{k})) \tag*{(233)} m k I ; x 0 , … , x k : H i 0 ( x 0 , x 1 ) ⊗ R H i 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H i 0 ( x k − 1 , x k ) ⟶ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k )) (233) graded of degree 2 − M − k = 3 − ∣ I ∣ − k 2-M-k=3-|I|-k 2 − M − k = 3 − ∣ I ∣ − k .
We require m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k to be filtration-preserving for all k ≥ 1 k\geq1 k ≥ 1 and x 0 , … , x k x_{0},\ldots,x_{k} x 0 , … , x k in O i 0 \mathcal{O}^{i_{0}} O i 0 , that is, for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ we have
m k x 0 , … , x k ( F γ ( H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x k − 1 , x k ) ) ) ⊆ F γ ( H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k ) ) ) . m^{x_{0},\ldots,x_{k}}_{k}\left(F^{\gamma}\left(\mathcal{H}^{i_{0}}(x_{0},x_{1})\otimes\cdots\otimes\mathcal{H}^{i_{0}}(x_{k-1},x_{k})\right)\right)
\subseteq F^{\gamma}\left(\mathcal{H}^{i_{M}}(f^{i_{0}i_{M}}(x_{0}),f^{i_{0}i_{M}}(x_{k}))\right). m k x 0 , … , x k ( F γ ( H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ H i 0 ( x k − 1 , x k ) ) ) ⊆ F γ ( H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x k )) ) .
For all x 0 ∈ O i 0 x_0 \in\mathcal{O}^{i_0} x 0 ∈ O i 0 we require that m 0 I ; x 0 = 0 m_0^{I;x_0}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\leq K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) ∈ F ≽ 0 Hom A i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) m_0^{I;x_0}(1)\in F^{\succcurlyeq0}\operatorname{Hom}_{\mathcal{A}^{i_M}}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0)) m 0 I ; x 0 ( 1 ) ∈ F ≽ 0 Hom A i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 )) if K < ∣ I ∣ ≤ L K<|I|\leq L K < ∣ I ∣ ≤ L , and m 0 I ; x 0 ( 1 ) m_0^{I;x_0}(1) m 0 I ; x 0 ( 1 ) lies in F 0 H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) F^0\mathcal{H}^{i_M}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0)) F 0 H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 )) if ∣ I ∣ > L |I|>L ∣ I ∣ > L .
We require these maps m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k to satisfy the following identity generalizing (7.25) for all I = { i 0 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_0<\cdots<i_M\}\subseteq\{0,\ldots,N\} I = { i 0 < ⋯ < i M } ⊆ { 0 , … , N } , n ≥ 0 n\geq0 n ≥ 0 and objects x 0 , … , x n ∈ O i 0 x_0,\ldots,x_n\in\mathcal{O}^{i_0} x 0 , … , x n ∈ O i 0
∑ 0 ≤ j ≤ M , k ≥ 0 , y 0 , … , y k ∈ O i j ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = ∑ j = 1 M − 1 ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n . (234) \sum_{\substack{0\leq j\leq M,\ k\geq0,\\ y_0,\ldots,y_k\in\mathcal{O}^{i_j}}}(-1)^j m_k^{\{i_j,\ldots,i_M\};y_0,\ldots,y_k}\circ d_{n,k;y_0,\ldots,y_k}^{\{i_0,\ldots,i_j\};x_0,\ldots,x_n}
=
\sum_{j=1}^{M-1}(-1)^j m_n^{I\setminus\{i_j\};x_0,\ldots,x_n}. \tag*{(234)} 0 ≤ j ≤ M , k ≥ 0 , y 0 , … , y k ∈ O i j ∑ ( − 1 ) j m k { i j , … , i M } ; y 0 , … , y k ∘ d n , k ; y 0 , … , y k { i 0 , … , i j } ; x 0 , … , x n = j = 1 ∑ M − 1 ( − 1 ) j m n I ∖ { i j } ; x 0 , … , x n . (234)
Here d n , k ; y 0 , … , y k I ; x 0 , … , x n d_{n,k;y_0,\ldots,y_k}^{I;x_0,\ldots,x_n} d n , k ; y 0 , … , y k I ; x 0 , … , x n is defined using the m l J ; w 0 , … , w l m_l^{J;w_0,\ldots,w_l} m l J ; w 0 , … , w l as in Definition 7.2, but modified as in Remark 8.1 to sum over filtered balanced trees Γ i \Gamma_i Γ i in (7.9). Equation (8.11) converges in the filtered sense as for (8.5).
(v) Definition 8.3(a)(v) holds.
(b) If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , we define
A ˙ ∞ f i Cat R , K , L u n , ∞ ( α ) : ( A ˙ ∞ f i Cat R , K , L u n , ∞ ) N ⟶ ( A ˙ ∞ f i Cat R , K , L u n , ∞ ) N ‾ \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}(\alpha):
\left(\dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\right)_N
\longrightarrow
\left(\dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\right)_{\overline{N}} A ˙ ∞ fi Cat R , K , L un , ∞ ( α ) : ( A ˙ ∞ fi Cat R , K , L un , ∞ ) N ⟶ ( A ˙ ∞ fi Cat R , K , L un , ∞ ) N
as in Definition 7.22(b). Essentially the same proof as Proposition 7.23 shows that A ˙ ∞ f i Cat R , K , L u n , ∞ ( α ) \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}(\alpha) A ˙ ∞ fi Cat R , K , L un , ∞ ( α ) maps to ( A ˙ ∞ f i Cat R , K , L u n , ∞ ) N ‾ \left(\dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\right)_{\overline{N}} ( A ˙ ∞ fi Cat R , K , L un , ∞ ) N . As in Definition 7.22(b), A ˙ ∞ f i Cat R , K , L u n , ∞ : Δ o p → S e t \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}:\Delta^{\mathrm{op}}\to\mathrm{Set} A ˙ ∞ fi Cat R , K , L un , ∞ : Δ op → Set is a functor, i.e. A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty} A ˙ ∞ fi Cat R , K , L un , ∞ is a simplicial set.
Theorem 8.20 will show that A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty} A ˙ ∞ fi Cat R , K , L un , ∞ is an ∞ \infty ∞ -category. We say that A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty} A ˙ ∞ fi Cat R , K , L un , ∞ has type ( K , L ) (K,L) ( K , L ) . Clearly A ˙ ∞ f i Cat R , K , L u n , ∞ ⊆ A ˙ ∞ f i Cat R , K ′ , L ′ u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\subseteq\dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K',L'}^{\mathrm{un},\infty} A ˙ ∞ fi Cat R , K , L un , ∞ ⊆ A ˙ ∞ fi Cat R , K ′ , L ′ un , ∞ if K ≥ K ′ K\geq K' K ≥ K ′ and L ≥ L ′ L\geq L' L ≥ L ′ , as in (8.1).
As for Definition 8.4, we also give an analogue with weakly filtered N N N -morphisms for N ≥ 2 N\geq2 N ≥ 2 , used to define (8.2).
Definition 8.15. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\leq L K ≤ L and L ≥ 3 L\geq3 L ≥ 3 . We will define a simplicial set A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{wf}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty} A ˙ ∞ wf Cat R , K , L un , ∞ of unital filtered A ∞ A_\infty A ∞ -categories over R R R .
For N ≥ 0 N\geq0 N ≥ 0 , the N N N -simplices ( A ˙ ∞ w f Cat R , K , L u n , ∞ ) N \left(\dot{\mathcal{A}}_{\infty}^{\mathrm{wf}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\right)_N ( A ˙ ∞ wf Cat R , K , L un , ∞ ) N in A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}_{\infty}^{\mathrm{wf}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty} A ˙ ∞ wf Cat R , K , L un , ∞ are defined as for ( A ˙ ∞ f i Cat R , K , L u n , ∞ ) N \left(\dot{\mathcal{A}}_{\infty}^{\mathrm{fi}}\operatorname{Cat}_{R,K,L}^{\mathrm{un},\infty}\right)_N ( A ˙ ∞ fi Cat R , K , L un , ∞ ) N in Definition 8.14(a), but with the following changes to (a)(iv): there should exist δ ⪯ 0 \delta\preceq0 δ ⪯ 0 in Γ \Gamma Γ independent of I , k , x 0 , … , x k I,k,x_0,\ldots,x_k I , k , x 0 , … , x k such that the following holds.
The R R R -linear maps m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k in (8.10) should be filtration-preserving for ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 and k ≥ 1 k\geq1 k ≥ 1 , but for ∣ I ∣ ≥ 3 |I|\geq3 ∣ I ∣ ≥ 3 and k ≥ 1 k\geq1 k ≥ 1 they need not preserve filtrations, but should instead map F γ ( ⋯ ) → F γ + δ ( ⋯ ) F^\gamma(\cdots)\to F^{\gamma+\delta}(\cdots) F γ ( ⋯ ) → F γ + δ ( ⋯ ) for all γ ∈ Γ \gamma\in\Gamma γ ∈ Γ .
We require that m 0 I ; x 0 = 0 m_0^{I;x_0}=0 m 0 I ; x 0 = 0 if ∣ I ∣ ≤ K |I|\leq K ∣ I ∣ ≤ K , and m 0 I ; x 0 ( 1 ) ∈ F ≽ 0 Hom A i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) m_0^{I;x_0}(1)\in F^{\succcurlyeq0}\operatorname{Hom}_{\mathcal{A}^{i_M}}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0)) m 0 I ; x 0 ( 1 ) ∈ F ≽ 0 Hom A i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 )) if K < ∣ I ∣ ≤ L K<|I|\leq L K < ∣ I ∣ ≤ L , and m 0 I ; x 0 ( 1 ) ∈ F δ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) m_0^{I;x_0}(1)\in F^\delta\mathcal{H}^{i_M}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0)) m 0 I ; x 0 ( 1 ) ∈ F δ H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 )) if ∣ I ∣ > L |I|>L ∣ I ∣ > L (this replaces m 0 I ; x 0 ( 1 ) ∈ F 0 H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ) ) ) , f o r a l l m_0^{I;x_0}(1)\in F^0\mathcal{H}^{i_M}(f^{i_0i_M}(x_0),f^{i_0i_M}(x_0))), for all m 0 I ; x 0 ( 1 ) ∈ F 0 H i M ( f i 0 i M ( x 0 ) , f i 0 i M ( x 0 ))) , f or a l l x_0\in\mathcal{O}^{i_0}$. Then (8.11) converges in the filtered sense, by the argument in Definition 8.4. We also require that Definition 8.11(iv) holds. The rest of the definition of A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ wf Cat R , K , L un , ∞ follows Definition 8.14. Theorem 8.20 will show A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ wf Cat R , K , L un , ∞ is an ∞ \infty ∞ -category. Note that A ˙ ∞ f i Cat R , K , L u n , ∞ ⊂ A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L}\subset\dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ ⊂ A ˙ ∞ wf Cat R , K , L un , ∞ .
The next theorem, the analogue of Theorem 7.24, will be proved in §10.7 using equation (8.11).
Theorem 8.16. Let ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊂ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subset\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊂ { 0 , … , N } ) be an N N N -simplex in A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ or A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ wf Cat R , K , L un , ∞ as in Definitions 8.14 and 8.15, and define maps d n , k ; y 0 , … , y k I ; x 0 , … , x n d^{I;x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}} d n , k ; y 0 , … , y k I ; x 0 , … , x n from the m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k as in (7.6)–(7.9). Then for each nonempty subset I ⊂ { 0 , … , N } I\subset\{0,\ldots,N\} I ⊂ { 0 , … , N } , written I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M , and all k , n ≥ 0 k,n\ge0 k , n ≥ 0 , x 0 , … , x n ∈ O i 0 x_{0},\ldots,x_{n}\in\mathcal{O}^{i_{0}} x 0 , … , x n ∈ O i 0 and y 0 , … , y k ∈ O i M y_{0},\ldots,y_{k}\in\mathcal{O}^{i_{M}} y 0 , … , y k ∈ O i M , in R R R -linear maps as in (7.6) we have
∑ j = 0 M ∑ l ≥ 0 , w 0 , … , w l ∈ O i j ( − 1 ) j d l , k ; y 0 , … , y k { i j , … , i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 , … , i j } ; x 0 , … , x n = ∑ j = 1 M − 1 ( − 1 ) j d n , k ; y 0 , … , y k I ∖ { i j } ; x 0 , … , x n , (235) \sum_{j=0}^{M}\sum_{\substack{l\ge0,\ w_{0},\ldots,w_{l}\in\mathcal{O}^{i_{j}}}}(-1)^{j}d^{\{i_{j},\ldots,i_{M}\};w_{0},\ldots,w_{l}}_{l,k;y_{0},\ldots,y_{k}}\circ d^{\{i_{0},\ldots,i_{j}\};x_{0},\ldots,x_{n}}_{n,l;w_{0},\ldots,w_{l}}
=\sum_{j=1}^{M-1}(-1)^{j}d^{I\setminus\{i_{j}\};x_{0},\ldots,x_{n}}_{n,k;y_{0},\ldots,y_{k}}, \tag*{(235)} j = 0 ∑ M l ≥ 0 , w 0 , … , w l ∈ O i j ∑ ( − 1 ) j d l , k ; y 0 , … , y k { i j , … , i M } ; w 0 , … , w l ∘ d n , l ; w 0 , … , w l { i 0 , … , i j } ; x 0 , … , x n = j = 1 ∑ M − 1 ( − 1 ) j d n , k ; y 0 , … , y k I ∖ { i j } ; x 0 , … , x n , (235) where the first line is an infinite sum which converges in the filtered sense.
Here is the analogue of Examples 7.25 and 8.6.
Example 8.17. (a) A 0-simplex ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) (\mathcal{O}^{0},f^{00}=\operatorname{id}_{\mathcal{O}^{0}},\mathcal{H}^{0},m^{\{0\}}) ( O 0 , f 00 = id O 0 , H 0 , m { 0 } ) of A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ is equivalent to a unital filtered A ∞ A_{\infty} A ∞ -category A 0 \mathcal{A}^{0} A 0 over R R R , which is strict if K > 0 K>0 K > 0 , with objects Obj A 0 = O 0 \operatorname{Obj}\mathcal{A}^{0}=\mathcal{O}^{0} Obj A 0 = O 0 , Hom A 0 ( x , y ) = H 0 ( x , y ) \operatorname{Hom}_{\mathcal{A}^{0}}(x,y)=\mathcal{H}^{0}(x,y) Hom A 0 ( x , y ) = H 0 ( x , y ) , morphisms ( m k { 0 } ; x 0 , … , x k ) k ≥ 1 (m^{\{0\};x_{0},\ldots,x_{k}}_{k})_{k\ge1} ( m k { 0 } ; x 0 , … , x k ) k ≥ 1 , and units 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 1^{0}_{x}\in\operatorname{Hom}^{0}_{\mathcal{A}^{0}}(x,x)=\mathcal{H}^{0}(x,x)^{0} 1 x 0 ∈ Hom A 0 0 ( x , x ) = H 0 ( x , x ) 0 .
(b) A 1-simplex of A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ is equivalent to a unital filtered A ∞ A_{\infty} A ∞ -functor F 01 : A 0 → A 1 \mathcal{F}^{01}:\mathcal{A}^{0}\to\mathcal{A}^{1} F 01 : A 0 → A 1 , which is strict if K > 1 K>1 K > 1 , with F O b j 01 = f 01 \mathcal{F}^{01}_{\mathrm{Obj}}=f^{01} F Obj 01 = f 01 and morphisms F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k \mathcal{F}^{01;x_{0},\ldots,x_{k}}_{k}=m^{\{0,1\};x_{0},\ldots,x_{k}}_{k} F k 01 ; x 0 , … , x k = m k { 0 , 1 } ; x 0 , … , x k .
(c) A 2-simplex of A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{\mathcal{A}}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ is equivalent to unital filtered A ∞ A_{\infty} A ∞ -categories A 0 , A 1 , A 2 \mathcal{A}^{0},\mathcal{A}^{1},\mathcal{A}^{2} A 0 , A 1 , A 2 over R R R as in (a), which are strict if K > 0 K>0 K > 0 , and unital filtered A ∞ A_{\infty} A ∞ -functors F i j : A i → A j \mathcal{F}^{ij}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j for ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) (i,j)=(0,1),(0,2),(1,2) ( i , j ) = ( 0 , 1 ) , ( 0 , 2 ) , ( 1 , 2 ) as in (b), which are strict if K > 1 K>1 K > 1 , and R R R -linear maps for all k ≥ 0 k\ge0 k ≥ 0 and x 0 , … , x k ∈ O 0 x_{0},\ldots,x_{k}\in\mathcal{O}^{0} x 0 , … , x k ∈ O 0
m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 02 ( x 0 ) , f 02 ( x k ) ) m^{\{0,1,2\};x_{0},\ldots,x_{k}}_{k}:\mathcal{H}^{0}(x_{0},x_{1})\otimes_{R}\mathcal{H}^{0}(x_{1},x_{2})\otimes_{R}\cdots\otimes_{R}\mathcal{H}^{0}(x_{k-1},x_{k})\longrightarrow\mathcal{H}^{2}(f^{02}(x_{0}),f^{02}(x_{k})) m k { 0 , 1 , 2 } ; x 0 , … , x k : H 0 ( x 0 , x 1 ) ⊗ R H 0 ( x 1 , x 2 ) ⊗ R ⋯ ⊗ R H 0 ( x k − 1 , x k ) ⟶ H 2 ( f 02 ( x 0 ) , f 02 ( x k ))
graded of degree − k -k − k , satisfying (211) and
∑ i , j , k ≥ 0 i + j + k = n ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) (236) \sum_{\substack{i,j,k\geq0\\i+j+k=n}}(-1)^{i+jk}m_{i+k+1}^{\{0,1,2\};x_0,x_1,\ldots,x_i,x_{i+j},x_{i+j+1},\ldots,x_n}\circ\left(\operatorname{id}_{\mathcal H^0(x_0,x_1)}\otimes\cdots\otimes\operatorname{id}_{\mathcal H^0(x_{i-1},x_i)}\otimes m_j^{\{0\};x_i,x_{i+1},\ldots,x_{i+j}}\otimes\operatorname{id}_{\mathcal H^0(x_{i+j},x_{i+j+1})}\otimes\cdots\otimes\operatorname{id}_{\mathcal H^0(x_{n-1},x_n)}\right) \tag*{(236)} i , j , k ≥ 0 i + j + k = n ∑ ( − 1 ) i + j k m i + k + 1 { 0 , 1 , 2 } ; x 0 , x 1 , … , x i , x i + j , x i + j + 1 , … , x n ∘ ( id H 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H 0 ( x i − 1 , x i ) ⊗ m j { 0 } ; x i , x i + 1 , … , x i + j ⊗ id H 0 ( x i + j , x i + j + 1 ) ⊗ ⋯ ⊗ id H 0 ( x n − 1 , x n ) ) (236)
− ( m n { 1 , 2 } ∘ m n { 0 , 1 } ) x 0 , … , x n + ∑ 1 ≤ l ≤ k , n = n 1 + ⋯ + n k , n i ≥ 0 , i = 1 , … , k ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ l = 1 i − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i ) -\left(m_n^{\{1,2\}}\circ m_n^{\{0,1\}}\right)^{x_0,\ldots,x_n}
+\sum_{\substack{1\leq l\leq k,\\n=n_1+\cdots+n_k,\\n_i\geq0,\ i=1,\ldots,k}}(-1)^{k+1}\cdot(-1)^{\sum_{l=1}^{i-1}(n_i+1)}\cdot(-1)^{\sum_{i=1}^{k}(n_i+1)(k-i)} − ( m n { 1 , 2 } ∘ m n { 0 , 1 } ) x 0 , … , x n + 1 ≤ l ≤ k , n = n 1 + ⋯ + n k , n i ≥ 0 , i = 1 , … , k ∑ ( − 1 ) k + 1 ⋅ ( − 1 ) ∑ l = 1 i − 1 ( n i + 1 ) ⋅ ( − 1 ) ∑ i = 1 k ( n i + 1 ) ( k − i )
{ m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) ∘ [ m n 1 { 1 , 2 } ∘ m n 1 { 0 , 1 } ] n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ [ m n l − 1 { 1 , 2 } ∘ m n l − 1 { 0 , 1 } ] n l − 1 x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = − m n { 0 , 2 } ; x 0 , … , x n , \begin{aligned}
\left\{m_k^{\{2\};f^{\{0,1,2\}}(x_{p_0}),f^{\{0,1,2\}}(x_{p_1}),\ldots,f^{\{0,1,2\}}(x_{p_{l-1}}),f^{02}(x_{p_l}),\ldots,f^{02}(x_{p_k})}\circ\left[m_{n_1}^{\{1,2\}}\circ m_{n_1}^{\{0,1\}}\right]^{x_{p_0},\ldots,x_{p_1}}_{n_1}\otimes\cdots\otimes\left[m_{n_{l-1}}^{\{1,2\}}\circ m_{n_{l-1}}^{\{0,1\}}\right]^{x_{p_{l-2}},\ldots,x_{p_{l-1}}}_{n_{l-1}}\otimes m_{n_l}^{\{0,1,2\};x_{p_{l-1}},\ldots,x_{p_l}}\otimes m_{n_{l+1}}^{\{0,2\};x_{p_l},\ldots,x_{p_{l+1}}}\otimes\cdots\otimes m_{n_k}^{\{0,2\};x_{p_{k-1}},\ldots,x_{p_k}}\right] \\
&=-m_n^{\{0,2\};x_0,\ldots,x_n},
\end{aligned} { m k { 2 } ; f { 0 , 1 , 2 } ( x p 0 ) , f { 0 , 1 , 2 } ( x p 1 ) , … , f { 0 , 1 , 2 } ( x p l − 1 ) , f 02 ( x p l ) , … , f 02 ( x p k ) ∘ [ m n 1 { 1 , 2 } ∘ m n 1 { 0 , 1 } ] n 1 x p 0 , … , x p 1 ⊗ ⋯ ⊗ [ m n l − 1 { 1 , 2 } ∘ m n l − 1 { 0 , 1 } ] n l − 1 x p l − 2 , … , x p l − 1 ⊗ m n l { 0 , 1 , 2 } ; x p l − 1 , … , x p l ⊗ m n l + 1 { 0 , 2 } ; x p l , … , x p l + 1 ⊗ ⋯ ⊗ m n k { 0 , 2 } ; x p k − 1 , … , x p k ] = − m n { 0 , 2 } ; x 0 , … , x n ,
where p j = ∑ i = 1 j n i p_j=\sum_{i=1}^{j}n_i p j = ∑ i = 1 j n i , 0 ≤ j ≤ k 0\leq j\leq k 0 ≤ j ≤ k .
with m 0 { 0 , 1 , 2 } ; x 0 = 0 m_0^{\{0,1,2\};x_0}=0 m 0 { 0 , 1 , 2 } ; x 0 = 0 if K > 2 K>2 K > 2 .
By Definition 5.21 and equations (5.13) and (8.13), m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 0 m^{\{0,1,2\}}=(m_k^{\{0,1,2\}})_{k\geq0} m { 0 , 1 , 2 } = ( m k { 0 , 1 , 2 } ) k ≥ 0 is a unital filtered A ∞ A_\infty A ∞ -homotopy F { 1 , 2 } ∘ F { 0 , 1 } ⇒ F { 0 , 2 } \mathcal F^{\{1,2\}}\circ\mathcal F^{\{0,1\}}\Rightarrow\mathcal F^{\{0,2\}} F { 1 , 2 } ∘ F { 0 , 1 } ⇒ F { 0 , 2 } , which is strict if K > 2 K>2 K > 2 . Hence, a 2-simplex in A ∞ f i C a t R , K , L u n , ∞ \mathsf A^{\mathrm{fi}}_{\infty}\mathsf{Cat}^{\mathrm{un},\infty}_{R,K,L} A ∞ fi Cat R , K , L un , ∞ is a 2-commutative diagram of (strict if K > 0 K>0 K > 0 ) unital filtered A ∞ A_\infty A ∞ -categories, (strict if K > 1 K>1 K > 1 ) unital filtered A ∞ A_\infty A ∞ -functors, and a (strict if K > 2 K>2 K > 2 ) unital filtered A ∞ A_\infty A ∞ -homotopy
A 0 → F { 0 , 1 } A 1 → F { 1 , 2 } A 2 F { 0 , 2 } ↓ m { 0 , 1 , 2 } ↓ A 2 = A 2 \begin{CD}
\mathcal A^0 @>{\mathcal F^{\{0,1\}}}>> \mathcal A^1 @>{\mathcal F^{\{1,2\}}}>> \mathcal A^2\\
@V{\mathcal F^{\{0,2\}}}VV @V{m^{\{0,1,2\}}}VV @.\\
\mathcal A^2 @= \mathcal A^2 @.
\end{CD} A 0 F { 0 , 2 } ↓ ⏐ A 2 F { 0 , 1 } A 1 m { 0 , 1 , 2 } ↓ ⏐ A 2 F { 1 , 2 } A 2
The following analogue of Theorem 7.26 is proved in §10.8.
Theorem 8.18. Let K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L\in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K\leq L K ≤ L and L ≥ 3 L\geq3 L ≥ 3 . Then there is a strict isomorphism of ∞ \infty ∞ -categories
I R , K , L u n , f i : A ∞ f i C a t R , K , L u n , ∞ ⟶ A ˙ ∞ f i C a t R , K , L u n , ∞ (237) I^{\mathrm{un},\mathrm{fi}}_{R,K,L}:\mathsf A^{\mathrm{fi}}_{\infty}\mathsf{Cat}^{\mathrm{un},\infty}_{R,K,L}\longrightarrow\dot{\mathsf A}^{\mathrm{fi}}_{\infty}\mathsf{Cat}^{\mathrm{un},\infty}_{R,K,L} \tag*{(237)} I R , K , L un , fi : A ∞ fi Cat R , K , L un , ∞ ⟶ A ˙ ∞ fi Cat R , K , L un , ∞ (237) from Definitions 8.11 and 8.14, which acts on N N N -simplices by
I R , K , L u n , f i : ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ⟼ ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ ≠ I ⊆ { 0 , … , N } ) , (238) I^{\mathrm{un},\mathrm{fi}}_{R,K,L}:\left((\mathcal O^i)_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal H^i)_{i=0}^{N},(m^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right)\longmapsto\left((\mathcal O^i)_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal H^i)_{i=0}^{N},(\dot m^I)_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\right), \tag*{(238)} I R , K , L un , fi : ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) ⟼ ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m ˙ I ) ∅ = I ⊆ { 0 , … , N } ) , (238) where m ˙ k I ; x 0 , … , x k \dot m_k^{I;x_0,\ldots,x_k} m ˙ k I ; x 0 , … , x k is defined from m k I ; x 0 , … , x k m_k^{I;x_0,\ldots,x_k} m k I ; x 0 , … , x k by
m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { m 0 { i 0 , i 1 , i 2 } ; x 0 − 1 f i 2 i 0 ( x 0 ) , k = 0 and M = 2 , m k { i 0 , … , i M } ; x 0 , … , x k , otherwise. (239) \dot m_k^{\{i_0,\ldots,i_M\};x_0,\ldots,x_k}=
\begin{cases}
m_0^{\{i_0,i_1,i_2\};x_0}-1_{f^{i_2i_0}(x_0)}, & k=0\text{ and }M=2,\\
m_k^{\{i_0,\ldots,i_M\};x_0,\ldots,x_k}, & \text{otherwise.}
\end{cases} \tag*{(239)} m ˙ k { i 0 , … , i M } ; x 0 , … , x k = { m 0 { i 0 , i 1 , i 2 } ; x 0 − 1 f i 2 i 0 ( x 0 ) , m k { i 0 , … , i M } ; x 0 , … , x k , k = 0 and M = 2 , otherwise. (239)
Similarly, there is a strict isomorphism of ∞ \infty ∞ -categories
I R , K , L u n , w f : A ∞ w f Cat R , K , L u n , ∞ ⟶ A ˙ ∞ w f Cat R , K , L u n , ∞ (240) I^{\mathrm{un,wf}}_{R,K,L}: A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} \longrightarrow\dot{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} \tag*{(240)} I R , K , L un , wf : A ∞ wf Cat R , K , L un , ∞ ⟶ A ˙ ∞ wf Cat R , K , L un , ∞ (240)
from Definitions 8.11 and 8.15.
Here is the analogue of Definition 7.28.
Definition 8.19. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid, R R R a complete Γ \Gamma Γ -filtered ring, and K , L ∈ { 0 , 1 , 2 , … , ∞ } K,L \in\{0,1,2,\ldots,\infty\} K , L ∈ { 0 , 1 , 2 , … , ∞ } with K ≤ L K \le L K ≤ L and L ≥ 3 L \ge3 L ≥ 3 . Generalizing Definition 8.14, define a simplicial set A ∞ f i Cat R , K , L ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ fi Cat R , K , L ∞ of (possibly non-unital) filtered A ∞ A_{\infty} A ∞ -categories over R R R whose N N N -simplices
( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) \left((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}\right) ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) for N ≥ 0 N \ge0 N ≥ 0 satisfy Definition 8.14(a)(i)–(iv), but need not satisfy (a)(v). If α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] is a morphism in Δ \Delta Δ , define
A ∞ f i Cat R , K , L ∞ ( α ) : ( A ∞ f i Cat R , K , L ∞ ) N ⟶ ( A ∞ f i Cat R , K , L ∞ ) N ‾ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}(\alpha):(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L})^{N}\longrightarrow(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L})^{\overline{N}} A ∞ fi Cat R , K , L ∞ ( α ) : ( A ∞ fi Cat R , K , L ∞ ) N ⟶ ( A ∞ fi Cat R , K , L ∞ ) N as in Definition 8.14(b). Essentially the same proof as Proposition 7.23 shows A ∞ f i Cat R , K , L ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ fi Cat R , K , L ∞ is a simplicial set. Theorem 8.16 also holds in A ∞ f i Cat R , K , L ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ fi Cat R , K , L ∞ , as the proof does not use unitality. Theorem 8.20 will show that A ∞ f i Cat R , K , L ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ fi Cat R , K , L ∞ is an ∞ \infty ∞ -category. We say that A ∞ f i Cat R , K , L ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ fi Cat R , K , L ∞ has type ( K , L ) (K,L) ( K , L ) . We also write A ∞ f i Cat R ∞ = A ∞ f i Cat R , 0 , 3 ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R}=A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,0,3} A ∞ fi Cat R ∞ = A ∞ fi Cat R , 0 , 3 ∞ .
We also define ∞ \infty ∞ -categories A ∞ w f Cat R , K , L ∞ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L} A ∞ wf Cat R , K , L ∞ as in (8.2) exactly as above, but modifying the definition of A ˙ ∞ w f Cat R , K , L u n , ∞ \dot{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ wf Cat R , K , L un , ∞ in Definition 8.15 rather than A ˙ ∞ f i Cat R , K , L u n , ∞ \dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L} A ˙ ∞ fi Cat R , K , L un , ∞ in Definition 8.14. As in Definition 8.15, we require that Definition 8.11(iv) holds.
The following generalization of Theorem 7.29 is proved in §10.10.
Theorem 8.20. The following simplicial sets defined in Definitions 8.11, 8.14, 8.15, and 8.19, for all K , L K,L K , L with L ≥ 3 L \ge3 L ≥ 3 , are all ∞ \infty ∞ -categories:
A ∞ f i Cat R , K , L u n , ∞ , A ˙ ∞ f i Cat R , K , L u n , ∞ , A ∞ f i Cat R , K , L ∞ , A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\ \dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}, A ∞ fi Cat R , K , L un , ∞ , A ˙ ∞ fi Cat R , K , L un , ∞ , A ∞ fi Cat R , K , L ∞ , A ∞ w f Cat R , K , L u n , ∞ , A ˙ ∞ w f Cat R , K , L u n , ∞ , A ∞ w f Cat R , K , L ∞ . A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\ \dot{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}. A ∞ wf Cat R , K , L un , ∞ , A ˙ ∞ wf Cat R , K , L un , ∞ , A ∞ wf Cat R , K , L ∞ .
Here is the analogue of Definition 7.30. It completes our diagrams (8.1)–(8.2) of inclusions of ∞ \infty ∞ -categories of filtered A ∞ A_{\infty} A ∞ -categories.
Definition 8.21. Suppose ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) is a po-monoid, and R R R is a complete Γ \Gamma Γ -filtered ring, and K = 0 , 1 K=0,1 K = 0 , 1 or ∞ \infty ∞ . Let A ∞ f i Cat R , K u n A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K} A ∞ fi Cat R , K un be the ordinary category of (possibly strict) unital filtered A ∞ A_{\infty} A ∞ -categories over R R R from Definition 5.18, and N ( A ∞ f i Cat R , K u n ) \mathcal{N}(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K}) N ( A ∞ fi Cat R , K un ) be its nerve from Example 2.11, which is an ∞ \infty ∞ -category.
Then N N N -simplices in N ( A ∞ f i Cat R , K u n ) \mathcal{N}(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K}) N ( A ∞ fi Cat R , K un ) are of the form
( ( A i ) i = 0 N , ( F i ) i = 1 N ) , ((\mathcal{A}^{i})_{i=0}^{N},(\mathcal{F}^{i})_{i=1}^{N}), (( A i ) i = 0 N , ( F i ) i = 1 N ) , where A i \mathcal{A}^{i} A i is a (strict if K = 1 , ∞ K=1,\infty K = 1 , ∞ ) unital filtered A ∞ A_{\infty} A ∞ -category over R R R for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and F i : A i − 1 → A i \mathcal{F}^{i}:\mathcal{A}^{i-1}\to\mathcal{A}^{i} F i : A i − 1 → A i is a (strict if K = ∞ K=\infty K = ∞ ) unital filtered A ∞ A_{\infty} A ∞ -functor for i = 1 , … , N i=1,\ldots,N i = 1 , … , N .
Define an ∞ \infty ∞ -functor
inc R , K u n , f i : N ( A ∞ f i Cat R , K u n ) ↪ A ˙ ∞ f i Cat R , K , ∞ u n , ∞ \operatorname{inc}^{\mathrm{un,fi}}_{R,K}:\mathcal{N}(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K})\hookrightarrow\dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty} inc R , K un , fi : N ( A ∞ fi Cat R , K un ) ↪ A ˙ ∞ fi Cat R , K , ∞ un , ∞ to act on N N N -simplices for N ≥ 0 N\ge0 N ≥ 0 mapping
( ( A i ) i = 0 N , ( F i ) i = 1 N ) ⟼ ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) , ((\mathcal{A}^{i})_{i=0}^{N},(\mathcal{F}^{i})_{i=1}^{N})\longmapsto((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\le i\le j\le N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}), (( A i ) i = 0 N , ( F i ) i = 1 N ) ⟼ (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) ,
where we set O i = Obj A i \mathcal{O}^{i}=\operatorname{Obj}\mathcal{A}^{i} O i = Obj A i , and writing F ~ i i = id A i : A i → A i \widetilde{\mathcal{F}}^{ii}=\operatorname{id}_{\mathcal{A}^{i}}:\mathcal{A}^{i}\to\mathcal{A}^{i} F ii = id A i : A i → A i and F ~ i j = F ~ j ∘ F ~ j − 1 ∘ ⋯ ∘ F ~ i + 1 : A i → A j \widetilde{\mathcal{F}}^{ij}=\widetilde{\mathcal{F}}^{j}\circ\widetilde{\mathcal{F}}^{j-1}\circ\cdots\circ\widetilde{\mathcal{F}}^{i+1}:\mathcal{A}^{i}\to\mathcal{A}^{j} F ij = F j ∘ F j − 1 ∘ ⋯ ∘ F i + 1 : A i → A j for 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N , we define f i j = F ~ Obj i j : O i → O j f^{ij}=\widetilde{\mathcal{F}}^{ij}_{\operatorname{Obj}}:\mathcal{O}^{i}\to\mathcal{O}^{j} f ij = F Obj ij : O i → O j for 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N . We set H i ( x , y ) = Hom A i ( x , y ) \mathcal{H}^{i}(x,y)=\operatorname{Hom}_{\mathcal{A}^{i}}(x,y) H i ( x , y ) = Hom A i ( x , y ) for x , y ∈ O i = Obj A i x,y\in\mathcal{O}^{i}=\operatorname{Obj}\mathcal{A}^{i} x , y ∈ O i = Obj A i . We take m k { i } ; x 0 , … , x k m^{\{i\};x_{0},\ldots,x_{k}}_{k} m k { i } ; x 0 , … , x k to be the map m k x 0 , … , x k m^{x_{0},\ldots,x_{k}}_{k} m k x 0 , … , x k in A i \mathcal{A}^{i} A i for i = 0 , … , N i=0,\ldots,N i = 0 , … , N , and m k { i , j } ; x 0 , … , x k m^{\{i,j\};x_{0},\ldots,x_{k}}_{k} m k { i , j } ; x 0 , … , x k to be the map F ~ k i j ; x 0 , … , x k \widetilde{\mathcal{F}}^{ij;x_{0},\ldots,x_{k}}_{k} F k ij ; x 0 , … , x k in F ~ i j \widetilde{\mathcal{F}}^{ij} F ij for 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N , and m k I ; x 0 , … , x k = 0 m^{I;x_{0},\ldots,x_{k}}_{k}=0 m k I ; x 0 , … , x k = 0 if ∣ I ∣ ≥ 3 \lvert I\rvert\geq3 ∣ I ∣ ≥ 3 and k ≥ 0 k\geq0 k ≥ 0 .
We can prove as in Definition 7.30 that inc R , K u n , f i \operatorname{inc}^{\mathrm{un,fi}}_{R,K} inc R , K un , fi is an ∞ \infty ∞ -functor. It is a strict isomorphism with the ∞ \infty ∞ -subcategory of A ˙ ∞ f i Cat R , K , ∞ u n , ∞ \dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty} A ˙ ∞ fi Cat R , K , ∞ un , ∞ consisting of N N N -simplices ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) with m k I ; x 0 , … , x k = 0 m^{I;x_{0},\ldots,x_{k}}_{k}=0 m k I ; x 0 , … , x k = 0 if ∣ I ∣ ≥ 3 \lvert I\rvert\geq3 ∣ I ∣ ≥ 3 and k ≥ 0 k\geq0 k ≥ 0 . Similarly we define ∞ \infty ∞ -functors inc R , K f i : N ( A ∞ f i Cat R , K ) ↪ A ∞ f i Cat R , K , ∞ ∞ \operatorname{inc}^{\mathrm{fi}}_{R,K}:\mathcal{N}(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R,K})\hookrightarrow A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,\infty} inc R , K fi : N ( A ∞ fi Cat R , K ) ↪ A ∞ fi Cat R , K , ∞ ∞ for general filtered A ∞ A_{\infty} A ∞ -categories.
We also define an ∞ \infty ∞ -functor inc R , K , ∞ u n , f i : N ( A ∞ f i Cat R , K u n ) → A ∞ f i Cat R , K , ∞ u n , ∞ \operatorname{inc}^{\mathrm{un,fi}}_{R,K,\infty}:\mathcal{N}(A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R,K})\to A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,\infty} inc R , K , ∞ un , fi : N ( A ∞ fi Cat R , K un ) → A ∞ fi Cat R , K , ∞ un , ∞ which is identified with inc R , K u n , f i \operatorname{inc}^{\mathrm{un,fi}}_{R,K} inc R , K un , fi under the isomorphism I R , K , ∞ u n , f i I^{\mathrm{un,fi}}_{R,K,\infty} I R , K , ∞ un , fi in (8.14). The difference is that as in (7.23), when we define ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) above we set m 0 { i 0 , i 1 , i 2 } ; x 0 = 1 f i 0 i 2 ( x 0 ) i 0 i 2 m^{\{i_{0},i_{1},i_{2}\};x_{0}}_{0}=\mathbf{1}^{i_{0}i_{2}}_{f^{i_{0}i_{2}}(x_{0})} m 0 { i 0 , i 1 , i 2 } ; x 0 = 1 f i 0 i 2 ( x 0 ) i 0 i 2 , not m 0 { i 0 , i 1 , i 2 } ; x 0 = 0 m^{\{i_{0},i_{1},i_{2}\};x_{0}}_{0}=0 m 0 { i 0 , i 1 , i 2 } ; x 0 = 0 .
Definition 8.22. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring. For all K , L K,L K , L , define ∞ \infty ∞ -subcategories
A ∞ f i Alg R u n , ∞ ⊂ A ∞ f i Cat R u n , ∞ , A ∞ f i Alg R , K , L u n , ∞ ⊂ A ∞ f i Cat R , K , L u n , ∞ , A ˙ ∞ f i Alg R , K , L u n , ∞ ⊂ A ˙ ∞ f i Cat R , K , L u n , ∞ , A ∞ f i Alg R ∞ ⊂ A ∞ f i Cat R ∞ , A ∞ f i Alg R , K , L ∞ ⊂ A ∞ f i Cat R , K , L ∞ , A ∞ w f Alg R u n , ∞ ⊂ A ∞ w f Cat R u n , ∞ , A ∞ w f Alg R , K , L u n , ∞ ⊂ A ∞ w f Cat R , K , L u n , ∞ , A ˙ ∞ w f Alg R , K , L u n , ∞ ⊂ A ˙ ∞ w f Cat R , K , L u n , ∞ , A ∞ w f Alg R ∞ ⊂ A ∞ w f Cat R ∞ , A ∞ w f Alg R , K , L ∞ ⊂ A ∞ w f Cat R , K , L ∞ , (241) \begin{aligned}
&A^{\mathrm{fi}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R},\qquad
A^{\mathrm{fi}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R,K,L}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\\
&\dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R,K,L}\subset\dot{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\qquad
A^{\mathrm{fi}}_{\infty}\operatorname{Alg}^{\infty}_{R}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R},\\
&A^{\mathrm{fi}}_{\infty}\operatorname{Alg}^{\infty}_{R,K,L}\subset A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L},\qquad
A^{\mathrm{wf}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R}\subset A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R},\\
&A^{\mathrm{wf}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R,K,L}\subset A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\qquad
\dot{A}^{\mathrm{wf}}_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R,K,L}\subset\dot{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R,K,L},\\
&A^{\mathrm{wf}}_{\infty}\operatorname{Alg}^{\infty}_{R}\subset A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R},\qquad
A^{\mathrm{wf}}_{\infty}\operatorname{Alg}^{\infty}_{R,K,L}\subset A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\infty}_{R,K,L}, \tag*{(241)}
\end{aligned} A ∞ fi Alg R un , ∞ ⊂ A ∞ fi Cat R un , ∞ , A ∞ fi Alg R , K , L un , ∞ ⊂ A ∞ fi Cat R , K , L un , ∞ , A ˙ ∞ fi Alg R , K , L un , ∞ ⊂ A ˙ ∞ fi Cat R , K , L un , ∞ , A ∞ fi Alg R ∞ ⊂ A ∞ fi Cat R ∞ , A ∞ fi Alg R , K , L ∞ ⊂ A ∞ fi Cat R , K , L ∞ , A ∞ wf Alg R un , ∞ ⊂ A ∞ wf Cat R un , ∞ , A ∞ wf Alg R , K , L un , ∞ ⊂ A ∞ wf Cat R , K , L un , ∞ , A ˙ ∞ wf Alg R , K , L un , ∞ ⊂ A ˙ ∞ wf Cat R , K , L un , ∞ , A ∞ wf Alg R ∞ ⊂ A ∞ wf Cat R ∞ , A ∞ wf Alg R , K , L ∞ ⊂ A ∞ wf Cat R , K , L ∞ , (241)
to be the full ∞ \infty ∞ -subcategories of filtered A ∞ A_{\infty} A ∞ -categories A \mathcal{A} A with Obj A = { ∗ } \operatorname{Obj}\mathcal{A}=\{\ast\} Obj A = { ∗ } . These are equivalent to filtered A ∞ A_{\infty} A ∞ -algebras over R R R , so (8.18) are filtered analogues of the ∞ \infty ∞ -categories A ∞ Alg R u n , ∞ , … , A ∞ Alg R , K ∞ A_{\infty}\operatorname{Alg}^{\mathrm{un},\infty}_{R},\ldots,A_{\infty}\operatorname{Alg}^{\infty}_{R,K} A ∞ Alg R un , ∞ , … , A ∞ Alg R , K ∞ from §6.3.
For N N N -simplices ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) in these ∞ \infty ∞ -categories we have O i = { ∗ } \mathcal{O}^{i}=\{\ast\} O i = { ∗ } and f i j = id { ∗ } f^{ij}=\operatorname{id}_{\{\ast\}} f ij = id { ∗ } for all i , j i,j i , j , and we use the notation H i ( ∗ , ∗ ) = A i \mathcal{H}^{i}(\ast,\ast)=A^{i} H i ( ∗ , ∗ ) = A i and m k I ; ∗ , … , ∗ = m k I m^{I;\ast,\ldots,\ast}_{k}=m^{I}_{k} m k I ; ∗ , … , ∗ = m k I , so that N N N -simplices may alternatively be written ( ( A i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((A^{i})_{i=0}^{N},(m^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( A i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , as in Chapter 6.
∞ \infty ∞ -category constructions for A ∞ A_{\infty} A ∞ -categories
Describing mapping spaces in A ∞ Cat R u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞
The next theorem, proved in §11.1, generalizes equation (4.45) in Theorem 4.67 to our ∞ \infty ∞ -categories A ∞ Cat R u n , ∞ A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , A ∞ f i Cat R u n , ∞ A^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , A ∞ w f Cat R u n , ∞ A^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ .
Theorem 9.1. (a) Let A , B \mathcal{A},\mathcal{B} A , B be unital A ∞ A_{\infty} A ∞ -categories over a commutative ring R R R . Then there is a canonical strict isomorphism of Kan complexes
Hom A ∞ Cat R u n , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ u n ( A , B ) ) . (242) \operatorname{Hom}^{L}_{A_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B})\cong\mathcal{N}_{A_{\infty}}\left(\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{A},\mathcal{B})\right). \tag*{(242)} Hom A ∞ Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ un ( A , B ) ) . (242) :::`
Here Hom A ^ ∞ Cat R u n , ∞ L ( A , B ) \operatorname{Hom}^{L}_{\widehat{\mathcal{A}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un , ∞ L ( A , B ) is the space of left morphisms in A ∞ Cat R u n , ∞ \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ from Definition 2.15, one model for the mapping space Hom A ∞ Cat R u n , ∞ ( A , B ) \operatorname{Hom}_{\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B}) Hom A ∞ Cat R un , ∞ ( A , B ) in the ∞ \infty ∞ -category A ∞ Cat R u n , ∞ \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , and N A ∞ ( Fun A ∞ u n ( A , B ) ) \mathcal{N}_{\mathcal{A}_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}_{\infty}}(\mathcal{A},\mathcal{B})) N A ∞ ( Fun A ∞ un ( A , B )) is the simplicial nerve from Definition 4.53 of the unital A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ un ( A , B ) from Definitions 4.10–4.11, and ( ⋅ ) K a n (\mathord{\cdot})^{\mathrm{Kan}} ( ⋅ ) Kan denotes the maximal Kan complex in an ∞ \infty ∞ -category. We prove (9.1) by first showing that
Hom A ^ ∞ Cat R u n , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ u n ( A , B ) ) . (243) \operatorname{Hom}^{L}_{\widehat{\mathcal{A}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B}) \cong\mathcal{N}_{\mathcal{A}_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}_{\infty}}(\mathcal{A},\mathcal{B})). \tag*{(243)} Hom A ∞ Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ un ( A , B )) . (243)
Here we use the left morphisms Hom C L ( X , Y ) \operatorname{Hom}^{L}_{\mathcal{C}}(X,Y) Hom C L ( X , Y ) defined as in Definition 2.15, but in the simplicial set A ^ ∞ Cat R u n , ∞ \widehat{\mathcal{A}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ from Definition 7.7, which is not an ∞ \infty ∞ -category, so that its left morphisms need not be a Kan complex. As ( A ^ ∞ Cat R u n , ∞ , T ) (\widehat{\mathcal{A}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) is an ∞ \infty ∞ -bicategory by Theorem 7.17, its left morphisms are an ∞ \infty ∞ -category.
(b) Now let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring, as in §5.2, and suppose A , B \mathcal{A},\mathcal{B} A , B are unital filtered A ∞ A_{\infty} A ∞ -categories over R R R . Then generalizing (9.1), there are canonical strict isomorphisms of Kan complexes
Hom A ∞ f i Cat R u n , ∞ L ( A , B ) ≅ N A ∞ ( F 0 Fun A ∞ f i u n ( A , B ) ) K a n , (244) \operatorname{Hom}^{L}_{\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B}) \cong\mathcal{N}_{\mathcal{A}_{\infty}}\left(\operatorname{F}^{0}\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B})\right)^{\mathrm{Kan}}, \tag*{(244)} Hom A ∞ fi Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( F 0 Fun A ∞ fi un ( A , B ) ) Kan , (244)
Hom A ∞ w f Cat R u n , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ f i u n ( A , B ) ) K a n , (245) \operatorname{Hom}^{L}_{\mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}}(\mathcal{A},\mathcal{B}) \cong\mathcal{N}_{\mathcal{A}_{\infty}}\left(\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B})\right)^{\mathrm{Kan}}, \tag*{(245)} Hom A ∞ wf Cat R un , ∞ L ( A , B ) ≅ N A ∞ ( Fun A ∞ fi un ( A , B ) ) Kan , (245)
where Fun A ∞ f i u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi un ( A , B ) , F 0 Fun A ∞ f i ( A , B ) \operatorname{F}^{0}\operatorname{Fun}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ fi ( A , B ) are as in Definition 5.19(e),(f). Note that as in Remark 5.20, Fun A ∞ f i u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B}) Fun A ∞ fi un ( A , B ) is an honest unital A ∞ A_{\infty} A ∞ -category over R R R (it is not curved, it has m 0 F ~ = 0 m^{\widetilde{\mathcal{F}}}_{0}=0 m 0 F = 0 for all objects F ~ \widetilde{\mathcal{F}} F ), and F 0 Fun A ∞ f i u n ( A , B ) \operatorname{F}^{0}\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B}) F 0 Fun A ∞ fi un ( A , B ) is a unital A ∞ A_{\infty} A ∞ -category over F 0 R \operatorname{F}^{0}R F 0 R , so N A ∞ ( Fun A ∞ f i u n ( A , B ) ) \mathcal{N}_{\mathcal{A}_{\infty}}(\operatorname{Fun}^{\mathrm{un}}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B})) N A ∞ ( Fun A ∞ fi un ( A , B )) , N A ∞ ( F 0 Fun A ∞ f i ( A , B ) ) \mathcal{N}_{\mathcal{A}_{\infty}}(\operatorname{F}^{0}\operatorname{Fun}_{\mathcal{A}^{\mathrm{fi}}_{\infty}}(\mathcal{A},\mathcal{B})) N A ∞ ( F 0 Fun A ∞ fi ( A , B )) make sense. The analogues of (9.2) also hold.
The author views Theorem 9.1 as evidence that A ∞ Cat R u n , ∞ \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , A ∞ f i Cat R u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , A ∞ w f Cat R u n , ∞ \mathcal{A}^{\mathrm{wf}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ wf Cat R un , ∞ are the ‘right’ definitions of ∞ \infty ∞ -categories generalizing A ∞ Cat R u n \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un , A ∞ f i Cat R \mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}_{R} A ∞ fi Cat R . In particular, (9.1)–(9.4) hold not just up to homotopy equivalence, but on the nose. Combining Definition 7.34 and Theorem 9.1(a), we can also describe mapping spaces in the ∞ \infty ∞ -category DGCat R ∞ \operatorname{DGCat}^{\infty}_{R} DGCat R ∞ from Definition 7.34:
Corollary 9.2. Let A , B \mathcal{A},\mathcal{B} A , B be dg-categories over a commutative ring R R R . Then there is a canonical strict isomorphism of Kan complexes
Hom DGCat R ∞ L ( A , B ) ≅ N A ∞ ( Fun u n ( A , B ) d g ) K a n . \operatorname{Hom}^{L}_{\operatorname{DGCat}^{\infty}_{R}}(\mathcal{A},\mathcal{B}) \cong\mathcal{N}_{\mathcal{A}_{\infty}}\left(\operatorname{Fun}^{\mathrm{un}}(\mathcal{A},\mathcal{B})^{\mathrm{dg}}\right)^{\mathrm{Kan}}. Hom DGCat R ∞ L ( A , B ) ≅ N A ∞ ( Fun un ( A , B ) dg ) Kan . Here Fun u n ( A , B ) d g \operatorname{Fun}^{\mathrm{un}}(\mathcal{A},\mathcal{B})^{\mathrm{dg}} Fun un ( A , B ) dg is the full A ∞ A_{\infty} A ∞ -subcategory of Fun u n ( A , B ) \operatorname{Fun}^{\mathrm{un}}(\mathcal{A},\mathcal{B}) Fun un ( A , B ) with objects dg-functors F ~ : A → B \widetilde{\mathcal{F}}:\mathcal{A}\to\mathcal{B} F : A → B .
Comparing our ∞ \infty ∞ -categories with localizations
Definition 7.30 gives an ∞ \infty ∞ -functor inc R u n : N ( A ∞ Cat R u n ) → A ∞ Cat R u n , ∞ \operatorname{inc}^{\mathrm{un}}_{R}:\mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R})\to\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} inc R un : N ( A ∞ Cat R un ) → A ∞ Cat R un , ∞ , which maps unital homotopy equivalences in A ∞ Cat R u n \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un in Definition 4.9 to equivalences in A ∞ C a t R u n , ∞ \mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ by Remark 7.31. Thus by the universal property of ∞ \infty ∞ -category localizations in § 2.6 , there exists an ∞ \infty ∞ -functor Φ \Phi Φ in the diagram
N ( A ∞ C a t R u n ) → Π l o c ∞ A ∞ C a t R u n , ∞ [ H E u n − 1 ] ∞ ↘ i n c R u n ↓ Φ A ∞ C a t R u n , ∞ (246) \begin{aligned}
\mathcal{N}\left(\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}}\right)&\xrightarrow{\Pi_{\mathrm{loc}}^{\infty}}\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\left[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}\right]_{\infty}\\
&\searrow^{\mathrm{inc}_{R}^{\mathrm{un}}}\qquad\qquad\qquad\qquad\qquad\quad\downarrow^{\Phi}\\
&\qquad\qquad\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} \tag*{(246)}
\end{aligned} N ( A ∞ Cat R un ) Π loc ∞ A ∞ Cat R un , ∞ [ H E un − 1 ] ∞ ↘ inc R un ↓ Φ A ∞ Cat R un , ∞ (246)
unique up to homotopy, where H E u n \mathcal{H}\mathcal{E}_{\mathrm{un}} H E un is the class of unital homotopy equivalences in A ∞ C a t R u n \mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un , in the sense of Definition 4.27.
Theorem 9.7 will show that Φ \Phi Φ is an equivalence of ∞ \infty ∞ -categories. To do this, we will show that Φ \Phi Φ induces homotopy equivalences of mapping spaces. Mapping spaces in A ∞ C a t R u n , ∞ \mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ are given in Theorem 9.1. To describe mapping spaces in A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}}\left[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}\right]_{\infty} A ∞ Cat R un [ H E un − 1 ] ∞ , in the next definition and theorem we will construct simplicial resolutions of objects in ( A ∞ C a t R u n , H E u n ) (\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}},\mathcal{H}\mathcal{E}_{\mathrm{un}}) ( A ∞ Cat R un , H E un ) as in Definition 2.36, and then apply Theorem 2.37.
Definition 9.3. Let R R R be a commutative ring, and B \mathcal{B} B an object in A ∞ C a t R u n \mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un . We will define a functor S B : Δ o p → A ∞ C a t R u n S_{\mathcal{B}}:\Delta^{\mathrm{op}}\to\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} S B : Δ op → A ∞ Cat R un and an isomorphism i B : B → S B ( [ 0 ] ) i_{\mathcal{B}}:\mathcal{B}\to S_{\mathcal{B}}([0]) i B : B → S B ([ 0 ]) . For each N ≥ 0 N\geq0 N ≥ 0 , Definition 4.53 gives a unital A ∞ A_{\infty} A ∞ -category C [ N ] Δ \mathcal{C}^{\Delta}_{[N]} C [ N ] Δ over R R R . This has objects 0 , … , N 0,\ldots,N 0 , … , N , and for all 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N there is a distinguished closed degree 0 morphism ω i j : i → j \omega_{ij}:i\to j ω ij : i → j in C [ N ] Δ \mathcal{C}^{\Delta}_{[N]} C [ N ] Δ , with ω i i = 1 i \omega_{ii}=1_i ω ii = 1 i .
Definitions 4.10–4.11 define a unital A ∞ A_{\infty} A ∞ -category Fun A ∞ u n ( C [ N ] Δ , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[N]},\mathcal{B}) Fun A ∞ un ( C [ N ] Δ , B ) whose objects are unital A ∞ A_{\infty} A ∞ -functors F : C [ N ] Δ → B \mathcal{F}:\mathcal{C}^{\Delta}_{[N]}\to\mathcal{B} F : C [ N ] Δ → B . For the object [ N ] [N] [ N ] in Δ o p \Delta^{\mathrm{op}} Δ op , define S B ( [ N ] ) ∈ A ∞ C a t R u n S_{\mathcal{B}}([N])\in\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} S B ([ N ]) ∈ A ∞ Cat R un to be the full A ∞ A_{\infty} A ∞ -subcategory of Fun A ∞ u n ( C [ N ] Δ , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[N]},\mathcal{B}) Fun A ∞ un ( C [ N ] Δ , B ) with objects those F : C [ N ] Δ → B \mathcal{F}:\mathcal{C}^{\Delta}_{[N]}\to\mathcal{B} F : C [ N ] Δ → B such that if 0 ≤ i < j ≤ N 0\leq i<j\leq N 0 ≤ i < j ≤ N then F ( ω i j ) \mathcal{F}(\omega_{ij}) F ( ω ij ) is a homotopy equivalence in B \mathcal{B} B , in the sense of Definition 4.8.
If α : [ N ] → [ M ] \alpha:[N]\to[M] α : [ N ] → [ M ] is a morphism in Δ \Delta Δ then Definition 4.53 gives a unital A ∞ A_{\infty} A ∞ -functor C [ N ] Δ → C [ M ] Δ \mathcal{C}^{\Delta}_{[N]}\to\mathcal{C}^{\Delta}_{[M]} C [ N ] Δ → C [ M ] Δ . By Definition 4.13(a),(e) this induces a unital A ∞ A_{\infty} A ∞ -functor ( C α Δ ) ∗ : Fun A ∞ u n ( C [ M ] Δ , B ) → Fun A ∞ u n ( C [ N ] Δ , B ) (\mathcal{C}^{\Delta}_{\alpha})^{*}:\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[M]},\mathcal{B})\to\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[N]},\mathcal{B}) ( C α Δ ) ∗ : Fun A ∞ un ( C [ M ] Δ , B ) → Fun A ∞ un ( C [ N ] Δ , B ) . Since C α Δ ( ω i j ) = ω α ( i ) , α ( j ) \mathcal{C}^{\Delta}_{\alpha}(\omega_{ij})=\omega_{\alpha(i),\alpha(j)} C α Δ ( ω ij ) = ω α ( i ) , α ( j ) , this ( C α Δ ) ∗ (\mathcal{C}^{\Delta}_{\alpha})^{*} ( C α Δ ) ∗ restricts to a unital A ∞ A_{\infty} A ∞ -functor S B ( [ M ] ) → S B ( [ N ] ) S_{\mathcal{B}}([M])\to S_{\mathcal{B}}([N]) S B ([ M ]) → S B ([ N ]) . Define S B ( α ) = ( C α Δ ) ∗ ∣ S B ( [ M ] ) : S B ( [ M ] ) → S B ( [ N ] ) S_{\mathcal{B}}(\alpha)=(\mathcal{C}^{\Delta}_{\alpha})^{*}|_{S_{\mathcal{B}}([M])}:S_{\mathcal{B}}([M])\to S_{\mathcal{B}}([N]) S B ( α ) = ( C α Δ ) ∗ ∣ S B ([ M ]) : S B ([ M ]) → S B ([ N ]) . The functoriality of the C α Δ \mathcal{C}^{\Delta}_{\alpha} C α Δ in Definition 4.53 and of pullbacks imply that S B : Δ o p → A ∞ C a t R u n S_{\mathcal{B}}:\Delta^{\mathrm{op}}\to\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} S B : Δ op → A ∞ Cat R un is a functor.
Now C [ 0 ] Δ \mathcal{C}^{\Delta}_{[0]} C [ 0 ] Δ is isomorphic to the unital point A ∞ A_{\infty} A ∞ -category P u n \mathcal{P}^{\mathrm{un}} P un in Definition 4.17, so Proposition 4.18 gives a canonical isomorphism B → Fun A ∞ u n ( C [ 0 ] Δ , B ) \mathcal{B}\to\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[0]},\mathcal{B}) B → Fun A ∞ un ( C [ 0 ] Δ , B ) . As S B ( [ 0 ] ) = Fun A ∞ u n ( C [ 0 ] Δ , B ) S_{\mathcal{B}}([0])=\operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[0]},\mathcal{B}) S B ([ 0 ]) = Fun A ∞ un ( C [ 0 ] Δ , B ) , we define i B : B → S B ( [ 0 ] ) i_{\mathcal{B}}:\mathcal{B}\to S_{\mathcal{B}}([0]) i B : B → S B ([ 0 ]) to be this.
Now suppose that B \mathcal{B} B lies in D G C a t R ⊂ A ∞ C a t R u n \mathbf{DGCat}_{R}\subset\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} DGCat R ⊂ A ∞ Cat R un . As in Definition 4.53, the A ∞ A_{\infty} A ∞ -categories C [ N ] Δ \mathcal{C}^{\Delta}_{[N]} C [ N ] Δ are actually dg-categories. Hence Fun A ∞ u n ( C [ N ] Δ , B ) \operatorname{Fun}^{\mathrm{un}}_{A_{\infty}}(\mathcal{C}^{\Delta}_{[N]},\mathcal{B}) Fun A ∞ un ( C [ N ] Δ , B ) is a dg-category, as in Remark 4.34. It is immediate from Definition 4.13(a) that the A ∞ A_{\infty} A ∞ -functors ( C α Δ ) ∗ (\mathcal{C}^{\Delta}_{\alpha})^{*} ( C α Δ ) ∗ above are dg-functors. Thus S B S_{\mathcal{B}} S B is a functor S B : Δ o p → D G C a t R ⊂ A ∞ C a t R u n S_{\mathcal{B}}:\Delta^{\mathrm{op}}\to\mathbf{DGCat}_{R}\subset\mathsf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} S B : Δ op → DGCat R ⊂ A ∞ Cat R un .
The next theorem will be proved in §11.2. Note that (a),(b) refer to two different, though closely related, notions of simplicial resolution.
Theorem 9.4. Work in the situation of Definition 9.3.
(a) ( S B , i B ) (S_{\mathcal{B}},i_{\mathcal{B}}) ( S B , i B ) is a simplicial resolution of B \mathcal{B} B in ( A ∞ C a t R u n , H E u n ) (\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}},\mathcal{H}\mathcal{E}_{\mathrm{un}}) ( A ∞ Cat R un , H E un ) , in the sense of Definition 2.36.
(b) Suppose B \mathcal{B} B lies in D G C a t R ⊂ A ∞ C a t R u n \mathsf{DGCat}_{R}\subset\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} DGCat R ⊂ A ∞ Cat R un , and is fibrant in the Tabuada model structure on D G C a t R \mathsf{DGCat}_{R} DGCat R from Theorem 4.35 . Then ( S B , i B ) (S_{\mathcal{B}},i_{\mathcal{B}}) ( S B , i B ) is a simplicial resolution of B \mathcal{B} B in the model category D G C a t R \mathsf{DGCat}_{R} DGCat R , in the sense of Remark 2.35(c). This simplicial resolution was constructed by Holstein [43, §3].
(c) Define F S B : ( A ∞ C a t R u n ) o p → S S e t F_{S_{\mathcal{B}}}\colon(\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\to\mathsf{SSet} F S B : ( A ∞ Cat R un ) op → SSet as in Definition 2.36. Then for each unital A ∞ A_{\infty} A ∞ -category A \mathcal{A} A over R R R there is a canonical strict isomorphism of simplicial sets
F S B u n ( A ) ≅ N A ∞ ( Fun A ∞ u n ( A , B ) ) K a n . (247) F_{S_{\mathcal{B}}}^{\mathrm{un}}(\mathcal{A})\cong\mathcal{N}_{A_{\infty}}\bigl(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\bigr)^{\mathrm{Kan}}. \tag*{(247)} F S B un ( A ) ≅ N A ∞ ( Fun A ∞ un ( A , B ) ) Kan . (247) If F : A 1 → A 2 \mathcal{F}\colon\mathcal{A}_{1}\to\mathcal{A}_{2} F : A 1 → A 2 is a morphism in A ∞ C a t R u n \mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un then the isomorphism (9.6) identifies F S B u n ( F ) : F S B u n ( A 2 ) → F S B u n ( A 1 ) F_{S_{\mathcal{B}}}^{\mathrm{un}}(\mathcal{F})\colon F_{S_{\mathcal{B}}}^{\mathrm{un}}(\mathcal{A}_{2})\to F_{S_{\mathcal{B}}}^{\mathrm{un}}(\mathcal{A}_{1}) F S B un ( F ) : F S B un ( A 2 ) → F S B un ( A 1 ) with
N A ∞ ( F ∗ ) K a n : N A ∞ ( Fun A ∞ u n ( A 2 , B ) ) K a n ⟶ N A ∞ ( Fun A ∞ u n ( A 1 , B ) ) K a n , (248) \mathcal{N}_{A_{\infty}}(\mathcal{F}^{*})^{\mathrm{Kan}}\colon\mathcal{N}_{A_{\infty}}\bigl(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{2},\mathcal{B})\bigr)^{\mathrm{Kan}}\longrightarrow\mathcal{N}_{A_{\infty}}\bigl(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{1},\mathcal{B})\bigr)^{\mathrm{Kan}}, \tag*{(248)} N A ∞ ( F ∗ ) Kan : N A ∞ ( Fun A ∞ un ( A 2 , B ) ) Kan ⟶ N A ∞ ( Fun A ∞ un ( A 1 , B ) ) Kan , (248) where F ∗ : Fun A ∞ u n ( A 2 , B ) → Fun A ∞ u n ( A 1 , B ) \mathcal{F}^{*}\colon\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{2},\mathcal{B})\to\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A}_{1},\mathcal{B}) F ∗ : Fun A ∞ un ( A 2 , B ) → Fun A ∞ un ( A 1 , B ) is as in Definition 4.13(a),(e).
(d) Combining (a),(c) and Theorem 2.37 gives a homotopy equivalence for all unital A ∞ A_{\infty} A ∞ -categories A , B \mathcal{A},\mathcal{B} A , B over R R R
Hom A ∞ C a t R u n [ H E u n − 1 ] ∞ ( A , B ) ≃ N A ∞ ( Fun A ∞ u n ( A , B ) ) K a n . (249) \operatorname{Hom}_{\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty}(\mathcal{A},\mathcal{B})\simeq\mathcal{N}_{A_{\infty}}\bigl(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{A},\mathcal{B})\bigr)^{\mathrm{Kan}}.} \tag*{(249)} Hom A ∞ Cat R un [ H E un − 1 ] ∞ ( A , B ) ≃ N A ∞ ( Fun A ∞ un ( A , B ) ) Kan . (249) (e) Suppose C \mathcal{C} C is any of the full subcategories of A ∞ C a t R u n \mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un
D G C a t R A ∞ , D G C a t R A ∞ , s u , D G C a t R A ∞ , h p , D G C a t R A ∞ , h s , \mathsf{DGCat}_{R}^{A_{\infty}},\quad\mathsf{DGCat}_{R}^{A_{\infty},\mathrm{su}},\quad\mathsf{DGCat}_{R}^{A_{\infty},\mathrm{hp}},\quad\mathsf{DGCat}_{R}^{A_{\infty},\mathrm{hs}}, DGCat R A ∞ , DGCat R A ∞ , su , DGCat R A ∞ , hp , DGCat R A ∞ , hs , A ∞ C a t R u n , s u , A ∞ C a t R u n , h p , A ∞ C a t R u n , h s , A ∞ △ C a t R u n , \mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un,su}},\quad\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un,hp}},\quad\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un,hs}},\quad\mathsf{A}_{\infty}^{\triangle}\mathsf{Cat}_{R}^{\mathrm{un}}, A ∞ Cat R un , su , A ∞ Cat R un , hp , A ∞ Cat R un , hs , A ∞ △ Cat R un , A ∞ △ C a t R u n , s u , A ∞ △ C a t R u n , h p , A ∞ △ C a t R u n , h s , (250) \mathsf{A}_{\infty}^{\triangle}\mathsf{Cat}_{R}^{\mathrm{un,su}},\quad\mathsf{A}_{\infty}^{\triangle}\mathsf{Cat}_{R}^{\mathrm{un,hp}},\quad\mathsf{A}_{\infty}^{\triangle}\mathsf{Cat}_{R}^{\mathrm{un,hs}}, \tag*{(250)} A ∞ △ Cat R un , su , A ∞ △ Cat R un , hp , A ∞ △ Cat R un , hs , (250) from Definitions 4.32, 4.47, and 4.60, and let B \mathcal{B} B be an object of C \mathcal{C} C . Then S B S_{\mathcal{B}} S B maps to C ⊂ A ∞ C a t R u n \mathcal{C}\subset\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} C ⊂ A ∞ Cat R un , so that ( S B , i B ) (S_{\mathcal{B}},i_{\mathcal{B}}) ( S B , i B ) is also a simplicial resolution of B \mathcal{B} B in ( C , H E u n ∩ C ) (\mathcal{C},\mathcal{H}\mathcal{E}_{\mathrm{un}}\cap\mathcal{C}) ( C , H E un ∩ C ) , in the sense of Definition 2.36. Hence the computation of mapping spaces (9.8) also holds in the localized ∞ \infty ∞ -category C [ ( H E u n ∩ C ) − 1 ] ∞ \mathcal{C}[(\mathcal{H}\mathcal{E}_{\mathrm{un}}\cap\mathcal{C})^{-1}]_{\infty} C [( H E un ∩ C ) − 1 ] ∞ . This implies that the natural morphism C [ ( H E u n ∩ C ) − 1 ] ∞ → A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathcal{C}[(\mathcal{H}\mathcal{E}_{\mathrm{un}}\cap\mathcal{C})^{-1}]_{\infty}\to\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_{\infty} C [( H E un ∩ C ) − 1 ] ∞ → A ∞ Cat R un [ H E un − 1 ] ∞ is fully faithful.
Remark 9.5. (a) The simplicial resolution ( S B , i B ) (S_{\mathcal{B}},i_{\mathcal{B}}) ( S B , i B ) in Definition 9.3 is functorial in B \mathcal{B} B over A ∞ C a t R u n \mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un . We can define a functor S : Δ o p × A ∞ C a t R u n → A ∞ C a t R u n S\colon\Delta^{\mathrm{op}}\times\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}}\to\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}} S : Δ op × A ∞ Cat R un → A ∞ Cat R un and a natural isomorphism i : Id A ∞ C a t R u n → S ∣ { [ 0 ] } × A ∞ C a t R u n i\colon\operatorname{Id}_{\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}}\to S|_{\{[0]\}\times\mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un}}}} i : Id A ∞ Cat R un → S ∣ {[ 0 ]} × A ∞ Cat R un with S ∣ Δ o p × B = S B S|_{\Delta^{\mathrm{op}}\times\mathcal{B}}=S_{\mathcal{B}} S ∣ Δ op × B = S B and i ∣ B = i B i|_{\mathcal{B}}=i_{\mathcal{B}} i ∣ B = i B for all B \mathcal{B} B . But we will not need this.
(b) Although it is a ‘classical’ result, our proof of Theorem 9.4 (a) uses Theorems 7.13 and 9.1 (a) on our new ∞ \infty ∞ -category A ∞ C a t R u n , ∞ \mathsf{A}_{\infty}\mathsf{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ .
The next theorem will be proved in §11.3. It proves equation (4.47) in Remark 4.69.
Theorem 9.6. Let R R R be a commutative ring. Then inclusions of subcategories between the categories D G C a t R h s \mathrm{DGCat}^{\mathrm{hs}}_{R} DGCat R hs , D G C a t R \mathrm{DGCat}_{R} DGCat R , D G C a t R A ∞ , h s \mathrm{DGCat}^{A_{\infty},\mathrm{hs}}_{R} DGCat R A ∞ , hs , D G C a t R A ∞ \mathrm{DGCat}^{A_{\infty}}_{R} DGCat R A ∞ , A ∞ C a t R u n , h s A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R} A ∞ Cat R un , hs , A ∞ C a t R u n A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un from § 4 , and between the corresponding classes of homotopy equivalences H E ⋯ \mathcal{HE}_{\cdots} HE ⋯ or quasi-equivalences Q E ⋯ \mathcal{QE}_{\cdots} QE ⋯ , induce a homotopy commutative diagram of ∞ \infty ∞ -functors of localizations:
D G C a t R h s [ H E d g , h s − 1 ] ∞ → ≃ D G C a t R [ Q E d g − 1 ] ∞ ↓ ↓ ≃ D G C a t R A ∞ , h s [ H E d g , A ∞ , h s − 1 ] ∞ → D G C a t R A ∞ [ Q E d g , A ∞ − 1 ] ∞ ↓ ↓ ≃ A ∞ C a t R u n , h s [ H E u n , h s − 1 ] ∞ → A ∞ C a t R u n [ Q E u n − 1 ] ∞ (251) \begin{CD}
\mathrm{DGCat}^{\mathrm{hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{dg,hs}}]^{\infty} @>{\simeq}>> \mathrm{DGCat}_{R}[\mathcal{QE}^{-1}_{\mathrm{dg}}]^{\infty} \\
@VVV @VV{\simeq}V \\
\mathrm{DGCat}^{A_{\infty},\mathrm{hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{dg},A_{\infty},\mathrm{hs}}]^{\infty} @>>> \mathrm{DGCat}^{A_{\infty}}_{R}[\mathcal{QE}^{-1}_{\mathrm{dg},A_{\infty}}]^{\infty} \\
@VVV @VV{\simeq}V \\
A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{un,hs}}]^{\infty} @>>> A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{QE}^{-1}_{\mathrm{un}}]^{\infty} \tag*{(251)}
\end{CD} DGCat R hs [ HE dg , hs − 1 ] ∞ ↓ ⏐ DGCat R A ∞ , hs [ HE dg , A ∞ , hs − 1 ] ∞ ↓ ⏐ A ∞ Cat R un , hs [ HE un , hs − 1 ] ∞ ≃ DGCat R [ QE dg − 1 ] ∞ ↓ ⏐ ≃ DGCat R A ∞ [ QE dg , A ∞ − 1 ] ∞ ↓ ⏐ ≃ A ∞ Cat R un [ QE un − 1 ] ∞ (251) The top and right morphisms in the diagram, written ‘⟶ \longrightarrow ⟶ ’, are Joyal equivalences by Theorems 4.64 and 4.68, as in equations (130) and (132) . The other morphisms in (251) , written ‘⇢ \dashrightarrow ⇢ ’, are also Joyal equivalences.
The next theorem will be proved in §11.4.
Theorem 9.7. Let R R R be a commutative ring. Then the ∞ \infty ∞ -functor from equation (246) is a Joyal equivalence:
Φ : A ∞ C a t R u n [ H E u n − 1 ] ∞ ⟶ A ∞ C a t R u n , ∞ . (252) \Phi\colon A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{HE}^{-1}_{\mathrm{un}}]^{\infty}\longrightarrow A_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}. \tag*{(252)} Φ : A ∞ Cat R un [ HE un − 1 ] ∞ ⟶ A ∞ Cat R un , ∞ . (252) The analogue holds for all the full subcategories of A ∞ C a t R u n A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un in (250) , giving Joyal equivalences of their localizations with the full ∞ \infty ∞ -subcategories of A ∞ C a t R u n , ∞ A_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ with the same objects. So for example we have equivalences
A ∞ C a t R u n , h s [ H E u n , h s − 1 ] ∞ ≃ A ∞ C a t R u n , h s , ∞ , A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{un,hs}}]^{\infty}\simeq A_{\infty}\mathrm{Cat}^{\mathrm{un,hs},\infty}_{R}, A ∞ Cat R un , hs [ HE un , hs − 1 ] ∞ ≃ A ∞ Cat R un , hs , ∞ , A ∞ C a t R u n , s u [ H E u n , s u − 1 ] ∞ ≃ A ∞ C a t R u n , s u , ∞ , A_{\infty}\mathrm{Cat}^{\mathrm{un,su}}_{R}[\mathcal{HE}^{-1}_{\mathrm{un,su}}]^{\infty}\simeq A_{\infty}\mathrm{Cat}^{\mathrm{un,su},\infty}_{R}, A ∞ Cat R un , su [ HE un , su − 1 ] ∞ ≃ A ∞ Cat R un , su , ∞ , A ∞ C a t R Δ [ H E u n , Δ − 1 ] ∞ ≃ A ∞ Δ C a t R u n , ∞ . (253) A_{\infty}\mathrm{Cat}^{\Delta}_{R}[\mathcal{HE}^{-1}_{\mathrm{un},\Delta}]^{\infty}\simeq A^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}. \tag*{(253)} A ∞ Cat R Δ [ HE un , Δ − 1 ] ∞ ≃ A ∞ Δ Cat R un , ∞ . (253)
Combining Theorems 4.64, 4.68, 7.35, 9.6, 9.7 and Corollary 4.65, we deduce:
Corollary 9.8. (a) Let R R R be a commutative ring. Then all the following ∞ \infty ∞ -categories are Joyal equivalent:
A ∞ C a t R u n , h s , ∞ , D G C a t R A ∞ , h s , ∞ , A ∞ C a t R u n [ Q E u n − 1 ] ∞ , A ∞ C a t R c u [ Q E c u − 1 ] ∞ , A ∞ C a t R w c [ Q E w c − 1 ] ∞ , D G C a t R [ Q E d g − 1 ] ∞ , D G C a t R A ∞ [ Q E d g , A ∞ − 1 ] ∞ , A ∞ C a t R u n , h s [ H E u n , h s − 1 ] ∞ , D G C a t R h s [ H E d g , h s − 1 ] ∞ , D G C a t R A ∞ , h s [ H E A ∞ , d g , h s − 1 ] ∞ . (254) \begin{aligned}
&A_{\infty}\mathrm{Cat}^{\mathrm{un,hs},\infty}_{R},\quad
\mathrm{DGCat}^{A_{\infty},\mathrm{hs},\infty}_{R},\quad
A_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{QE}^{-1}_{\mathrm{un}}]^{\infty},\\
&A_{\infty}\mathrm{Cat}^{\mathrm{cu}}_{R}[\mathcal{QE}^{-1}_{\mathrm{cu}}]^{\infty},\quad
A_{\infty}\mathrm{Cat}^{\mathrm{wc}}_{R}[\mathcal{QE}^{-1}_{\mathrm{wc}}]^{\infty},\quad
\mathrm{DGCat}_{R}[\mathcal{QE}^{-1}_{\mathrm{dg}}]^{\infty},\\
&\mathrm{DGCat}^{A_{\infty}}_{R}[\mathcal{QE}^{-1}_{\mathrm{dg},A_{\infty}}]^{\infty},\quad
A_{\infty}\mathrm{Cat}^{\mathrm{un,hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{un,hs}}]^{\infty},\\
&\mathrm{DGCat}^{\mathrm{hs}}_{R}[\mathcal{HE}^{-1}_{\mathrm{dg,hs}}]^{\infty},\quad
\mathrm{DGCat}^{A_{\infty},\mathrm{hs}}_{R}[\mathcal{HE}^{-1}_{A_{\infty,\mathrm{dg,hs}}}]^{\infty}. \tag*{(254)}
\end{aligned} A ∞ Cat R un , hs , ∞ , DGCat R A ∞ , hs , ∞ , A ∞ Cat R un [ QE un − 1 ] ∞ , A ∞ Cat R cu [ QE cu − 1 ] ∞ , A ∞ Cat R wc [ QE wc − 1 ] ∞ , DGCat R [ QE dg − 1 ] ∞ , DGCat R A ∞ [ QE dg , A ∞ − 1 ] ∞ , A ∞ Cat R un , hs [ HE un , hs − 1 ] ∞ , DGCat R hs [ HE dg , hs − 1 ] ∞ , DGCat R A ∞ , hs [ HE A ∞ , dg , hs − 1 ] ∞ . (254) They satisfy the conclusions of Corollary 4.65, so they are presentable and closed symmetric monoidal.
(b) The following ∞ \infty ∞ -categories are Joyal equivalent:
D G C a t R A ∞ , s u , ∞ , A ∞ C a t R u n , s u , ∞ , \mathrm{DGCat}_{R}^{A_{\infty},\mathrm{su},\infty},\quad A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\mathrm{su},\infty}, DGCat R A ∞ , su , ∞ , A ∞ Cat R un , su , ∞ ,
D G C a t R A ∞ , s u [ H E d g , A ∞ , s u − 1 ] ∞ , A ∞ C a t R u n , s u [ H E u n , s u − 1 ] ∞ . (255) \mathrm{DGCat}_{R}^{A_{\infty},\mathrm{su}}[\mathcal{H}E^{-1}_{\mathrm{dg},A_{\infty},\mathrm{su}}]^{\infty},\quad A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\mathrm{su}}[\mathcal{H}E^{-1}_{\mathrm{un},\mathrm{su}}]^{\infty}. \tag*{(255)} DGCat R A ∞ , su [ H E dg , A ∞ , su − 1 ] ∞ , A ∞ Cat R un , su [ H E un , su − 1 ] ∞ . (255)
Remark 9.9. The ∞ \infty ∞ -categories A ∞ C a t R u n [ Q u n − 1 ] ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[Q_{\mathrm{un}}^{-1}]^{\infty} A ∞ Cat R un [ Q un − 1 ] ∞ and A ∞ C a t R c u [ Q c u − 1 ] ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{cu}}[Q_{\mathrm{cu}}^{-1}]^{\infty} A ∞ Cat R cu [ Q cu − 1 ] ∞ in (9.13) are considered in the literature [11, 12, 24, 72, 73, 76, 88] to be ‘the’ ∞ \infty ∞ -category of A ∞ A_{\infty} A ∞ -categories over R R R , as in §4.9. We have shown that these can be identified with the full ∞ \infty ∞ -subcategory A ∞ C a t R u n , h s , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\mathrm{hs},\infty} A ∞ Cat R un , hs , ∞ of our explicit ∞ \infty ∞ -category A ∞ C a t R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ from Chapter 7.
Remark 9.10. We did not treat filtered A ∞ A_{\infty} A ∞ -categories from Chapter 8 in the material above. This is because the method of Definition 9.3 does not work to define simplicial resolutions in the filtered case; to define the isomorphism i B : B → S B ( [ 0 ] ) i_{\mathcal{B}}:\mathcal{B}\to S_{\mathcal{B}}([0]) i B : B → S B ([ 0 ]) we used that C Δ D ≅ P u n \mathcal{C}_{\Delta}^{\mathrm{D}}\cong\mathcal{P}^{\mathrm{un}} C Δ D ≅ P un and Fun A ∞ u n ( P u n , B ) ≅ B \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},\mathcal{B})\cong\mathcal{B} Fun A ∞ un ( P un , B ) ≅ B , but as in Proposition 5.29, in the filtered case we have Fun A ∞ u n ( P u n , B ) ≅ Str ( B ) \operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},\mathcal{B})\cong\operatorname{Str}(\mathcal{B}) Fun A ∞ un ( P un , B ) ≅ Str ( B ) . The author experimented with other methods to define simplicial resolutions in ( A ∞ C a t R u n , H E u n , f i ) (A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}},\mathcal{H}E_{\mathrm{un},\mathrm{fi}}) ( A ∞ Cat R un , H E un , fi ) , but did not succeed. It seems plausible that
A ∞ f i C a t R u n [ H E u n , f i − 1 ] ∞ ≃ A ∞ f i C a t R u n , ∞ , A_{\infty}^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathcal{H}E_{\mathrm{un},\mathrm{fi}}^{-1}]^{\infty}\simeq A_{\infty}^{\mathrm{fi}}\mathrm{Cat}_{R}^{\mathrm{un},\infty}, A ∞ fi Cat R un [ H E un , fi − 1 ] ∞ ≃ A ∞ fi Cat R un , ∞ ,
but the author does not have a proof of this.
The ∞ \infty ∞ -categorical simplicial nerve functor
This section will construct an ∞ \infty ∞ -functor N A ∞ ∞ : A ∞ C a t R u n , ∞ → C a t ∞ \mathcal{N}_{A_{\infty}}^{\infty}:A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\to\mathcal{C}\mathrm{at}_{\infty} N A ∞ ∞ : A ∞ Cat R un , ∞ → C at ∞ which generalizes the simplicial nerve functor N A ∞ : A ∞ C a t R u n → C a t ∞ \mathcal{N}_{A_{\infty}}:A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}\to\mathcal{C}\mathrm{at}_{\infty} N A ∞ : A ∞ Cat R un → C at ∞ from §4.8. Recall that Definition 4.17 defined the unital point A ∞ A_{\infty} A ∞ -category P u n \mathcal{P}^{\mathrm{un}} P un , and Proposition 4.18 showed that A ≅ Fun A ∞ u n ( P u n , A ) \mathcal{A}\cong\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) A ≅ Fun A ∞ un ( P un , A ) for any unital A ∞ A_{\infty} A ∞ -category A \mathcal{A} A . Combining (9.2) and Proposition 4.18, we see that if we could define an ∞ \infty ∞ -functor N A ∞ ∞ : A ∞ C a t R u n , ∞ → C a t ∞ \mathcal{N}_{A_{\infty}}^{\infty}:A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\to\mathcal{C}\mathrm{at}_{\infty} N A ∞ ∞ : A ∞ Cat R un , ∞ → C at ∞ mapping A ↦ Hom A ∞ C a t R u n , ∞ ^ L ( P u n , A ) \mathcal{A}\mapsto\operatorname{Hom}_{\widehat{A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}}}^{L}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) A ↦ Hom A ∞ Cat R un , ∞ L ( P un , A ) , then on objects it would map A ↦ N A ∞ ( Fun A ∞ u n ( P u n , A ) ) ≅ N A ∞ ( A ) \mathcal{A}\mapsto\mathcal{N}_{A_{\infty}}(\operatorname{Fun}_{A_{\infty}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}))\cong\mathcal{N}_{A_{\infty}}(\mathcal{A}) A ↦ N A ∞ ( Fun A ∞ un ( P un , A )) ≅ N A ∞ ( A ) , as we want. We do this using the ‘straightening construction’ of Theorem 2.23.
Definition 9.11. Let the simplicial sets A ∞ C a t R u n , ∞ ⊂ A ∞ C a t ^ R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\subset\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ be as in Definitions 7.7 and 7.12, for R R R a commutative ring. Then P u n \mathcal{P}^{\mathrm{un}} P un in Definition 4.17 is a 0-simplex in A ∞ C a t R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ . Define a simplicial set N ∞ C a t R u n , ∞ \mathcal{N}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} N ∞ Cat R un , ∞ and a map of simplicial sets p : N ∞ C a t R u n , ∞ → A ∞ C a t R u n , ∞ p:\mathcal{N}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\to A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} p : N ∞ Cat R un , ∞ → A ∞ Cat R un , ∞ , such that N N N -simplices of N ∞ C a t R u n , ∞ \mathcal{N}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} N ∞ Cat R un , ∞ are ( N + 1 ) (N+1) ( N + 1 ) -simplices f : Δ N + 1 → A ∞ C a t ^ R u n , ∞ f:\Delta^{N+1}\to\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} f : Δ N + 1 → A ∞ Cat R un , ∞ such that f ( Δ { 0 } ) = P u n f(\Delta^{\{0\}})=\mathcal{P}^{\mathrm{un}} f ( Δ { 0 } ) = P un and f ( Δ { 1 , 2 , … , N + 1 } ) f(\Delta^{\{1,2,\ldots,N+1\}}) f ( Δ { 1 , 2 , … , N + 1 } ) is an N N N -simplex in A ∞ C a t R u n , ∞ ⊂ A ∞ C a t ^ R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\subset\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ , and p p p maps f ↦ f ( Δ { 1 , 2 , … , N + 1 } ) f\mapsto f(\Delta^{\{1,2,\ldots,N+1\}}) f ↦ f ( Δ { 1 , 2 , … , N + 1 } ) . Maps between N N N -simplices corresponding to morphisms α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] in Δ \Delta Δ , to make N ∞ C a t R u n , ∞ \mathcal{N}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} N ∞ Cat R un , ∞ into a simplicial set, are defined as for Hom C L ( x , y ) \operatorname{Hom}_{\mathcal{C}}^{L}(x,y) Hom C L ( x , y ) in Definition 2.15.
This is an example of a well known construction: as in [60, §1.2.9] we have a simplicial set ( A ∞ C a t ^ R u n , ∞ ) P u n / (\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty})_{\mathcal{P}^{\mathrm{un}}/} ( A ∞ Cat R un , ∞ ) P un / with a morphism Π : ( A ∞ C a t ^ R u n , ∞ ) P u n / → A ∞ C a t ^ R u n , ∞ \Pi:(\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty})_{\mathcal{P}^{\mathrm{un}}/}\to\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} Π : ( A ∞ Cat R un , ∞ ) P un / → A ∞ Cat R un , ∞ , the undercategory of P u n \mathcal{P}^{\mathrm{un}} P un in A ∞ C a t ^ R u n , ∞ \widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , and p p p is the restriction of Π \Pi Π to the preimage of A ∞ C a t R u n , ∞ ⊂ A ∞ C a t ^ R u n , ∞ A_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\subset\widehat{A_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ .
The next theorem will be proved in §11.5.
Theorem 9.12. (a) In Definition 9.11, p : N C a t R u n , ∞ ⟶ A ∞ C a t R u n , ∞ p:\mathbf{N}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\longrightarrow\mathop{\mathbf{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} p : N Cat R un , ∞ ⟶ A ∞ Cat R un , ∞ is a cocartesian fibration in the sense of Definition 2.21. Therefore Theorem 2.23(b) defines an ∞ \infty ∞ -functor N A ∞ ∞ : A ∞ C a t R u n , ∞ ⟶ C a t ∞ \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty}:\mathop{\mathbf{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\longrightarrow\mathcal{C}\mathrm{at}_{\infty} N A ∞ ∞ : A ∞ Cat R un , ∞ ⟶ C at ∞ , the straightening of p p p , uniquely up to a contractible choice. We call N A ∞ ∞ \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty} N A ∞ ∞ the ∞ \infty ∞ -categorical simplicial nerve functor, as it extends N A ∞ \mathcal{N}_{\mathbf{A}_{\infty}} N A ∞ in §4.8 to ∞ \infty ∞ -categories.
(b) Let A \mathcal{A} A be an object of A ∞ C a t R u n , ∞ \mathop{\mathbf{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ . Then there is a strict isomorphism of ∞ \infty ∞ -categories p − 1 ( A ) ≅ N A ∞ ( A ) p^{-1}(\mathcal{A})\cong\mathcal{N}_{\mathbf{A}_{\infty}}(\mathcal{A}) p − 1 ( A ) ≅ N A ∞ ( A ) , where N A ∞ ( A ) \mathcal{N}_{\mathbf{A}_{\infty}}(\mathcal{A}) N A ∞ ( A ) is as in Definition 4.53. Hence by Theorem 2.23(a), N A ∞ ∞ ( A ) \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty}(\mathcal{A}) N A ∞ ∞ ( A ) is Joyal equivalent to N A ∞ ( A ) \mathcal{N}_{\mathbf{A}_{\infty}}(\mathcal{A}) N A ∞ ( A ) .
(c) The following diagram of ∞ \infty ∞ -categories commutes up to homotopy:
N ( A ∞ C a t R u n ) → N ( N A ∞ ) N ( C a t ∞ 1 ) i n c R u n ↓ ↓ I C a t ∞ A ∞ C a t R u n , ∞ → N A ∞ ∞ C a t ∞ (256) \begin{CD}
\mathcal{N}(\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) @>{\mathcal{N}(\mathcal{N}_{\mathbf{A}_{\infty}})}>> \mathcal{N}(\mathrm{Cat}_{\infty}^{1}) \\
@V{\mathrm{inc}_{R}^{\mathrm{un}}}VV @VV{I^{\mathrm{Cat}_{\infty}}}V \\
\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} @>{\mathcal{N}_{\mathbf{A}_{\infty}}^{\infty}}>> \mathcal{C}\mathrm{at}_{\infty} \tag*{(256)}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ N ( N A ∞ ) N A ∞ ∞ N ( Cat ∞ 1 ) ↓ ⏐ I Cat ∞ C at ∞ (256) using the notation of Definitions 2.18, 4.53 and 7.30. We prove this by showing that the pullback of p : N C a t R u n , ∞ ⟶ A ∞ C a t R u n , ∞ p:\mathbf{N}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\longrightarrow\mathop{\mathbf{A}_{\infty}\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} p : N Cat R un , ∞ ⟶ A ∞ Cat R un , ∞ to N ( A ∞ C a t R u n ) \mathcal{N}(\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) N ( A ∞ Cat R un ) is strictly isomorphic to the relative simplicial nerve π : N N A ∞ ( A ∞ C a t R u n ) ⟶ N ( A ∞ C a t R u n ) \pi:\mathcal{N}_{\mathcal{N}_{\mathbf{A}_{\infty}}}(\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}})\longrightarrow\mathcal{N}(\mathbf{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) π : N N A ∞ ( A ∞ Cat R un ) ⟶ N ( A ∞ Cat R un ) from Example 2.24, which is an explicit model for the unstraightening of I C a t ∞ ∘ N ( N A ∞ ) I^{\mathrm{Cat}_{\infty}}\circ\mathcal{N}(\mathcal{N}_{\mathbf{A}_{\infty}}) I Cat ∞ ∘ N ( N A ∞ ) in Theorem 2.23(c).
Combining Theorems 4.61 and 9.12 yields:
Corollary 9.13. In Theorem 9.12, N A ∞ ∞ \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty} N A ∞ ∞ restricts to
N A ∞ ∞ : A ∞ △ C a t R u n , ∞ ⟶ C a t ∞ s t (257) \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty}:\mathbf{A}_{\infty}^{\triangle}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\longrightarrow\mathcal{C}\mathrm{at}_{\infty}^{\mathrm{st}} \tag*{(257)} N A ∞ ∞ : A ∞ △ Cat R un , ∞ ⟶ C at ∞ st (257) from the ∞ \infty ∞ -category A ∞ △ C a t R u n , ∞ \mathbf{A}_{\infty}^{\triangle}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ △ Cat R un , ∞ of pretriangulated unital A ∞ A_{\infty} A ∞ -categories over R R R in Definition 7.33 to the ∞ \infty ∞ -category C a t ∞ s t \mathcal{C}\mathrm{at}_{\infty}^{\mathrm{st}} C at ∞ st of stable ∞ \infty ∞ -categories in Definition 2.38.
Remark 9.14. (a) As in Remark 4.63, if ( X , ω ) (X,\omega) ( X , ω ) is a suitable symplectic manifold, the derived Fukaya category is D F b ( X , ω ) = H 0 ( Tw ( F ( X , ω ) ) ) D^{b}_{\mathcal{F}}(X,\omega)=H^{0}(\operatorname{Tw}(\mathcal{F}(X,\omega))) D F b ( X , ω ) = H 0 ( Tw ( F ( X , ω ))) , a triangulated category. Kontsevich’s Homological Mirror Symmetry Conjecture [53] explains Mirror Symmetry of Calabi–Yau manifolds X , X ˇ X,\check{X} X , X ˇ in terms of equivalences of triangulated categories D F b ( X , ω ) ≃ D c o h b X ˇ D^{b}_{\mathcal{F}}(X,\omega)\simeq D^{b}_{\mathrm{coh}}\check{X} D F b ( X , ω ) ≃ D coh b X ˇ .
As in §2.7, stable ∞ \infty ∞ -categories are the ∞ \infty ∞ -categorical version of triangulated categories, and are better-behaved than triangulated categories in some ways. Corollary 9.13 is important as it will allow construction of derived Fukaya categories as stable ∞ \infty ∞ -categories, in an ∞ \infty ∞ -functorial way.
(b) We can interpret N A ∞ ∞ \mathcal{N}_{\mathbf{A}_{\infty}}^{\infty} N A ∞ ∞ in Theorem 9.12 in terms of the ∞ \infty ∞ -bicategory ( A ∞ C a t ^ R u n , ∞ , T ) (\widehat{\mathbf{A}_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty},\mathcal{T}) ( A ∞ Cat R un , ∞ , T ) in Theorem 7.17, as mapping A ↦ Hom A ∞ C a t ^ R u n , ∞ ( P u n , A ) \mathcal{A}\mapsto\operatorname{Hom}_{\widehat{\mathbf{A}_{\infty}\mathrm{Cat}}_{R}^{\mathrm{un},\infty}}(\mathcal{P}^{\mathrm{un}},\mathcal{A}) A ↦ Hom A ∞ Cat R un , ∞ ( P un , A ) . Probably Theorem 9.12(a) could be proved using the theory of ∞ \infty ∞ -bicategories, but we give a direct proof.
(c) An alternative, less explicit way to construct the ∞ \infty ∞ -functor N A ∞ ∞ \mathcal{N}_{A_\infty}^{\infty} N A ∞ ∞ in Theorem 9.12 is via the homotopy commutative diagram:
\begin{CD}
\mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R) @>{\mathcal{N}(\mathsf{N}_{A_\infty})}>> \mathcal{N}(\mathsf{Cat}^{1}_\infty) \\
@V{\Pi_{\mathrm{loc}}^\infty}VV @VV{\Pi_{\mathrm{loc}}^\infty}V \\
\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_\infty@>{\mathcal{N}(\mathsf{N}_{A_\infty})_{\mathrm{loc}}^\infty}>> \mathsf{Cat}^{1}_\infty[\mathsf{Joy}^{-1}]_\infty\\
@V{\simeq\ (9.11)}VV @VV{\simeq\ (2.15)}V \\
\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un},\infty}_R @>{\mathcal{N}_{A_\infty}^{\infty}}>> \mathsf{Cat}_\infty
\end{CD}
Here we first show that N A ∞ \mathcal{N}_{A_\infty} N A ∞ takes homotopy equivalences to Joyal equivalences, then the middle horizontal morphism exists by properties of localizations, and the bottom horizontal morphism exists as the bottom vertical morphisms are equivalences. Properties of localizations then imply that N A ∞ ∞ \mathcal{N}_{A_\infty}^{\infty} N A ∞ ∞ is characterized uniquely up to homotopy by the fact that (9.15) commutes.
(d) It would be interesting to strengthen Theorem 9.12 to make N A ∞ ∞ \mathcal{N}_{A_\infty}^{\infty} N A ∞ ∞ map to R R R -linear ∞ \infty ∞ -categories, in a suitable sense. For the restriction (9.16) to A ∞ △ C a t R u n , ∞ \mathsf{A}_\infty^{\triangle}\mathsf{Cat}^{\mathrm{un},\infty}_R A ∞ △ Cat R un , ∞ , these could be tensored over the stable ∞ \infty ∞ -category D b m o d D^b\mathrm{mod} D b mod -R R R .
An ‘Inner Hom functor’ for A ∞ C a t R u n \mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R A ∞ Cat R un
Theorem 9.15. Let R R R be a commutative ring. Then there exist ∞ \infty ∞ -functors F u n ~ A ∞ u n , ∞ \widetilde{\mathrm{Fun}}_{A_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ , F u n A ∞ u n , ∞ \mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ , unique up to homotopy, such that the following diagrams of ∞ \infty ∞ -categories commute up to homotopy:
\begin{CD}
\mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R)^{\mathrm{op}} \times\mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R) @>{\mathcal{N}(\mathrm{Fun}_{A_\infty}^{\mathrm{un}})}>> \mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R) \\
@V{(\Pi_{\mathrm{loc}}^\infty)^{\mathrm{op}} \times\Pi_{\mathrm{loc}}^\infty}VV @VV{\Pi_{\mathrm{loc}}^\infty}V \\
(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_\infty)^{\mathrm{op}} \times\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_\infty@>{\widetilde{\mathrm{Fun}}_{A_\infty}^{\mathrm{un}}}>> \mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]_\infty
\end{CD}
N ( A ∞ C a t R u n ) o p × N ( A ∞ C a t R u n ) → N ( F u n A ∞ u n ) N ( A ∞ C a t R u n ) ( i n c R u n ) o p × i n c R u n ↓ ↓ i n c R u n ( A ∞ C a t R u n , ∞ ) o p × A ∞ C a t R u n , ∞ → F u n A ∞ u n , ∞ A ∞ C a t R u n , ∞ \begin{CD}
\mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R)^{\mathrm{op}} \times\mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R) @>{\mathcal{N}(\mathrm{Fun}_{A_\infty}^{\mathrm{un}})}>> \mathcal{N}(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R) \\
@V{(\mathrm{inc}^{\mathrm{un}}_R)^{\mathrm{op}} \times\mathrm{inc}^{\mathrm{un}}_R}VV @VV{\mathrm{inc}^{\mathrm{un}}_R}V \\
(\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un},\infty}_R)^{\mathrm{op}} \times\mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un},\infty}_R @>{\mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty}}>> \mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un},\infty}_R
\end{CD} N ( A ∞ Cat R un ) op × N ( A ∞ Cat R un ) ( inc R un ) op × inc R un ↓ ⏐ ( A ∞ Cat R un , ∞ ) op × A ∞ Cat R un , ∞ N ( Fun A ∞ un ) Fun A ∞ un , ∞ N ( A ∞ Cat R un ) ↓ ⏐ inc R un A ∞ Cat R un , ∞ Here the ordinary functor F u n A ∞ u n \mathrm{Fun}_{A_\infty}^{\mathrm{un}} Fun A ∞ un is as in (4.24), and i n c R u n \mathrm{inc}^{\mathrm{un}}_R inc R un is as in Definition 7.30.
Proof. Corollary 4.31 implies that F u n A ∞ u n \mathrm{Fun}_{A_\infty}^{\mathrm{un}} Fun A ∞ un maps ( H E u n ) o p × H E u n → H E u n (\mathcal{H}\mathcal{E}_{\mathrm{un}})^{\mathrm{op}} \times\mathcal{H}\mathcal{E}_{\mathrm{un}} \to\mathcal{H}\mathcal{E}_{\mathrm{un}} ( H E un ) op × H E un → H E un . Thus by the universal property of localizations of ∞ \infty ∞ -categories, there exists an ∞ \infty ∞ -functor F u n ~ A ∞ u n \widetilde{\mathrm{Fun}}_{A_\infty}^{\mathrm{un}} Fun A ∞ un , unique up to homotopy, making (9.17) commute up to homotopy. Equation (9.18) then follows from (9.17) by the equivalence (9.11), which identifies Π l o c ∞ \Pi_{\mathrm{loc}}^\infty Π loc ∞ with i n c R u n \mathrm{inc}^{\mathrm{un}}_R inc R un up to homotopy.
Remark 9.16. (a) The ∞ \infty ∞ -functor F u n A ∞ u n , ∞ \mathrm{Fun}_{A_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ is an analogue of the internal mapping objects H o m ( C , D ) \mathop{\mathrm{Hom}}(\mathcal{C},\mathcal{D}) Hom ( C , D ) in D G C a t R [ Q E d g − 1 ] ∞ \mathsf{DGCat}_R[\mathcal{Q}\mathcal{E}_{\mathrm{dg}}^{-1}]_\infty DGCat R [ Q E dg − 1 ] ∞ , A ∞ C a t R u n [ Q E u n − 1 ] ∞ \mathsf{A}_\infty\mathsf{Cat}^{\mathrm{un}}_R[\mathcal{Q}\mathcal{E}_{\mathrm{un}}^{-1}]_\infty A ∞ Cat R un [ Q E un − 1 ] ∞ discussed in Theorems 4.38, 4.66 and 4.67, and should coincide with them on restriction to the ∞ \infty ∞ -subcategory A ∞ C a t R u n , h s , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un,hs},\infty} A ∞ Cat R un , hs , ∞ which is equivalent to D G C a t R [ Q E d g − 1 ] ∞ \mathrm{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]^{\infty} DGCat R [ QE dg − 1 ] ∞ , A ∞ C a t R u n [ Q E u n − 1 ] ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}[\mathcal{QE}_{\mathrm{un}}^{-1}]^{\infty} A ∞ Cat R un [ QE un − 1 ] ∞ by Corollary 9.8(a).
However, our ∞ \infty ∞ -functors are both more general – they are defined on larger ∞ \infty ∞ -categories – and are easier to construct. One reason is that we are inverting homotopy equivalences, not quasi-equivalences, so we can apply Corollary 4.31.
(b) It is an obvious question whether we can define a symmetric monoidal structure ⊗ ∞ \otimes^{\infty} ⊗ ∞ on A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ for which F u n A ∞ u n , ∞ \mathrm{Fun}_{\mathrm{A}_{\infty}}^{\mathrm{un},\infty} Fun A ∞ un , ∞ is the associated internal mapping object, as for D G C a t R [ Q E d g − 1 ] ∞ \mathrm{DGCat}_{R}[\mathcal{QE}_{\mathrm{dg}}^{-1}]^{\infty} DGCat R [ QE dg − 1 ] ∞ in Theorem 4.38. This is possible for the ∞ \infty ∞ -subcategory A ∞ C a t R u n , h s , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un,hs},\infty} A ∞ Cat R un , hs , ∞ by Theorem 4.38 and Corollary 9.8(a). But the author does not know whether ⊗ ∞ \otimes^{\infty} ⊗ ∞ can be extended to A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ . See Saneblidze–Umble [80, 81], Markl–Shnider [67] and Ornaghi [74] for results related to this question, on defining tensor products of A ∞ A_{\infty} A ∞ -algebras and A ∞ A_{\infty} A ∞ -categories.
Another interesting question is whether we can find an explicit formula for the ∞ \infty ∞ -functor F u n A ∞ u n , ∞ \mathrm{Fun}_{\mathrm{A}_{\infty}}^{\mathrm{un},\infty} Fun A ∞ un , ∞ in (9.18), written in terms of the presentation of A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ in §7.2 as an explicit simplicial set. The author has not managed to find such a general formula.
However, we can write down explicit formulae for F u n A ∞ u n , ∞ \mathrm{Fun}_{\mathrm{A}_{\infty}}^{\mathrm{un},\infty} Fun A ∞ un , ∞ on the ∞ \infty ∞ -subcategories ( N ( A ∞ C a t R u n ) ) o p × A ∞ C a t R u n , ∞ (\mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}))^{\mathrm{op}}\times\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} ( N ( A ∞ Cat R un ) ) op × A ∞ Cat R un , ∞ and ( A ∞ C a t R u n , ∞ ) o p × N ( A ∞ C a t R u n ) (\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) ( A ∞ Cat R un , ∞ ) op × N ( A ∞ Cat R un ) .
Definition 9.17. (a) Let R R R be a commutative ring. We will define an ∞ \infty ∞ -functor
F u n ^ A ∞ u n , ∞ : N ( A ∞ C a t R u n ) o p × A ∞ C a t R u n , ∞ ⟶ A ∞ C a t R u n , ∞ . (258) \widehat{\mathrm{Fun}}_{\mathrm{A}_{\infty}}^{\mathrm{un},\infty}\colon\mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}\longrightarrow\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}. \tag*{(258)} Fun A ∞ un , ∞ : N ( A ∞ Cat R un ) op × A ∞ Cat R un , ∞ ⟶ A ∞ Cat R un , ∞ . (258)
Form the opposite ∞ \infty ∞ -category N ( A ∞ C a t R u n ) o p \mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}} N ( A ∞ Cat R un ) op as in Definitions 2.8 and 2.13. Then N N N -simplices of the product ∞ \infty ∞ -category N ( A ∞ C a t R u n ) o p × A ∞ C a t R u n , ∞ \mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} N ( A ∞ Cat R un ) op × A ∞ Cat R un , ∞ are pairs
( S , T ) : = ( ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) , ( ( P i ) i = 0 N , ( g i j ) 0 ≤ i ≤ j ≤ N , ( I i ) i = 0 N , ( n I ) ∅ ≠ I ⊆ { 0 , … , N } ) ) . \begin{aligned}
(S,T):={}&\bigl(((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(\mathfrak{m}^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}}),\\
&\qquad((\mathcal{P}^{i})_{i=0}^{N},(g^{ij})_{0\leq i\leq j\leq N},(\mathcal{I}^{i})_{i=0}^{N},(\mathfrak{n}^{I})_{\varnothing\ne I\subseteq\{0,\ldots,N\}})\bigr).
\end{aligned} ( S , T ) := ( (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) , (( P i ) i = 0 N , ( g ij ) 0 ≤ i ≤ j ≤ N , ( I i ) i = 0 N , ( n I ) ∅ = I ⊆ { 0 , … , N } ) ) .
where S S S is an N N N -simplex in N ( A ∞ C a t R u n ) ⊂ A ∞ C a t R u n , ∞ \mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}})\subset\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} N ( A ∞ Cat R un ) ⊂ A ∞ Cat R un , ∞ and T T T is an N N N -simplex in A ∞ C a t R u n , ∞ \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ , as in §7.2. (Here we regard i n c R u n \mathrm{inc}_{R}^{\mathrm{un}} inc R un in Definition 7.30 as embedding N ( A ∞ C a t R u n ) \mathcal{N}(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}}) N ( A ∞ Cat R un ) as a simplicial subset of A ∞ C a t R u n \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un}} A ∞ Cat R un .) A morphism α : [ N ‾ ] → [ N ] \alpha\colon[\overline{N}]\to[N] α : [ N ] → [ N ] in Δ \Delta Δ acts on ( S , T ) (S,T) ( S , T ) as
α : ( S , T ) ⟼ ( A ∞ C a t R u n , ∞ ( α o p ) ( S ) , A ∞ C a t R u n , ∞ ( α ) ( T ) ) , \alpha\colon(S,T)\longmapsto\bigl(\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}(\alpha^{\mathrm{op}})(S),\mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}(\alpha)(T)\bigr), α : ( S , T ) ⟼ ( A ∞ Cat R un , ∞ ( α op ) ( S ) , A ∞ Cat R un , ∞ ( α ) ( T ) ) ,
where A ∞ C a t R u n , ∞ ( α ) \mathrm{A}_{\infty}\mathrm{Cat}_{R}^{\mathrm{un},\infty}(\alpha) A ∞ Cat R un , ∞ ( α ) is as in Definition 7.7(b), and α o p ( i ) = N − α ( N ‾ − i ) \alpha^{\mathrm{op}}(i)=N-\alpha(\overline{N}-i) α op ( i ) = N − α ( N − i ) .
For i = 0 , … , N i=0,\ldots,N i = 0 , … , N , write A i , B i \mathcal{A}^{i},\mathcal{B}^{i} A i , B i for the A ∞ A_{\infty} A ∞ -categories over R R R at vertex i i i of S , T S,T S , T , so that Obj A i = O i \operatorname{Obj}\mathcal{A}^{i}=\mathcal{O}^{i} Obj A i = O i , Obj B i = P i \operatorname{Obj}\mathcal{B}^{i}=\mathcal{P}^{i} Obj B i = P i , Hom A i ( x 0 , x 1 ) = H i ( x 0 , x 1 ) \operatorname{Hom}_{\mathcal{A}^{i}}(x_{0},x_{1})=\mathcal{H}^{i}(x_{0},x_{1}) Hom A i ( x 0 , x 1 ) = H i ( x 0 , x 1 ) , Hom B i ( y 0 , y 1 ) = I i ( y 0 , y 1 ) \operatorname{Hom}_{\mathcal{B}^{i}}(y_{0},y_{1})=\mathcal{I}^{i}(y_{0},y_{1}) Hom B i ( y 0 , y 1 ) = I i ( y 0 , y 1 ) , m k A i ; x 0 , … , x k = m k { i } ; x 0 , … , x k m_{k}^{\mathcal{A}^{i};x_{0},\ldots,x_{k}}=m_{k}^{\{i\};x_{0},\ldots,x_{k}} m k A i ; x 0 , … , x k = m k { i } ; x 0 , … , x k , and m k B i ; y 0 , … , y k = n k { i } ; y 0 , … , y k m_{k}^{\mathcal{B}^{i};y_{0},\ldots,y_{k}}=n_{k}^{\{i\};y_{0},\ldots,y_{k}} m k B i ; y 0 , … , y k = n k { i } ; y 0 , … , y k . For 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , write F i j : A i → A j \mathcal{F}^{ij}\colon\mathcal{A}^{i}\to\mathcal{A}^{j} F ij : A i → A j and G i j : B i → B j \mathcal{G}^{ij}\colon\mathcal{B}^{i}\to\mathcal{B}^{j} G ij : B i → B j for the A ∞ A_{\infty} A ∞ -functors at edges [ i , j ] [i,j] [ i , j ] of S , T S,T S , T , so that F O b j i j = f i j \mathcal{F}^{ij}_{\mathrm{Obj}}=f^{ij} F Obj ij = f ij , G O b j i j = g i j \mathcal{G}^{ij}_{\mathrm{Obj}}=g^{ij} G Obj ij = g ij , F k ; x 0 , … , x k i j = m k { i , j } ; x 0 , … , x k \mathcal{F}^{ij}_{k;x_{0},\ldots,x_{k}}=m^{\{i,j\};x_{0},\ldots,x_{k}}_{k} F k ; x 0 , … , x k ij = m k { i , j } ; x 0 , … , x k , and G k ; y 0 , … , y k i j = n k { i , j } ; y 0 , … , y k \mathcal{G}^{ij}_{k;y_{0},\ldots,y_{k}}=n^{\{i,j\};y_{0},\ldots,y_{k}}_{k} G k ; y 0 , … , y k ij = n k { i , j } ; y 0 , … , y k .
As S S S is an N N N -simplex in N ( A ∞ C a t R u n ) ⊂ A ∞ C a t R u n \mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R})\subset A_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R} N ( A ∞ Cat R un ) ⊂ A ∞ Cat R un , we have F i k = F j k ∘ F i j \mathcal{F}^{ik}=\mathcal{F}^{jk}\circ\mathcal{F}^{ij} F ik = F j k ∘ F ij for 0 ≤ i < j < k ≤ N 0\leq i<j<k\leq N 0 ≤ i < j < k ≤ N , and m 0 { i , j , k } ; x = 1 F i k ( x ) k m^{\{i,j,k\};x}_{0}=1^{k}_{\mathcal{F}^{ik}(x)} m 0 { i , j , k } ; x = 1 F ik ( x ) k , and m k I ; x 0 , … , x k = 0 m^{I;x_{0},\ldots,x_{k}}_{k}=0 m k I ; x 0 , … , x k = 0 if ∣ I ∣ ≥ 4 |I|\geq4 ∣ I ∣ ≥ 4 or ∣ I ∣ = 3 |I|=3 ∣ I ∣ = 3 and k ≥ 1 k\geq1 k ≥ 1 .
To deal with N N N -simplices S S S in the opposite ∞ \infty ∞ -categories ( N ( A ∞ C a t R u n ) ) o p (\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}))^{\mathrm{op}} ( N ( A ∞ Cat R un ) ) op or ( A ∞ C a t R u n ) o p (A_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R})^{\mathrm{op}} ( A ∞ Cat R un ) op we will regard S S S as an N N N -simplex in N ( A ∞ C a t R u n ) \mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R}) N ( A ∞ Cat R un ) or A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un}}_{R} A ∞ Cat R un , but we will consistently reverse the order of the vertices 0 , 1 , … , N 0,1,\ldots,N 0 , 1 , … , N in Δ N \Delta^{N} Δ N , replacing vertex i i i by vertex N − i N-i N − i . So our formulae will involve O N − i \mathcal{O}^{N-i} O N − i , f ( N − i ) ( N − j ) f^{(N-i)(N-j)} f ( N − i ) ( N − j ) , A N − i \mathcal{A}^{N-i} A N − i , F ( N − i ) ( N − j ) \mathcal{F}^{(N-i)(N-j)} F ( N − i ) ( N − j ) , and so on. We write d k , l ; x 0 , … , x l A , I ; w 0 , … , w k d^{\mathcal{A},I;w_{0},\ldots,w_{k}}_{k,l;x_{0},\ldots,x_{l}} d k , l ; x 0 , … , x l A , I ; w 0 , … , w k for the operators d k , l ; x 0 , … , x l I ; w 0 , … , w k d^{I;w_{0},\ldots,w_{k}}_{k,l;x_{0},\ldots,x_{l}} d k , l ; x 0 , … , x l I ; w 0 , … , w k in Definition 7.2 defined using the data in the N N N -simplex S S S .
Given such an N N N -simplex ( S , T ) (S,T) ( S , T ) for N ≥ 0 N\geq0 N ≥ 0 we will define another N N N -simplex
F u n ^ A ∞ u n , ∞ ( S , T ) : = ( ( Q i ) i = 0 N , ( h i j ) 0 ≤ i ≤ j ≤ N , ( J i ) i = 0 N , ( o I ) ∅ ≠ I ⊆ { 0 , … , N } ) (259) \widehat{\mathrm{Fun}}_{A_{\infty}^{\mathrm{un},\infty}}(S,T):=\bigl((\mathcal{Q}^{i})^{N}_{i=0},(h^{ij})_{0\leq i\leq j\leq N},(\mathcal{J}^{i})^{N}_{i=0},(o^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\bigr) \tag*{(259)} Fun A ∞ un , ∞ ( S , T ) := ( ( Q i ) i = 0 N , ( h ij ) 0 ≤ i ≤ j ≤ N , ( J i ) i = 0 N , ( o I ) ∅ = I ⊆ { 0 , … , N } ) (259)
in A ∞ C a t R u n , ∞ \mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ . Theorem 9.18 shows that these define an ∞ \infty ∞ -functor (9.19). Define the data in (9.20) as follows:
For i = 0 , … , N i=0,\ldots,N i = 0 , … , N , Q i \mathcal{Q}^{i} Q i is the set of unital A ∞ A_{\infty} A ∞ -functors F i : A N − i ⟶ B i \mathcal{F}^{i}:\mathcal{A}^{N-i}\longrightarrow\mathcal{B}^{i} F i : A N − i ⟶ B i .
For 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , h i j : Q i ⟶ Q j h^{ij}:\mathcal{Q}^{i}\longrightarrow\mathcal{Q}^{j} h ij : Q i ⟶ Q j maps h i j : F i ⟼ G j ∘ F i ∘ F ( N − j ) ( N − i ) h^{ij}:\mathcal{F}^{i}\longmapsto\mathcal{G}^{j}\circ\mathcal{F}^{i}\circ\mathcal{F}^{(N-j)(N-i)} h ij : F i ⟼ G j ∘ F i ∘ F ( N − j ) ( N − i ) .
Suppose ∅ ≠ I ⊆ { 0 , 1 , … , N } \varnothing\neq I\subseteq\{0,1,\ldots,N\} ∅ = I ⊆ { 0 , 1 , … , N } with I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } for i 0 < i 1 < ⋯ < i M i_{0}<i_{1}<\cdots<i_{M} i 0 < i 1 < ⋯ < i M . Define h I : Q i 0 ⟶ Q i M h^{I}:\mathcal{Q}^{i_{0}}\longrightarrow\mathcal{Q}^{i_{M}} h I : Q i 0 ⟶ Q i M using the h i j h^{ij} h ij as in (7.2).
For i = 0 , … , N i=0,\ldots,N i = 0 , … , N and F , F ′ \mathcal{F},\mathcal{F}' F , F ′ in Q i \mathcal{Q}^{i} Q i , define J i ( F , F ′ ) = Hom A ∞ u n ( F , F ′ ) \mathcal{J}^{i}(\mathcal{F},\mathcal{F}')=\operatorname{Hom}_{A_{\infty}^{\mathrm{un}}}(\mathcal{F},\mathcal{F}') J i ( F , F ′ ) = Hom A ∞ un ( F , F ′ ) as a graded R R R -module, where Hom A ∞ u n ( − , − ) \operatorname{Hom}_{A_{\infty}^{\mathrm{un}}}(-,-) Hom A ∞ un ( − , − ) is as in Definition 4.11.
Suppose I = { i 0 , i 1 , … , i M } I=\{i_{0},i_{1},\ldots,i_{M}\} I = { i 0 , i 1 , … , i M } with 0 ≤ i 0 < i 1 < ⋯ < i M ≤ N 0\leq i_{0}<i_{1}<\cdots<i_{M}\leq N 0 ≤ i 0 < i 1 < ⋯ < i M ≤ N , and p ≥ 0 p\geq0 p ≥ 0 , and F 0 , … , F p ∈ Q i 0 \mathcal{F}^{0},\ldots,\mathcal{F}^{p}\in\mathcal{Q}^{i_{0}} F 0 , … , F p ∈ Q i 0 . We will define an R R R -linear map
o p I ; F 0 , … , F p : J i 0 ( F 0 , F 1 ) ⊗ R J i 0 ( F 1 , F 2 ) ⊗ R ⋯ ⊗ R J i 0 ( F p − 1 , F p ) ⟶ J i M ( h I ( F 0 ) , h i 0 i M ( F p ) ) (260) o^{I;\mathcal{F}^{0},\ldots,\mathcal{F}^{p}}_{p}:\mathcal{J}^{i_{0}}(\mathcal{F}^{0},\mathcal{F}^{1})\otimes_{R}\mathcal{J}^{i_{0}}(\mathcal{F}^{1},\mathcal{F}^{2})\otimes_{R}\cdots\otimes_{R}\mathcal{J}^{i_{0}}(\mathcal{F}^{p-1},\mathcal{F}^{p})
\longrightarrow\mathcal{J}^{i_{M}}(h^{I}(\mathcal{F}^{0}),h^{i_{0}i_{M}}(\mathcal{F}^{p})) \tag*{(260)} o p I ; F 0 , … , F p : J i 0 ( F 0 , F 1 ) ⊗ R J i 0 ( F 1 , F 2 ) ⊗ R ⋯ ⊗ R J i 0 ( F p − 1 , F p ) ⟶ J i M ( h I ( F 0 ) , h i 0 i M ( F p )) (260)
graded of degree 2 − M − p = 3 − ∣ I ∣ − p 2-M-p=3-|I|-p 2 − M − p = 3 − ∣ I ∣ − p . If T j ∈ J i 0 , d j ( F j − 1 , F j ) \mathcal{T}^{j}\in\mathcal{J}^{i_{0},d_{j}}(\mathcal{F}^{j-1},\mathcal{F}^{j}) T j ∈ J i 0 , d j ( F j − 1 , F j ) for j = 1 , … , p j=1,\ldots,p j = 1 , … , p then we will construct
o p I ; F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) = U o^{I;\mathcal{F}^{0},\ldots,\mathcal{F}^{p}}_{p}(\mathcal{T}^{1}\otimes\cdots\otimes\mathcal{T}^{p})=\mathcal{U} o p I ; F 0 , … , F p ( T 1 ⊗ ⋯ ⊗ T p ) = U
in J i M , d 1 + ⋯ + d p + 2 − M − p ( h I ( F 0 ) , h i 0 i M ( F p ) ) \mathcal{J}^{i_{M},d_{1}+\cdots+d_{p}+2-M-p}(h^{I}(\mathcal{F}^{0}),h^{i_{0}i_{M}}(\mathcal{F}^{p})) J i M , d 1 + ⋯ + d p + 2 − M − p ( h I ( F 0 ) , h i 0 i M ( F p )) .
To define U \mathcal{U} U in (9.22), for all k ≥ 0 k\geq0 k ≥ 0 and w 0 , … , w k ∈ O N − i M w_{0},\ldots,w_{k}\in\mathcal{O}^{N-i_{M}} w 0 , … , w k ∈ O N − i M , we define
U k w 0 , … , w k : Hom A N − i M ( w 0 , w 1 ) ⊗ R Hom A N − i M ( w 1 , w 2 ) ⊗ R ⋯ \mathcal{U}^{w_{0},\ldots,w_{k}}_{k}:\operatorname{Hom}_{\mathcal{A}^{N-i_{M}}}(w_{0},w_{1})\otimes_{R}\operatorname{Hom}_{\mathcal{A}^{N-i_{M}}}(w_{1},w_{2})\otimes_{R}\cdots U k w 0 , … , w k : Hom A N − i M ( w 0 , w 1 ) ⊗ R Hom A N − i M ( w 1 , w 2 ) ⊗ R ⋯
⊗ R Hom A N − i M ( w k − 1 , w k ) ⟶ Hom B i M ( h I ( F 0 ) ( w 0 ) , h i 0 i M ( F p ) ( w k ) ) \otimes_{R}\operatorname{Hom}_{\mathcal{A}^{N-i_{M}}}(w_{k-1},w_{k})\longrightarrow\operatorname{Hom}_{\mathcal{B}^{i_{M}}}(h^{I}(\mathcal{F}^{0})(w_{0}),h^{i_{0}i_{M}}(\mathcal{F}^{p})(w_{k})) ⊗ R Hom A N − i M ( w k − 1 , w k ) ⟶ Hom B i M ( h I ( F 0 ) ( w 0 ) , h i 0 i M ( F p ) ( w k ))
by, if M = 0 M=0 M = 0 ,
L k w 0 , … , w k = ∑ m : p ≤ m ≤ k + p , y 0 , … , y m ∈ P i 0 n m { i 0 } ; y 0 , … , y m ∘ D k ; m ; y 0 , … , y m p ; w 0 , … , w k ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) + ( − 1 ) d 1 + ⋯ + d p ∑ l : 0 ≤ l ≤ k , x 0 , … , x l ∈ O N − i 0 D l , 1 ; S 0 ( x 0 ) , S p ( x l ) p ; x 0 , … , x l ∘ d k , l ; x 0 , … , x l A , { N − i 0 } ; w 0 , … , w k ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) . (261) \begin{aligned}
\mathcal{L}^{w_0,\ldots,w_k}_k =
\sum_{\substack{m:\,p\leq m\leq k+p,\\ y_0,\ldots,y_m\in\mathcal{P}^{i_0}}}
n^{\{i_0\};y_0,\ldots,y_m}_m \circ D^{p;w_0,\ldots,w_k}_{k;m;y_0,\ldots,y_m}(\mathcal{S}^0,\ldots,\mathcal{S}^p,\mathcal{T}^1\otimes\cdots\otimes\mathcal{T}^p) \\
&+(-1)^{d_1+\cdots+d_p}
\sum_{\substack{l:\,0\leq l\leq k,\\ x_0,\ldots,x_l\in\mathcal{O}^{N-i_0}}}
D^{p;x_0,\ldots,x_l}_{l,1;\mathcal{S}^0(x_0),\mathcal{S}^p(x_l)}
\circ d^{\mathcal{A},\{N-i_0\};w_0,\ldots,w_k}_{k,l;x_0,\ldots,x_l}
(\mathcal{S}^0,\ldots,\mathcal{S}^p,\mathcal{T}^1\otimes\cdots\otimes\mathcal{T}^p). \tag*{(261)}
\end{aligned} L k w 0 , … , w k = m : p ≤ m ≤ k + p , y 0 , … , y m ∈ P i 0 ∑ n m { i 0 } ; y 0 , … , y m ∘ D k ; m ; y 0 , … , y m p ; w 0 , … , w k ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) + ( − 1 ) d 1 + ⋯ + d p l : 0 ≤ l ≤ k , x 0 , … , x l ∈ O N − i 0 ∑ D l , 1 ; S 0 ( x 0 ) , S p ( x l ) p ; x 0 , … , x l ∘ d k , l ; x 0 , … , x l A , { N − i 0 } ; w 0 , … , w k ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) . (261)
and if M > 0 M>0 M > 0
L k w 0 , … , w k = ∑ l , m : 0 ≤ l ≤ k , p ≤ m ≤ l + p , x 0 , … , x l ∈ O N − i 0 , y 0 , … , y m ∈ P i 0 n m I ; y 0 , … , y m ∘ D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) ∘ d k , l ; x 0 , … , x l A , { N − i M , N − i 0 } ; w 0 , … , w k . (262) \mathcal{L}^{w_0,\ldots,w_k}_k =
\sum_{\substack{l,m:\,0\leq l\leq k,\ p\leq m\leq l+p,\\ x_0,\ldots,x_l\in\mathcal{O}^{N-i_0},\\ y_0,\ldots,y_m\in\mathcal{P}^{i_0}}}
n^{I;y_0,\ldots,y_m}_m\circ D^{p;x_0,\ldots,x_l}_{l,m;y_0,\ldots,y_m}
(\mathcal{S}^0,\ldots,\mathcal{S}^p,\mathcal{T}^1\otimes\cdots\otimes\mathcal{T}^p)
\circ d^{\mathcal{A},\{N-i_M,N-i_0\};w_0,\ldots,w_k}_{k,l;x_0,\ldots,x_l}. \tag*{(262)} L k w 0 , … , w k = l , m : 0 ≤ l ≤ k , p ≤ m ≤ l + p , x 0 , … , x l ∈ O N − i 0 , y 0 , … , y m ∈ P i 0 ∑ n m I ; y 0 , … , y m ∘ D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) ∘ d k , l ; x 0 , … , x l A , { N − i M , N − i 0 } ; w 0 , … , w k . (262)
Here d k , l ; x 0 , … , x l A , { N − i M , N − i 0 } ; w 0 , … , w k d^{\mathcal{A},\{N-i_M,N-i_0\};w_0,\ldots,w_k}_{k,l;x_0,\ldots,x_l} d k , l ; x 0 , … , x l A , { N − i M , N − i 0 } ; w 0 , … , w k is d k , l ; x 0 , … , x l I ; w 0 , … , w k d^{I;w_0,\ldots,w_k}_{k,l;x_0,\ldots,x_l} d k , l ; x 0 , … , x l I ; w 0 , … , w k from §7.1 for the data S = ( ( O i ) , ( f i j ) , ( M i ) ) S=((\mathcal{O}^i),(f^{ij}),(\mathcal{M}^i)) S = (( O i ) , ( f ij ) , ( M i )) and I = { N − i M , N − i 0 } I=\{N-i_M,N-i_0\} I = { N − i M , N − i 0 } , and if l , m , p ≥ 0 l,m,p\geq0 l , m , p ≥ 0 , and x 0 , … , x l ∈ O N − i 0 x_0,\ldots,x_l\in\mathcal{O}^{N-i_0} x 0 , … , x l ∈ O N − i 0 , and y 0 , … , y m ∈ P i 0 y_0,\ldots,y_m\in\mathcal{P}^{i_0} y 0 , … , y m ∈ P i 0 , and S 0 , … , S p ∈ Q i 0 \mathcal{S}^0,\ldots,\mathcal{S}^p\in\mathcal{Q}^{i_0} S 0 , … , S p ∈ Q i 0 , we define
D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) = ∑ l = b 1 + ⋯ + b m , 0 = c 0 < c 1 < ⋯ < c p + 1 = m + 1 : b c i ≥ 0 , 1 ≤ i ≤ p , b j ≥ 1 , j ≠ c i ( − 1 ) ∑ i = 1 p ( m + 1 ) ( d i + 1 ) ⋅ ( − 1 ) ∑ j = 1 p ∑ i = 1 c j − 1 ( b j + 1 ) ( d i + 1 ) ⋅ ( − 1 ) ∑ i = 1 m ( b i + 1 ) ( m − i ) S 0 ; x s 0 , … , x s 1 ⊗ S 0 ; x s 1 , … , x s 2 ⊗ ⋯ ⊗ S 0 ; x s c 1 − 2 , … , x s c 1 − 1 ⊗ T b c 1 − 1 1 ; x s c 1 − 1 , … , x s c 1 ⊗ T b c 1 + 1 1 ; x s c 1 , … , x s c 1 + 1 ⊗ ⋯ ⊗ T b c 2 − 1 1 ; x s c 2 − 2 , … , x s c 2 − 1 ⊗ T b c 2 2 ; x s c 2 − 1 , … , x s c 2 ⊗ ⋯ ⊗ T b c p p ; x s c p − 1 , … , x s c p ⊗ ⋯ ⊗ T b c p + 1 p ; x s c p , … , x s c p + 1 ⊗ ⋯ ⊗ S b m p ; x s m − 1 , … , x s m . (263) \begin{aligned}
D^{p;x_0,\ldots,x_l}_{l,m;y_0,\ldots,y_m}(\mathcal{S}^0,\ldots,\mathcal{S}^p,\mathcal{T}^1\otimes\cdots\otimes\mathcal{T}^p) \\
&=\sum_{\substack{l=b_1+\cdots+b_m,\\ 0=c_0<c_1<\cdots<c_{p+1}=m+1:\\ b_{c_i}\geq0,\ 1\leq i\leq p,\\ b_j\geq1,\ j\ne c_i}}
(-1)^{\sum_{i=1}^{p}(m+1)(d_i+1)}
\cdot(-1)^{\sum_{j=1}^{p}\sum_{i=1}^{c_j-1}(b_j+1)(d_i+1)}
\cdot(-1)^{\sum_{i=1}^{m}(b_i+1)(m-i)}
\mathcal{S}^{0;x_{s_0},\ldots,x_{s_1}}\otimes\mathcal{S}^{0;x_{s_1},\ldots,x_{s_2}}\otimes\cdots\otimes\mathcal{S}^{0;x_{s_{c_1-2}},\ldots,x_{s_{c_1-1}}}
\otimes\mathcal{T}^{1;x_{s_{c_1-1}},\ldots,x_{s_{c_1}}}_{b_{c_1-1}}\otimes\mathcal{T}^{1;x_{s_{c_1}},\ldots,x_{s_{c_1+1}}}_{b_{c_1+1}}\otimes\cdots\otimes\mathcal{T}^{1;x_{s_{c_2-2}},\ldots,x_{s_{c_2-1}}}_{b_{c_2-1}}
\otimes\mathcal{T}^{2;x_{s_{c_2-1}},\ldots,x_{s_{c_2}}}_{b_{c_2}}\otimes\cdots\otimes\mathcal{T}^{p;x_{s_{c_p-1}},\ldots,x_{s_{c_p}}}_{b_{c_p}}
\otimes\cdots\otimes\mathcal{T}^{p;x_{s_{c_p}},\ldots,x_{s_{c_p+1}}}_{b_{c_{p+1}}}\otimes\cdots\otimes\mathcal{S}^{p;x_{s_{m-1}},\ldots,x_{s_m}}_{b_m}. \tag*{(263)}
\end{aligned} D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) = l = b 1 + ⋯ + b m , 0 = c 0 < c 1 < ⋯ < c p + 1 = m + 1 : b c i ≥ 0 , 1 ≤ i ≤ p , b j ≥ 1 , j = c i ∑ ( − 1 ) ∑ i = 1 p ( m + 1 ) ( d i + 1 ) ⋅ ( − 1 ) ∑ j = 1 p ∑ i = 1 c j − 1 ( b j + 1 ) ( d i + 1 ) ⋅ ( − 1 ) ∑ i = 1 m ( b i + 1 ) ( m − i ) S 0 ; x s 0 , … , x s 1 ⊗ S 0 ; x s 1 , … , x s 2 ⊗ ⋯ ⊗ S 0 ; x s c 1 − 2 , … , x s c 1 − 1 ⊗ T b c 1 − 1 1 ; x s c 1 − 1 , … , x s c 1 ⊗ T b c 1 + 1 1 ; x s c 1 , … , x s c 1 + 1 ⊗ ⋯ ⊗ T b c 2 − 1 1 ; x s c 2 − 2 , … , x s c 2 − 1 ⊗ T b c 2 2 ; x s c 2 − 1 , … , x s c 2 ⊗ ⋯ ⊗ T b c p p ; x s c p − 1 , … , x s c p ⊗ ⋯ ⊗ T b c p + 1 p ; x s c p , … , x s c p + 1 ⊗ ⋯ ⊗ S b m p ; x s m − 1 , … , x s m . (263)
Set s j = ∑ i = 1 j b i s_j=\sum_{i=1}^{j}b_i s j = ∑ i = 1 j b i . Require y j = S i ( x s j ) y_j=\mathcal{S}^{i}(x_{s_j}) y j = S i ( x s j ) , 0 ≤ j ≤ m 0\leq j\leq m 0 ≤ j ≤ m , c i ≤ j < c i + 1 c_i\leq j<c_{i+1} c i ≤ j < c i + 1 .
This defines F u n ^ A ∞ u n , ∞ \widehat{\mathrm{Fun}}_{\mathcal{A}_\infty}^{\mathrm{un},\infty} Fun A ∞ un , ∞ in (9.19). Note that when N = 0 N=0 N = 0 , the 0 0 0 -simplex F u n ^ A ∞ u n , ∞ ( S , T ) \widehat{\mathrm{Fun}}_{\mathcal{A}_\infty}^{\mathrm{un},\infty}(S,T) Fun A ∞ un , ∞ ( S , T ) is the unital A ∞ A_\infty A ∞ -category F u n A ∞ u n ( A 0 , B 0 ) \mathrm{Fun}_{\mathcal{A}_\infty}^{\mathrm{un}}(\mathcal{A}^0,\mathcal{B}^0) Fun A ∞ un ( A 0 , B 0 ) from Definition 4.11.
(b) Similarly we define an ∞ \infty ∞ -functor
F u n A ∞ u n , ∞ : ( A ∞ C a t R u n , ∞ ) o p × N ( A ∞ C a t R u n ) ⟶ A ∞ C a t R u n , ∞ . (264) \mathrm{Fun}_{\mathcal{A}_\infty}^{\mathrm{un},\infty}:
(A_\infty\mathrm{Cat}_{\mathcal{R}}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_\infty\mathrm{Cat}_{\mathcal{R}}^{\mathrm{un}})\longrightarrow A_\infty\mathrm{Cat}_{\mathcal{R}}^{\mathrm{un},\infty}. \tag*{(264)} Fun A ∞ un , ∞ : ( A ∞ Cat R un , ∞ ) op × N ( A ∞ Cat R un ) ⟶ A ∞ Cat R un , ∞ . (264)
Most of the definition is the same as in (a). The differences are that as now T T T is an N N N -simplex in N ( A ∞ C a t R u n ) ⊂ A ∞ C a t R u n \mathcal{N}(A_\infty\mathrm{Cat}_{\mathcal{R}}^{\mathrm{un}})\subset A_\infty\mathrm{Cat}_{\mathcal{R}}^{\mathrm{un}} N ( A ∞ Cat R un ) ⊂ A ∞ Cat R un , we have G i k = G j k ∘ G i j G^{ik}=G^{jk}\circ G^{ij} G ik = G j k ∘ G ij for 0 ≤ i ≤ j ≤ k ≤ N 0\leq i\leq j\leq k\leq N 0 ≤ i ≤ j ≤ k ≤ N , and n 0 i j k ; x = 1 G i k ( x ) n^{ijk;x}_0=1_{G^{ik}(x)} n 0 ij k ; x = 1 G ik ( x ) , and n k I ; x 0 , … , x k = 0 n^{I;x_0,\ldots,x_k}_k=0 n k I ; x 0 , … , x k = 0 if ∣ I ∣ ≥ 4 |I|\geq4 ∣ I ∣ ≥ 4 or ∣ I ∣ = 3 |I|=3 ∣ I ∣ = 3 and k ≥ 1 k\geq1 k ≥ 1 , and we leave (9.23) unchanged, but replace (9.24) by
L k w 0 , … , w k = ( − 1 ) ( M + 1 ) ( d 1 + ⋯ + d p ) ∑ l , m : 0 ≤ l ≤ k , p ≤ m ≤ l + p , x 0 , … , x l ∈ O N − i 0 , y 0 , … , y m ∈ P i 0 n m { i 0 , i M } ; y 0 , … , y m ∘ D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) ∘ d k , l ; x 0 , … , x l A , { N − i : i ∈ I } ; w 0 , … , w k . (265) \mathcal{L}^{w_0,\ldots,w_k}_k=(-1)^{(M+1)(d_1+\cdots+d_p)}
\sum_{\substack{l,m:\,0\leq l\leq k,\ p\leq m\leq l+p,\\ x_0,\ldots,x_l\in\mathcal{O}^{N-i_0},\\ y_0,\ldots,y_m\in\mathcal{P}^{i_0}}}
n^{\{i_0,i_M\};y_0,\ldots,y_m}_m\circ D^{p;x_0,\ldots,x_l}_{l,m;y_0,\ldots,y_m}
(\mathcal{S}^0,\ldots,\mathcal{S}^p,\mathcal{T}^1\otimes\cdots\otimes\mathcal{T}^p)
\circ d^{\mathcal{A},\{N-i:i\in I\};w_0,\ldots,w_k}_{k,l;x_0,\ldots,x_l}. \tag*{(265)} L k w 0 , … , w k = ( − 1 ) ( M + 1 ) ( d 1 + ⋯ + d p ) l , m : 0 ≤ l ≤ k , p ≤ m ≤ l + p , x 0 , … , x l ∈ O N − i 0 , y 0 , … , y m ∈ P i 0 ∑ n m { i 0 , i M } ; y 0 , … , y m ∘ D l , m ; y 0 , … , y m p ; x 0 , … , x l ( S 0 , … , S p , T 1 ⊗ ⋯ ⊗ T p ) ∘ d k , l ; x 0 , … , x l A , { N − i : i ∈ I } ; w 0 , … , w k . (265)
The next theorem will be proved in §11.6.
Theorem 9.18. The ∞ \infty ∞ -functors in (258) and (264) are well defined. Relating (9.18) with (258) , in the following diagram of ∞ \infty ∞ -categories, the top rectangle commutes strictly, and the bottom commutes up to homotopy:
N ( A ∞ C a t R u n ) o p × N ( A ∞ C a t R u n ) → N ( Fun ^ A ∞ u n , ∞ ) N ( A ∞ C a t R u n ) ↓ i d × i n c R u n ↓ i n c R u n N ( A ∞ C a t R u n ) o p × A ∞ C a t R u n , ∞ → Fun ^ A ∞ u n , ∞ A ∞ C a t R u n , ∞ ↓ ( i n c R u n ) o p × i d ↓ i d ( A ∞ C a t R u n , ∞ ) o p × A ∞ C a t R u n , ∞ → Fun A ∞ u n , ∞ A ∞ C a t R u n , ∞ . (266) \begin{aligned}
\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})&\xrightarrow{\ \mathcal{N}(\widehat{\operatorname{Fun}}_{\mathbf{A}_{\infty}}^{\mathrm{un},\infty})\ }\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\\
\downarrow{\scriptstyle\mathrm{id}\times\mathrm{inc}_{R}^{\mathrm{un}}}&&\downarrow{\scriptstyle\mathrm{inc}_{R}^{\mathrm{un}}}\\
\mathcal{N}(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}&\xrightarrow{\ \widehat{\operatorname{Fun}}_{\mathbf{A}_{\infty}}^{\mathrm{un},\infty}\ }\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\\
\downarrow{\scriptstyle(\mathrm{inc}_{R}^{\mathrm{un}})^{\mathrm{op}}\times\mathrm{id}}&&\downarrow{\scriptstyle\mathrm{id}}\\
(\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}&\xrightarrow{\ \operatorname{Fun}_{\mathbf{A}_{\infty}}^{\mathrm{un},\infty}\ }\mathbf{A}_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}. \tag*{(266)}
\end{aligned} N ( A ∞ Cat R un ) op × N ( A ∞ Cat R un ) ↓ id × inc R un N ( A ∞ Cat R un ) op × A ∞ Cat R un , ∞ ↓ ( inc R un ) op × id ( A ∞ Cat R un , ∞ ) op × A ∞ Cat R un , ∞ N ( Fun A ∞ un , ∞ ) N ( A ∞ Cat R un ) Fun A ∞ un , ∞ A ∞ Cat R un , ∞ Fun A ∞ un , ∞ A ∞ Cat R un , ∞ . ↓ inc R un ↓ id (266) The analogue holds relating (9.18) with (264) .
We did not treat filtered A ∞ A_{\infty} A ∞ -categories from Chapter 8 in the material above, because as in Remark 9.10 the author does not know how to write A ∞ f i C a t R u n , ∞ A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ fi Cat R un , ∞ , A ∞ w f C a t R u n , ∞ A_{\infty}^{\mathrm{wf}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ wf Cat R un , ∞ as localizations, so we have no filtered analogues of (9.17). However, we can generalize Definition 9.17 to the filtered case.
Definition 9.19. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring. We will define ∞ \infty ∞ -functors
Fun ^ A ∞ f i u n , ∞ : N ( A ∞ f i C a t R u n ) o p × A ∞ f i C a t R u n , ∞ ⟶ A ∞ f i C a t R , 2 , 2 u n , ∞ , (267) \widehat{\operatorname{Fun}}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un},\infty}:\mathcal{N}(A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\longrightarrow A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R,2,2}^{\mathrm{un},\infty}, \tag*{(267)} Fun A ∞ fi un , ∞ : N ( A ∞ fi Cat R un ) op × A ∞ fi Cat R un , ∞ ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ , (267) Fun ^ A ∞ w f u n , ∞ : N ( A ∞ f i C a t R u n ) o p × A ∞ w f C a t R u n , ∞ ⟶ A ∞ w f C a t R , 2 , 2 u n , ∞ , (268) \widehat{\operatorname{Fun}}_{A_{\infty}^{\mathrm{wf}}}^{\mathrm{un},\infty}:\mathcal{N}(A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})^{\mathrm{op}}\times A_{\infty}^{\mathrm{wf}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty}\longrightarrow A_{\infty}^{\mathrm{wf}}\mathop{\mathrm{Cat}}\nolimits_{R,2,2}^{\mathrm{un},\infty}, \tag*{(268)} Fun A ∞ wf un , ∞ : N ( A ∞ fi Cat R un ) op × A ∞ wf Cat R un , ∞ ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ , (268) Fun ~ A ∞ f i u n , ∞ : ( A ∞ f i C a t R u n , ∞ ) o p × N ( A ∞ f i C a t R u n ) ⟶ A ∞ f i C a t R , 2 , 2 u n , ∞ , (269) \widetilde{\operatorname{Fun}}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un},\infty}:(A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\longrightarrow A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R,2,2}^{\mathrm{un},\infty}, \tag*{(269)} Fun A ∞ fi un , ∞ : ( A ∞ fi Cat R un , ∞ ) op × N ( A ∞ fi Cat R un ) ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ , (269) Fun ~ A ∞ w f u n , ∞ : ( A ∞ w f C a t R u n , ∞ ) o p × N ( A ∞ f i C a t R u n ) ⟶ A ∞ w f C a t R , 2 , 2 u n , ∞ . (270) \widetilde{\operatorname{Fun}}_{A_{\infty}^{\mathrm{wf}}}^{\mathrm{un},\infty}:(A_{\infty}^{\mathrm{wf}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty})^{\mathrm{op}}\times\mathcal{N}(A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}})\longrightarrow A_{\infty}^{\mathrm{wf}}\mathop{\mathrm{Cat}}\nolimits_{R,2,2}^{\mathrm{un},\infty}. \tag*{(270)} Fun A ∞ wf un , ∞ : ( A ∞ wf Cat R un , ∞ ) op × N ( A ∞ fi Cat R un ) ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ . (270)
To define (267) and (269) we follow Definition 9.17 with the following differences: A i \mathcal{A}^{i} A i , B i \mathcal{B}^{i} B i , H i j \mathcal{H}^{ij} H ij , F i j \mathcal{F}^{ij} F ij , G i j \mathcal{G}^{ij} G ij are filtered unital A ∞ A_{\infty} A ∞ -categories and unital A ∞ A_{\infty} A ∞ -functors, and we replace A ∞ C a t R u n A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ Cat R un , A ∞ C a t R u n , ∞ A_{\infty}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ Cat R un , ∞ by A ∞ f i C a t R u n A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un}} A ∞ fi Cat R un , A ∞ f i C a t R u n , ∞ A_{\infty}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits_{R}^{\mathrm{un},\infty} A ∞ fi Cat R un , ∞ , and Hom A ∞ u n ( H , H ′ ) \operatorname{Hom}_{A_{\infty}}^{\mathrm{un}}(\mathcal{H},\mathcal{H}') Hom A ∞ un ( H , H ′ ) by Hom A ∞ f i u n ( H , H ′ ) \operatorname{Hom}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{H},\mathcal{H}') Hom A ∞ fi un ( H , H ′ ) , and we change the limits of sums in (261) –(263) and (265) :
(i) In (261) the first sum is over m ≥ p m\ge p m ≥ p and the second over l ≥ 0 l\ge0 l ≥ 0 .
(ii) In (262) and (265) we sum over l ≥ 0 l\ge0 l ≥ 0 and m ≥ p m\ge p m ≥ p .
(iii) In (263) we do not require b j ≥ 1 b_j\ge1 b j ≥ 1 , j ≠ c i j\ne c_i j = c i .
Thus these equations become infinite sums, which converge in the filtered sense.
Then for (267) the 0 0 0 -simplex at vertex i i i of the N N N -simplex we define is the unital filtered A ∞ A_{\infty} A ∞ -category Fun A ∞ f i u n ( A N − i , B i ) \operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{A}^{N-i},\mathcal{B}^{i}) Fun A ∞ fi un ( A N − i , B i ) , and the 1 1 1 -simplex on edge i → j i\to j i → j is the unital filtered A ∞ A_{\infty} A ∞ -functor ( F ∗ ( N − j ) ( N − i ) ∘ G i j ) : Fun A ∞ f i u n ( A N − i , B i ) → Fun A ∞ f i u n ( A N − j , B j ) (\mathcal{F}^{(N-j)(N-i)}_{*}\circ\mathcal{G}^{ij}):\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{A}^{N-i},\mathcal{B}^{i})\to\operatorname{Fun}_{A_{\infty}^{\mathrm{fi}}}^{\mathrm{un}}(\mathcal{A}^{N-j},\mathcal{B}^{j}) ( F ∗ ( N − j ) ( N − i ) ∘ G ij ) : Fun A ∞ fi un ( A N − i , B i ) → Fun A ∞ fi un ( A N − j , B j ) , as in §5.5. Note as in Definition 5.19 and Remark 5.20 that these are strict filtered A ∞ A_{\infty} A ∞ -categories and strict filtered A ∞ A_{\infty} A ∞ -functors, so that m 0 I ; x 0 = 0 m_{0}^{I;x_{0}}=0 m 0 I ; x 0 = 0 for ∣ I ∣ = 1 , 2 |I|=1,2 ∣ I ∣ = 1 , 2 . Hence the target ∞ \infty ∞ -category in (267) is A ∞ , R , 2 , 2 f i C a t u n , ∞ A_{\infty, R,2,2}^{\mathrm{fi}}\mathop{\mathrm{Cat}}\nolimits^{\mathrm{un},\infty} A ∞ , R , 2 , 2 fi Cat un , ∞ from Definition 8.11. Equations (268) –(270) are similar.
The proof of the next theorem is the same as that of the first part of Theorem 9.18 in §11.6, except that the limits in sums are slightly different, which leads to infinite sums convergent in the filtered sense.
Theorem 9.20. The ∞ \infty ∞ -functors in (9.29)–(9.32) are well defined.
Definition 9.21. Let ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) be a po-monoid and R R R a complete Γ \Gamma Γ -filtered ring. Let P u n \mathcal{P}^{\mathrm{un}} P un be the unital point A ∞ A_\infty A ∞ -category over R R R from Definition 4.17, made into a strict filtered A ∞ A_\infty A ∞ -category in the obvious way. Following Definition 5.27 and Proposition 5.29 and using (9.29)–(9.30), define ∞ \infty ∞ -functors
Str ∞ = Fun ^ A ∞ f i u n , ∞ ( P u n , − ) : A ∞ f i Cat R u n , ∞ ⟶ A ∞ f i Cat R , 2 , 2 u n , ∞ (271) \operatorname{Str}^{\infty}=\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{fi}}_\infty}^{\mathrm{un},\infty}(\mathcal{P}^{\mathrm{un}},-):\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\longrightarrow\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2} \tag*{(271)} Str ∞ = Fun A ∞ fi un , ∞ ( P un , − ) : A ∞ fi Cat R un , ∞ ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ (271)
Str w f ∞ = Fun ^ A ∞ w f u n , ∞ ( P u n , − ) : A ∞ w f Cat R u n , ∞ ⟶ A ∞ w f Cat R , 2 , 2 u n , ∞ . (272) \operatorname{Str}^{\infty}_{\mathrm{wf}}=\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{wf}}_\infty}^{\mathrm{un},\infty}(\mathcal{P}^{\mathrm{un}},-):\mathcal{A}^{\mathrm{wf}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\longrightarrow\mathcal{A}^{\mathrm{wf}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}. \tag*{(272)} Str wf ∞ = Fun A ∞ wf un , ∞ ( P un , − ) : A ∞ wf Cat R un , ∞ ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ . (272)
By Proposition 5.29, these map an object A \mathcal{A} A to the associated strict filtered A ∞ A_\infty A ∞ -category Str ( A ) \operatorname{Str}(\mathcal{A}) Str ( A ) from Definition 5.27, and are ∞ \infty ∞ -categorical analogues of the functor Str \operatorname{Str} Str in Proposition 5.29(b),(d).
Remark 9.22. (a) Note that we can compose Fun ^ A ∞ f i u n , ∞ , … , Fun ^ A ∞ w f u n , ∞ \widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{fi}}_\infty}^{\mathrm{un},\infty},\ldots,\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{wf}}_\infty}^{\mathrm{un},\infty} Fun A ∞ fi un , ∞ , … , Fun A ∞ wf un , ∞ in (9.29)–(9.32) with Π A ∞ f i A ∞ w f \Pi_{\mathcal{A}^{\mathrm{fi}}_\infty}^{\mathcal{A}^{\mathrm{wf}}_\infty} Π A ∞ fi A ∞ wf in (8.8)–(8.9) to map to A ∞ Cat R u n , ∞ \mathcal{A}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ . The same applies to Str ∞ \operatorname{Str}^{\infty} Str ∞ and Str w f ∞ \operatorname{Str}^{\infty}_{\mathrm{wf}} Str wf ∞ in (9.33)–(9.34).
(b) It seems likely that (9.29)–(9.30) and (9.31)–(9.32) extend (at least up to homotopy) to ∞ \infty ∞ -functors
Fun A ∞ f i u n , ∞ : ( A ∞ f i Cat R u n , ∞ ) o p × A ∞ f i Cat R u n , ∞ ⟶ A ∞ f i Cat R , 2 , 2 u n , ∞ , \operatorname{Fun}_{\mathcal{A}^{\mathrm{fi}}_\infty}^{\mathrm{un},\infty}:\left(\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\right)^{\mathrm{op}}\times\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\longrightarrow\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}, Fun A ∞ fi un , ∞ : ( A ∞ fi Cat R un , ∞ ) op × A ∞ fi Cat R un , ∞ ⟶ A ∞ fi Cat R , 2 , 2 un , ∞ ,
Fun A ∞ w f u n , ∞ : ( A ∞ w f Cat R u n , ∞ ) o p × A ∞ w f Cat R u n , ∞ ⟶ A ∞ w f Cat R , 2 , 2 u n , ∞ , \operatorname{Fun}_{\mathcal{A}^{\mathrm{wf}}_\infty}^{\mathrm{un},\infty}:\left(\mathcal{A}^{\mathrm{wf}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\right)^{\mathrm{op}}\times\mathcal{A}^{\mathrm{wf}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\longrightarrow\mathcal{A}^{\mathrm{wf}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}, Fun A ∞ wf un , ∞ : ( A ∞ wf Cat R un , ∞ ) op × A ∞ wf Cat R un , ∞ ⟶ A ∞ wf Cat R , 2 , 2 un , ∞ ,
but we will not prove this.
The twist ∞ \infty ∞ -functor
We now give an ∞ \infty ∞ -categorical version of the twist functor Tw \operatorname{Tw} Tw in §4.9 and §5.8, in the unital case, by generalizing Definition 5.30.
Definition 9.23. Let R R R be a commutative ring. We will define an explicit ∞ \infty ∞ -functor
Tw ∞ : A ∞ Cat R u n , ∞ ⟶ A ∞ Δ Cat R u n , ∞ , (273) \operatorname{Tw}^{\infty}:\mathcal{A}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\longrightarrow\mathcal{A}^{\Delta}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R, \tag*{(273)} Tw ∞ : A ∞ Cat R un , ∞ ⟶ A ∞ Δ Cat R un , ∞ , (273)
where A ∞ Cat R u n , ∞ \mathcal{A}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Cat R un , ∞ is as in §7.2 and A ∞ Δ Cat R u n , ∞ \mathcal{A}^{\Delta}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R A ∞ Δ Cat R un , ∞ as in Definition 7.33. Make R R R into a complete filtered ring over the po-monoid ( Z , ≤ ) (\mathbb{Z},\leq) ( Z , ≤ ) by setting F λ R = R F^\lambda R=R F λ R = R if λ ≤ 0 \lambda\leq0 λ ≤ 0 in Z \mathbb{Z} Z and F λ R = 0 F^\lambda R=0 F λ R = 0 if λ > 0 \lambda>0 λ > 0 in Z \mathbb{Z} Z . Generalizing (5.20), we will define Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ in (9.35) as the composition of ∞ \infty ∞ -functors
A ∞ Cat R u n , ∞ → Σ ∞ A ∞ f i Cat R u n , ∞ → Fun ^ A ∞ f i u n ( P u n , − ) A ∞ f i Cat R , 2 , 2 u n , ∞ → Π A ∞ ∞ A ∞ Δ Cat R u n , ∞ . (274) \mathcal{A}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\xrightarrow{\Sigma^\infty}\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R\xrightarrow{\widehat{\operatorname{Fun}}_{\mathcal{A}^{\mathrm{fi}}_\infty}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},-)}\mathcal{A}^{\mathrm{fi}}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_{R,2,2}\xrightarrow{\Pi_{\mathcal{A}^{\infty}_\infty}}\mathcal{A}^{\Delta}_\infty\operatorname{Cat}^{\mathrm{un},\infty}_R. \tag*{(274)} A ∞ Cat R un , ∞ Σ ∞ A ∞ fi Cat R un , ∞ Fun A ∞ fi un ( P un , − ) A ∞ fi Cat R , 2 , 2 un , ∞ Π A ∞ ∞ A ∞ Δ Cat R un , ∞ . (274)
Here Σ ∞ \Sigma^\infty Σ ∞ is an ∞ \infty ∞ -categorical generalization of Σ \Sigma Σ in Definition 5.30 which we will define below. Also P u n \mathcal{P}^{\mathrm{un}} P un is the unital point A ∞ A_\infty A ∞ -category from Definition 4.17, with Z \mathbb{Z} Z -filtration on Hom P u n ( ∗ , ∗ ) = R \operatorname{Hom}_{\mathcal{P}_{\mathrm{un}}}(*,*)=R Hom P un ( ∗ , ∗ ) = R as above, and Fun ^ A ∞ u n ( P u n , − ) \widehat{\operatorname{Fun}}_{\mathcal{A}_{\infty}}^{\mathrm{un}}(\mathcal{P}^{\mathrm{un}},-) Fun A ∞ un ( P un , − ) is as in (9.29), restricted to the 0-simplex P u n \mathcal{P}^{\mathrm{un}} P un in N ( A ∞ f i Cat R u n ) o p \mathcal{N}(\mathcal{A}^{\mathrm{fi}}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R})^{\mathrm{op}} N ( A ∞ fi Cat R un ) op , and Π A ∞ A ∞ \Pi_{\mathcal{A}_{\infty}}^{\mathcal{A}_{\infty}} Π A ∞ A ∞ is as in (8.8).
It will follow by comparing the definition with Definition 5.30 that the following diagram of ∞ \infty ∞ -functors strictly commutes:
N ( A ∞ Cat R u n ) → N ( Tw ) N ( A ∞ Cat R u n ) inc R u n ↓ ↓ inc R u n A ∞ Cat R u n , ∞ → Tw ∞ A ∞ Cat R u n , ∞ (275) \begin{CD}
\mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) @>{\mathcal{N}(\operatorname{Tw})}>> \mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) \\
@V{\operatorname{inc}^{\mathrm{un}}_{R}}VV @VV{\operatorname{inc}^{\mathrm{un}}_{R}}V \\
\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} @>{\operatorname{Tw}^{\infty}}>> \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} \tag*{(275)}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ N ( Tw ) Tw ∞ N ( A ∞ Cat R un ) ↓ ⏐ inc R un A ∞ Cat R un , ∞ (275)
Since Tw \operatorname{Tw} Tw maps A ∞ Cat R u n → A ∞ △ Cat R u n ⊂ A ∞ Cat R u n \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}\to\mathcal{A}^{\triangle}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}\subset\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un → A ∞ △ Cat R un ⊂ A ∞ Cat R un , it follows that Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ maps to A ∞ △ Cat R u n , ∞ ⊂ A ∞ Cat R u n , ∞ \mathcal{A}^{\triangle}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\subset\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ △ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ , as in (9.35), so we can rewrite (9.37) as
N ( A ∞ Cat R u n ) → N ( Tw ) N ( A ∞ △ Cat R u n ) inc R u n ↓ ↓ inc R u n A ∞ Cat R u n , ∞ → Tw ∞ A ∞ △ Cat R u n , ∞ (276) \begin{CD}
\mathcal{N}(\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) @>{\mathcal{N}(\operatorname{Tw})}>> \mathcal{N}(\mathcal{A}^{\triangle}_{\infty}\operatorname{Cat}^{\mathrm{un}}_{R}) \\
@V{\operatorname{inc}^{\mathrm{un}}_{R}}VV @VV{\operatorname{inc}^{\mathrm{un}}_{R}}V \\
\mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} @>{\operatorname{Tw}^{\infty}}>> \mathcal{A}^{\triangle}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} \tag*{(276)}
\end{CD} N ( A ∞ Cat R un ) inc R un ↓ ⏐ A ∞ Cat R un , ∞ N ( Tw ) Tw ∞ N ( A ∞ △ Cat R un ) ↓ ⏐ inc R un A ∞ △ Cat R un , ∞ (276)
To define Σ ∞ \Sigma^{\infty} Σ ∞ in (9.36), suppose N ≥ 0 N\geq0 N ≥ 0 and ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) ((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(\mathfrak{m}^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}) (( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) is an N N N -simplex in A ∞ Cat R u n , ∞ ⊂ A ^ ∞ Cat R u n , ∞ \mathcal{A}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R}\subset\widehat{\mathcal{A}}_{\infty}\operatorname{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ , as in Definitions 7.7 and 7.12. Define
Σ ∞ ( ( O i ) i = 0 N , ( f i j ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ ≠ I ⊆ { 0 , … , N } ) = ( ( P i ) i = 0 N , ( g i j ) 0 ≤ i ≤ j ≤ N , ( I i ) i = 0 N , ( n I ) ∅ ≠ I ⊆ { 0 , … , N } ) . (277) \begin{aligned}
\Sigma^{\infty}\bigl((\mathcal{O}^{i})_{i=0}^{N},(f^{ij})_{0\leq i\leq j\leq N},(\mathcal{H}^{i})_{i=0}^{N},(\mathfrak{m}^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\bigr) \\
=\bigl((\mathcal{P}^{i})_{i=0}^{N},(g^{ij})_{0\leq i\leq j\leq N},(\mathcal{I}^{i})_{i=0}^{N},(\mathfrak{n}^{I})_{\varnothing\neq I\subseteq\{0,\ldots,N\}}\bigr). \tag*{(277)}
\end{aligned} Σ ∞ ( ( O i ) i = 0 N , ( f ij ) 0 ≤ i ≤ j ≤ N , ( H i ) i = 0 N , ( m I ) ∅ = I ⊆ { 0 , … , N } ) = ( ( P i ) i = 0 N , ( g ij ) 0 ≤ i ≤ j ≤ N , ( I i ) i = 0 N , ( n I ) ∅ = I ⊆ { 0 , … , N } ) . (277)
where, generalizing Definition 5.30, we set:
(a) Elements of P i \mathcal{P}^{i} P i are triples ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) ((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) where l ≥ 0 l\geq0 l ≥ 0 , and x 1 , … , x l ∈ O i x_{1},\ldots,x_{l}\in\mathcal{O}^{i} x 1 , … , x l ∈ O i , and s i , f i ∈ Z s_{i},f_{i}\in\mathbb{Z} s i , f i ∈ Z for i = 1 , … , l i=1,\ldots,l i = 1 , … , l . As a shorthand we write ( x , s , f ) = ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) (x,s,f)=((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) ( x , s , f ) = (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) .
(b) For 0 ≤ i ≤ j ≤ N 0\leq i\leq j\leq N 0 ≤ i ≤ j ≤ N , define g i j : P i → P j g^{ij}:\mathcal{P}^{i}\to\mathcal{P}^{j} g ij : P i → P j by
g i j ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) = ( ( f i j ( x 1 ) , … , f i j ( x l ) ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) . g^{ij}((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l}))=((f^{ij}(x_{1}),\ldots,f^{ij}(x_{l})),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})). g ij (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) = (( f ij ( x 1 ) , … , f ij ( x l )) , ( s 1 , … , s l ) , ( f 1 , … , f l )) .
When I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_{0}<i_{1}<\cdots<i_{M}\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } , we define g I : P i 0 → P i M g^{I}:\mathcal{P}^{i_{0}}\to\mathcal{P}^{i_{M}} g I : P i 0 → P i M from the g i j g^{ij} g ij as in (7.2).
(c) Let ( ( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l ) ) ((x_{1},\ldots,x_{l}),(s_{1},\ldots,s_{l}),(f_{1},\ldots,f_{l})) (( x 1 , … , x l ) , ( s 1 , … , s l ) , ( f 1 , … , f l )) and ( ( x 1 ′ , … , x l ′ ′ ) , ( s 1 ′ , … , s l ′ ′ ) , ( f 1 ′ , … , f l ′ ′ ) ) ((x'_{1},\ldots,x'_{l'}),(s'_{1},\ldots,s'_{l'}),(f'_{1},\ldots,f'_{l'})) (( x 1 ′ , … , x l ′ ′ ) , ( s 1 ′ , … , s l ′ ′ ) , ( f 1 ′ , … , f l ′ ′ )) be elements of P i \mathcal{P}^{i} P i . Define a graded R R R -module
I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) = ⨁ d ∈ Z I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d \mathcal{I}^{i}((x,s,f),(x',s',f'))=\bigoplus_{d\in\mathbb{Z}}\mathcal{I}^{i}((x,s,f),(x',s',f'))^{d} I i (( x , s , f ) , ( x ′ , s ′ , f ′ )) = d ∈ Z ⨁ I i (( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d
I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d = ⨁ i = 1 , … , l j = 1 , … , l ′ H i ( x i , x j ′ ) d + s i − s j ′ . \mathcal{I}^{i}((x,s,f),(x',s',f'))^{d}
=
\bigoplus_{\substack{i=1,\ldots,l\\ j=1,\ldots,l'}}
\mathcal{H}^{i}(x_{i},x'_{j})^{d+s_{i}-s'_{j}}. I i (( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d = i = 1 , … , l j = 1 , … , l ′ ⨁ H i ( x i , x j ′ ) d + s i − s j ′ .
We write elements of I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) \mathcal{I}^{i}\bigl((\mathbf{x},s,\mathbf{f}),(\mathbf{x}',s',\mathbf{f}')\bigr) I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) as ( T i j ) i = 1 , … , l j = 1 , … , l ′ (\mathcal{T}_{ij})^{j=1,\ldots,l'}_{i=1,\ldots,l} ( T ij ) i = 1 , … , l j = 1 , … , l ′ in the obvious way. Define a Z \mathbb{Z} Z -filtration on I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d \mathcal{I}^{i}\bigl((\mathbf{x},s,\mathbf{f}),(\mathbf{x}',s',\mathbf{f}')\bigr)^{d} I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d by, for λ ∈ Z \lambda\in\mathbb{Z} λ ∈ Z
F λ I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d = ⨁ i = 1 , … , l , j = 1 , … , l ′ f i − f j ′ ≥ λ H ( x i , x j ′ ) d + s i − s j ′ . F^{\lambda}\mathcal{I}^{i}\bigl((\mathbf{x},s,\mathbf{f}),(\mathbf{x}',s',\mathbf{f}')\bigr)^{d}
=
\bigoplus_{\substack{i=1,\ldots,l,\ j=1,\ldots,l'\\ f_i-f'_j\geq\lambda}}
\mathcal{H}(x_i,x'_j)^{d+s_i-s'_j}. F λ I i ( ( x , s , f ) , ( x ′ , s ′ , f ′ ) ) d = i = 1 , … , l , j = 1 , … , l ′ f i − f j ′ ≥ λ ⨁ H ( x i , x j ′ ) d + s i − s j ′ .
(d) For all I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } I=\{i_0<i_1<\cdots<i_M\}\subseteq\{0,\ldots,N\} I = { i 0 < i 1 < ⋯ < i M } ⊆ { 0 , … , N } , k ≥ 0 k\geq0 k ≥ 0 and elements x 0 , … , x k ∈ O i 0 x_0,\ldots,x_k\in\mathcal{O}^{i_0} x 0 , … , x k ∈ O i 0 and elements ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) (\mathbf{x}^{0},s^{0},\mathbf{f}^{0}),\ldots,(\mathbf{x}^{k},s^{k},\mathbf{f}^{k}) ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) of I i 0 \mathcal{I}^{i_0} I i 0 , define a graded R R R -linear map of degree 2 − k 2-k 2 − k
n k I ; ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : I i 0 ( ( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 ) ) ⊗ R ⋯ ⊗ R I i 0 ( ( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k ) ) ⟶ I i M ( g I ( x 0 , s 0 , f 0 ) , g i 0 i M ( x k , s k , f k ) ) n_k^{I;(\mathbf{x}^{0},s^{0},\mathbf{f}^{0}),\ldots,(\mathbf{x}^{k},s^{k},\mathbf{f}^{k})}:
\mathcal{I}^{i_0}\bigl((\mathbf{x}^{0},s^{0},\mathbf{f}^{0}),(\mathbf{x}^{1},s^{1},\mathbf{f}^{1})\bigr)\otimes_R\cdots\otimes_R
\mathcal{I}^{i_0}\bigl((\mathbf{x}^{k-1},s^{k-1},\mathbf{f}^{k-1}),(\mathbf{x}^{k},s^{k},\mathbf{f}^{k})\bigr)
\longrightarrow
\mathcal{I}^{i_M}\bigl(g^{I}(\mathbf{x}^{0},s^{0},\mathbf{f}^{0}),g^{i_0i_M}(\mathbf{x}^{k},s^{k},\mathbf{f}^{k})\bigr) n k I ; ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) : I i 0 ( ( x 0 , s 0 , f 0 ) , ( x 1 , s 1 , f 1 ) ) ⊗ R ⋯ ⊗ R I i 0 ( ( x k − 1 , s k − 1 , f k − 1 ) , ( x k , s k , f k ) ) ⟶ I i M ( g I ( x 0 , s 0 , f 0 ) , g i 0 i M ( x k , s k , f k ) )
by, if k = 0 k=0 k = 0 and j 0 , j ~ 0 = 1 , … , l 0 j^0,\widetilde{j}^0=1,\ldots,l^0 j 0 , j 0 = 1 , … , l 0 then
( n 0 I ; ( x 0 , s 0 , f 0 ) ( 1 ) ) j 0 j ~ 0 = δ j 0 j ~ 0 m 0 I ; x 0 ( 1 ) , (278) \left(n_0^{I;(\mathbf{x}^{0},s^{0},\mathbf{f}^{0})}(1)\right)_{j^0\widetilde{j}^0}
=
\delta_{j^0\widetilde{j}^0}m_0^{I;\mathbf{x}^{0}}(1), \tag*{(278)} ( n 0 I ; ( x 0 , s 0 , f 0 ) ( 1 ) ) j 0 j 0 = δ j 0 j 0 m 0 I ; x 0 ( 1 ) , (278)
and if k ≥ 1 k\geq1 k ≥ 1 and j 0 = 1 , … , l 0 j^0=1,\ldots,l^0 j 0 = 1 , … , l 0 and j k = 1 , … , l k j^k=1,\ldots,l^k j k = 1 , … , l k then
( n k I ; ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 0 = 1 , … , l 0 i 1 = 1 , … , l 1 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k − 1 = 1 , … , l k − 1 i k = 1 , … , l k ) ) j 0 j k = ∑ j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 m k I ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) . \begin{aligned}
&\left(n_k^{I;(\mathbf{x}^{0},s^{0},\mathbf{f}^{0}),\ldots,(\mathbf{x}^{k},s^{k},\mathbf{f}^{k})}
\left(
(\mathcal{T}^{1}_{i_0i_1})^{i_1=1,\ldots,l^1}_{i_0=1,\ldots,l^0}
\otimes\cdots\otimes
(\mathcal{T}^{k}_{i_{k-1}i_k})^{i_k=1,\ldots,l^k}_{i_{k-1}=1,\ldots,l^{k-1}}
\right)\right)_{j^0j^k}\\
&\qquad=
\sum_{\substack{j^1=1,\ldots,l^1,\ldots,\\j^{k-1}=1,\ldots,l^{k-1}}}
m_k^{I;\mathbf{x}^{0}_{j^0},\mathbf{x}^{1}_{j^1},\ldots,\mathbf{x}^{k-1}_{j^{k-1}},\mathbf{x}^{k}_{j^k}}
\left(\mathcal{T}^{1}_{j^0j^1}\otimes\cdots\otimes\mathcal{T}^{k}_{j^{k-1}j^k}\right).
\end{aligned} ( n k I ; ( x 0 , s 0 , f 0 ) , … , ( x k , s k , f k ) ( ( T i 0 i 1 1 ) i 0 = 1 , … , l 0 i 1 = 1 , … , l 1 ⊗ ⋯ ⊗ ( T i k − 1 i k k ) i k − 1 = 1 , … , l k − 1 i k = 1 , … , l k ) ) j 0 j k = j 1 = 1 , … , l 1 , … , j k − 1 = 1 , … , l k − 1 ∑ m k I ; x j 0 0 , x j 1 1 , … , x j k − 1 k − 1 , x j k k ( T j 0 j 1 1 ⊗ ⋯ ⊗ T j k − 1 j k k ) .
It is now straightforward to show that (277) is an N N N -simplex in A ^ ∞ C a t R u n , ∞ \widehat{\mathcal{A}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ Cat R un , ∞ , and then in A ∞ f i C a t R u n , ∞ \mathcal{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ . In particular, the ( ⋯ ) i 0 i n j 0 j n (\cdots)^{j_0j_n}_{i_0i_n} ( ⋯ ) i 0 i n j 0 j n component of (227) –(228) for (9.39) is the sum over $j^1=1,\ldots,l^1,\ldots,j^{n-1}=1,\ldots,l^{n-1}$ of (7.15)–{xref}(7.16) for $\bigl((\mathcal{O}^{i}),(\mathbf{f}^{ij}),(\mathcal{H}^{i}),(\mathbf{m}^{I})\bigr)$ at $x^{0}_{j_0},\ldots,x^{n}_{j_n}$. By {xref}(9.40)`, the isomorphism conditions on n { i 0 , i 1 , i 2 } n_{\{i_0,i_1,i_2\}} n { i 0 , i 1 , i 2 } in Definition 8.7 follows from a product of the isomorphism conditions on m { i 0 , i 1 , i 2 } m_{\{i_0,i_1,i_2\}} m { i 0 , i 1 , i 2 } in Definition 7.12. The above definitions are clearly compatible with the actions of morphisms α : [ N ‾ ] → [ N ] \alpha:[\overline{N}]\to[N] α : [ N ] → [ N ] in Δ \Delta Δ on the simplicial sets, so Σ ∞ \Sigma^{\infty} Σ ∞ is an ∞ \infty ∞ -functor.
Remark 9.24. Combining equations (9.11), (9.12) and (9.38), we see that there exists an ∞ \infty ∞ -functor T w ~ ∞ \widetilde{\mathrm{Tw}}^{\infty} Tw ∞ making the following commute up to homotopy:
N ( A ∞ C a t R u n ) → N ( T w ) N ( A ∞ Δ C a t R u n ) Π l o c ∞ ↓ ↓ Π l o c ∞ A ∞ C a t R u n , ∞ → T w ∞ A ∞ Δ C a t R u n , ∞ ≃ ( 9.11 ) ↓ ↓ ≃ ( 9.12 ) A ∞ C a t R u n [ H E u n − 1 ] ∞ → T w ~ ∞ A ∞ Δ C a t R u n [ H E u n , Δ − 1 ] ∞ (279) \begin{CD}
\mathcal{N}\bigl(\mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}\bigr) @>{\mathcal{N}(\mathrm{Tw})}>> \mathcal{N}\bigl(\mathcal{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}\bigr)\\
@V{\Pi^{\infty}_{\mathrm{loc}}}VV @VV{\Pi^{\infty}_{\mathrm{loc}}}V\\
\mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} @>{\mathrm{Tw}^{\infty}}>> \mathcal{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}\\
@V{\simeq\ (9.11)}VV @VV{\simeq\ (9.12)}V\\
\mathcal{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{un}}]_{\infty} @>{\widetilde{\mathrm{Tw}}^{\infty}}>> \mathcal{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}^{-1}_{\mathrm{un},\Delta}]_{\infty} \tag*{(279)}
\end{CD} N ( A ∞ Cat R un ) Π loc ∞ ↓ ⏐ A ∞ Cat R un , ∞ ≃ ( 9.11 ) ↓ ⏐ A ∞ Cat R un [ H E un − 1 ] ∞ N ( Tw ) Tw ∞ Tw ∞ N ( A ∞ Δ Cat R un ) ↓ ⏐ Π loc ∞ A ∞ Δ Cat R un , ∞ ↓ ⏐ ≃ ( 9.12 ) A ∞ Δ Cat R un [ H E un , Δ − 1 ] ∞ (279)
Therefore Tw \operatorname{Tw} Tw in §4.9 maps H E u n − 1 → H E u n − 1 \mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1} \to\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1} H E un − 1 → H E un − 1 , so the outer rectangle
N ( A ∞ C a t R u n ) → N ( Tw ) N ( A ∞ Δ C a t R u n ) ↓ Π l o c ∞ ↓ Π l o c ∞ A ∞ C a t R u n [ H E u n − 1 ] ∞ → Tw ~ ∞ A ∞ Δ C a t R u n [ H E u n , Δ − 1 ] ∞ (280) \begin{aligned}
\mathcal{N}(\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}) &\xrightarrow{\mathcal{N}(\operatorname{Tw})} \mathcal{N}(\mathsf{A}^{\Delta}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}) \\
\downarrow{\Pi^{\infty}_{\mathrm{loc}}} && \downarrow{\Pi^{\infty}_{\mathrm{loc}}} \\
\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty} &\xrightarrow{\widetilde{\operatorname{Tw}}^{\infty}} \mathsf{A}^{\Delta}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un},\Delta}^{-1}]^{\infty} \tag*{(280)}
\end{aligned} N ( A ∞ Cat R un ) ↓ Π loc ∞ A ∞ Cat R un [ H E un − 1 ] ∞ N ( Tw ) N ( A ∞ Δ Cat R un ) Tw ∞ A ∞ Δ Cat R un [ H E un , Δ − 1 ] ∞ ↓ Π loc ∞ (280)
is obtained by localization. Presumably we could have proved directly that Tw \operatorname{Tw} Tw maps H E u n − 1 → H E u n − 1 \mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1} \to\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1} H E un − 1 → H E un − 1 , and so obtained (9.42) without first constructing Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ . But Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ has the advantage of being explicit.
Theorem 9.25. Let R R R be a commutative ring, and Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ , Tw ~ ∞ \widetilde{\operatorname{Tw}}^{\infty} Tw ∞ be as in (9.35) and (9.41). Then:
(a) There are ∞ \infty ∞ -natural isomorphisms of ∞ \infty ∞ -functors η : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ \eta: \operatorname{Tw}^{\infty} \Rightarrow\operatorname{Tw}^{\infty} \circ\operatorname{Tw}^{\infty} η : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ and η ~ : Tw ~ ∞ ⇒ Tw ~ ∞ ∘ Tw ~ ∞ \widetilde{\eta} : \widetilde{\operatorname{Tw}}^{\infty} \Rightarrow\widetilde{\operatorname{Tw}}^{\infty} \circ\widetilde{\operatorname{Tw}}^{\infty} η : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ , in the sense of Definition 2.12.
(b) Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ and Tw ~ ∞ \widetilde{\operatorname{Tw}}^{\infty} Tw ∞ are left adjoint to the fully faithful inclusion ∞ \infty ∞ -functors
inc : A ∞ Δ C a t R u n , ∞ ⟶ A ∞ C a t R u n , ∞ , \operatorname{inc} : \mathsf{A}^{\Delta}_{\infty}\mathsf{Cat}^{\mathrm{un},\infty}_{R} \longrightarrow\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un},\infty}_{R}, inc : A ∞ Δ Cat R un , ∞ ⟶ A ∞ Cat R un , ∞ , inc : A ∞ Δ C a t R u n [ H E u n , Δ − 1 ] ∞ ⟶ A ∞ C a t R u n [ H E u n − 1 ] ∞ . \operatorname{inc} : \mathsf{A}^{\Delta}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un},\Delta}^{-1}]^{\infty} \longrightarrow\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty}. inc : A ∞ Δ Cat R un [ H E un , Δ − 1 ] ∞ ⟶ A ∞ Cat R un [ H E un − 1 ] ∞ .
Proof. Recall from §4.9 that natural transformation inc : Id ⇒ Tw \operatorname{inc} : \operatorname{Id} \Rightarrow\operatorname{Tw} inc : Id ⇒ Tw of functors A ∞ C a t R u n → A ∞ C a t R u n \mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R} \to\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un → A ∞ Cat R un . Composing on the left and right with Tw \operatorname{Tw} Tw give two natural transformations id Tw ∗ inc , inc ∗ id Tw : Tw ⇒ Tw ∘ Tw \operatorname{id}_{\operatorname{Tw}} * \operatorname{inc}, \operatorname{inc} * \operatorname{id}_{\operatorname{Tw}} : \operatorname{Tw} \Rightarrow\operatorname{Tw} \circ\operatorname{Tw} id Tw ∗ inc , inc ∗ id Tw : Tw ⇒ Tw ∘ Tw of functors A ∞ C a t R u n → A ∞ C a t R u n \mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R} \to\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R} A ∞ Cat R un → A ∞ Cat R un . For each object A \mathcal{A} A this gives two A ∞ A_{\infty} A ∞ -functors Tw ( inc A ) , inc Tw A : Tw A → Tw Tw A \operatorname{Tw}(\operatorname{inc}_{\mathcal{A}}),\operatorname{inc}_{\operatorname{Tw}\mathcal{A}} : \operatorname{Tw}\mathcal{A} \to\operatorname{Tw}\operatorname{Tw}\mathcal{A} Tw ( inc A ) , inc Tw A : Tw A → Tw Tw A , both homotopy equivalences.
By properties of localizations, there is a homotopy commutative diagram
N ( A ∞ C a t R u n ) × Δ 1 → N ( inc ) N ( A ∞ C a t R u n ) ↓ Π l o c ∞ × id Δ 1 ↓ Π l o c ∞ A ∞ C a t R u n [ H E u n − 1 ] ∞ × Δ 1 → inc ~ ∞ A ∞ C a t R u n [ H E u n − 1 ] ∞ . \begin{aligned}
\mathcal{N}(\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}) \times\Delta^{1} &\xrightarrow{\mathcal{N}(\operatorname{inc})} \mathcal{N}(\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}) \\
\downarrow{\Pi^{\infty}_{\mathrm{loc}}\times\operatorname{id}_{\Delta^{1}}} && \downarrow{\Pi^{\infty}_{\mathrm{loc}}} \\
\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty} \times\Delta^{1} &\xrightarrow{\widetilde{\operatorname{inc}}^{\infty}} \mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty}.
\end{aligned} N ( A ∞ Cat R un ) × Δ 1 ↓ Π loc ∞ × id Δ 1 A ∞ Cat R un [ H E un − 1 ] ∞ × Δ 1 N ( inc ) N ( A ∞ Cat R un ) inc ∞ A ∞ Cat R un [ H E un − 1 ] ∞ . ↓ Π loc ∞
where N ( inc ) : N ( Id A ∞ C a t R u n ) ⇒ N ( Tw ) \mathcal{N}(\operatorname{inc}) : \mathcal{N}(\operatorname{Id}_{\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}}) \Rightarrow\mathcal{N}(\operatorname{Tw}) N ( inc ) : N ( Id A ∞ Cat R un ) ⇒ N ( Tw ) and inc ~ ∞ : Id A ∞ C a t R u n [ H E u n − 1 ] ∞ ⇒ Tw ~ ∞ \widetilde{\operatorname{inc}}^{\infty} : \operatorname{Id}_{\mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty}} \Rightarrow\widetilde{\operatorname{Tw}}^{\infty} inc ∞ : Id A ∞ Cat R un [ H E un − 1 ] ∞ ⇒ Tw ∞ are ∞ \infty ∞ -natural transformations, in the sense of Definition 2.12. Composing on the left and right with Tw ~ ∞ \widetilde{\operatorname{Tw}}^{\infty} Tw ∞ gives two ∞ \infty ∞ -natural transformations id Tw ~ ∞ ∗ inc ~ ∞ , inc ~ ∞ ∗ id Tw ~ ∞ : Tw ~ ∞ ⇒ Tw ~ ∞ ∘ Tw ~ ∞ \operatorname{id}_{\widetilde{\operatorname{Tw}}^{\infty}} * \widetilde{\operatorname{inc}}^{\infty},\widetilde{\operatorname{inc}}^{\infty} * \operatorname{id}_{\widetilde{\operatorname{Tw}}^{\infty}} : \widetilde{\operatorname{Tw}}^{\infty} \Rightarrow\widetilde{\operatorname{Tw}}^{\infty} \circ\widetilde{\operatorname{Tw}}^{\infty} id Tw ∞ ∗ inc ∞ , inc ∞ ∗ id Tw ∞ : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ . Up to equivalences, these act on objects A \mathcal{A} A by Π l o c ∞ ( Tw ( inc A ) ) , Π l o c ∞ ( inc Tw A ) \Pi^{\infty}_{\mathrm{loc}}(\operatorname{Tw}(\operatorname{inc}_{\mathcal{A}})),\Pi^{\infty}_{\mathrm{loc}}(\operatorname{inc}_{\operatorname{Tw}\mathcal{A}}) Π loc ∞ ( Tw ( inc A )) , Π loc ∞ ( inc Tw A ) above. Since Tw ( inc A ) , inc Tw A \operatorname{Tw}(\operatorname{inc}_{\mathcal{A}}),\operatorname{inc}_{\operatorname{Tw}\mathcal{A}} Tw ( inc A ) , inc Tw A are homotopy equivalences, their images in A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathsf{A}_{\infty}\mathsf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty} A ∞ Cat R un [ H E un − 1 ] ∞ are equivalences. Hence id Tw ~ ∞ ∗ inc ~ ∞ , inc ~ ∞ ∗ id Tw ~ ∞ \operatorname{id}_{\widetilde{\operatorname{Tw}}^{\infty}} * \widetilde{\operatorname{inc}}^{\infty},\widetilde{\operatorname{inc}}^{\infty} * \operatorname{id}_{\widetilde{\operatorname{Tw}}^{\infty}} id Tw ∞ ∗ inc ∞ , inc ∞ ∗ id Tw ∞ are ∞ \infty ∞ -natural isomorphisms.
For (a), we define η ~ = id Tw ~ ∞ ∗ inc ~ ∞ : Tw ~ ∞ ⇒ Tw ~ ∞ ∘ Tw ~ ∞ \widetilde{\eta} = \operatorname{id}_{\widetilde{\operatorname{Tw}}^{\infty}} * \widetilde{\operatorname{inc}}^{\infty} : \widetilde{\operatorname{Tw}}^{\infty} \Rightarrow\widetilde{\operatorname{Tw}}^{\infty} \circ\widetilde{\operatorname{Tw}}^{\infty} η = id Tw ∞ ∗ inc ∞ : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ . Then by (9.41), there is an ∞ \infty ∞ -natural isomorphism η : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ \eta: \operatorname{Tw}^{\infty} \Rightarrow\operatorname{Tw}^{\infty} \circ\operatorname{Tw}^{\infty} η : Tw ∞ ⇒ Tw ∞ ∘ Tw ∞ .
For (b), Lurie [60, Prop. 5.2.74 proves the following result. Suppose C \mathcal{C} C is an ∞ \infty ∞ -category, F : C → C F : \mathcal{C} \to\mathcal{C} F : C → C is an ∞ \infty ∞ -functor, and D ⊂ C \mathcal{D} \subset\mathcal{C} D ⊂ C is the essential image of τ \tau τ . Then the following are equivalent:
There exists an ∞ \infty ∞ -natural transformation α : Id C ⇒ F \alpha: \operatorname{Id}_{\mathcal{C}} \Rightarrow F α : Id C ⇒ F such that for each object x x x in C \mathcal{C} C , the morphisms F ( α ( x ) ) , α ( F ( x ) ) : F ( x ) → F ∘ F ( x ) F(\alpha(x)),\alpha(F(x)) : F(x) \to F \circ F(x) F ( α ( x )) , α ( F ( x )) : F ( x ) → F ∘ F ( x ) are equivalences in C \mathcal{C} C , that is, id F ∗ α , α ∗ id F : F ⇒ F ∘ F \operatorname{id}_{F} * \alpha,\alpha* \operatorname{id}_{F} : F \Rightarrow F \circ F id F ∗ α , α ∗ id F : F ⇒ F ∘ F are ∞ \infty ∞ -natural isomorphisms.
(ii) Regarded as an ∞ \infty ∞ -functor F : C → D F:\mathcal{C}\to\mathcal{D} F : C → D , F F F is left adjoint to the inclusion inc : D → C \operatorname{inc}:\mathcal{D}\to\mathcal{C} inc : D → C .
Apply this with C = A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathcal{C}=\mathbf{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty} C = A ∞ Cat R un [ H E un − 1 ] ∞ , F = Tw ~ ∞ F=\widetilde{\operatorname{Tw}}^{\infty} F = Tw ∞ , D = A ∞ Δ C a t R u n [ H E u n , Δ − 1 ] ∞ \mathcal{D}=\mathbf{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un},\Delta}^{-1}]^{\infty} D = A ∞ Δ Cat R un [ H E un , Δ − 1 ] ∞ , and α = inc ~ ∞ \alpha=\widetilde{\operatorname{inc}}^{\infty} α = inc ∞ . Then (i) holds from above, so by (ii) Tw ~ ∞ \widetilde{\operatorname{Tw}}^{\infty} Tw ∞ is left adjoint to inc : A ∞ Δ C a t R u n [ H E u n , Δ − 1 ] ∞ → A ∞ C a t R u n [ H E u n − 1 ] ∞ \operatorname{inc}:\mathbf{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un},\Delta}^{-1}]^{\infty}\to\mathbf{A}_{\infty}\mathrm{Cat}^{\mathrm{un}}_{R}[\mathcal{H}\mathcal{E}_{\mathrm{un}}^{-1}]^{\infty} inc : A ∞ Δ Cat R un [ H E un , Δ − 1 ] ∞ → A ∞ Cat R un [ H E un − 1 ] ∞ . The analogue for Tw ∞ \operatorname{Tw}^{\infty} Tw ∞ follows by equation (279) .
On future applications to Symplectic Geometry
Finally we outline how our theory could be used to define and study Fukaya categories in Symplectic Geometry. Let ( X , ω ) (X,\omega) ( X , ω ) be a suitable symplectic manifold. Symplectic Geometers would like to construct the derived Fukaya category D b F ( X , ω ) D^{b}\mathcal{F}(X,\omega) D b F ( X , ω ) , which should be a triangulated category. These have important applications, e.g. Kontsevich’s Homological Mirror Symmetry Conjecture [53] explains Mirror Symmetry of Calabi–Yau manifolds X , X ˇ X,\check{X} X , X ˇ in terms of equivalences of triangulated categories D b F ( X , ω ) ≃ D b coh X ˇ D^{b}\mathcal{F}(X,\omega)\simeq D^{b}\operatorname{coh}\check{X} D b F ( X , ω ) ≃ D b coh X ˇ , D b F ( X ˇ , ω ˇ ) ≃ D b coh X D^{b}\mathcal{F}(\check{X},\check{\omega})\simeq D^{b}\operatorname{coh}X D b F ( X ˇ , ω ˇ ) ≃ D b coh X .
See Auroux [2] for an introduction to (derived) Fukaya categories, Seidel [84] for a definition of D b F ( X , ω ) D^{b}\mathcal{F}(X,\omega) D b F ( X , ω ) under strong restrictions on ( X , ω ) (X,\omega) ( X , ω ) , and Fukaya–Oh–Ohta–Ono [29,33] for the foundations of a more general construction.
In the world of ∞ \infty ∞ -categories one should replace triangulated categories by stable ∞ \infty ∞ -categories, as in §2.7. Here is how we envisage using our theory to produced derived Fukaya categories D b F ( X , ω ) D^{b}\mathcal{F}(X,\omega) D b F ( X , ω ) as stable ∞ \infty ∞ -categories. We fix a suitable symplectic manifold ( X , ω ) (X,\omega) ( X , ω ) .
(a) Take the po-monoid ( Γ , ⪯ ) (\Gamma,\preceq) ( Γ , ⪯ ) to be ( R , ≤ ) (\mathbb{R},\leq) ( R , ≤ ) , and the complete R \mathbb{R} R -filtered ring R R R to be the Novikov ring Λ Q n o v \Lambda_{\mathbb{Q}}^{\mathrm{nov}} Λ Q nov or Λ Z n o v \Lambda_{\mathbb{Z}}^{\mathrm{nov}} Λ Z nov from Example 5.9, where we use Λ Q n o v \Lambda_{\mathbb{Q}}^{\mathrm{nov}} Λ Q nov if the relevant counts of J J J -holomorphic curves in ( X , ω ) (X,\omega) ( X , ω ) lie in Q \mathbb{Q} Q , and Λ Z n o v \Lambda_{\mathbb{Z}}^{\mathrm{nov}} Λ Z nov if they lie in Z \mathbb{Z} Z (typically this requires ( X , ω ) (X,\omega) ( X , ω ) ‘semipositive’). After choosing some auxiliary extra data D \mathcal{D} D on ( X , ω ) (X,\omega) ( X , ω ) , as in Fukaya–Oh–Ohta–Ono [29,33] we should define the Fukaya category F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D , a unital filtered A ∞ A_{\infty} A ∞ -category F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D over R R R , and so an object of A ∞ f i C a t R u n , ∞ \mathbf{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ . Objects of F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D are compact Lagrangians in ( X , ω ) (X,\omega) ( X , ω ) (maybe equipped with orientations, gradings, . . . ).
The auxiliary data D \mathcal{D} D should include a choice of almost complex structure J J J on X X X compatible with ω \omega ω (and possibly domain-dependent). It could also include families of perturbations of Lagrangians as in Seidel [84], choices of Morse functions, . . . . The details depend on the set-up, and there are several different approaches in the literature.
(b) Having constructed the unital filtered A ∞ A_{\infty} A ∞ -category F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D , we define the derived Fukaya category D b F ( X , ω ) D D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D , a stable ∞ \infty ∞ -category to be the composition of ∞ \infty ∞ -functors from (8.8), (9.16), (9.33), and (9.35) applied to F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D :
A ∞ f i C a t R u n , ∞ → Str ∞ A ∞ f i C a t R , 2 , 2 u n , ∞ → Π A ∞ A ∞ C a t R u n , ∞ → Tw ∞ A ∞ Δ C a t R u n , ∞ → N A ∞ ∞ C a t ∞ s t . (281) \mathbf{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}
\xrightarrow{\operatorname{Str}^{\infty}}
\mathbf{A}^{\mathrm{fi}}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R,2,2}
\xrightarrow{\Pi^{A_{\infty}}}
\mathbf{A}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}
\xrightarrow{\operatorname{Tw}^{\infty}}
\mathbf{A}^{\Delta}_{\infty}\mathrm{Cat}^{\mathrm{un},\infty}_{R}
\xrightarrow{\mathcal{N}^{\infty}_{A_{\infty}}}
\mathcal{C}\mathit{at}^{\mathrm{st}}_{\infty}. \tag*{(281)} A ∞ fi Cat R un , ∞ Str ∞ A ∞ fi Cat R , 2 , 2 un , ∞ Π A ∞ A ∞ Cat R un , ∞ Tw ∞ A ∞ Δ Cat R un , ∞ N A ∞ ∞ C at ∞ st . (281)
We write Der R f i \operatorname{Der}^{\mathrm{fi}}_{R} Der R fi for this composition. So far we have not used ∞ \infty ∞ -categories in an interesting way, we could have defined D b F ( X , ω ) D D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D using ordinary categories.
(c) We now want to consider how F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D and D b F ( X , ω ) D D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D depend on the auxiliary data D \mathcal{D} D in (a). To do this, we should define a contractible Kan complex S X , ω \mathcal{S}_{X,\omega} S X , ω whose 0-simplices are choices of data D \mathcal{D} D , and an ∞ \infty ∞ -functor F : S X , ω → A ∞ f i C a t R u n , ∞ F:\mathcal{S}_{X,\omega}\to\mathbf{A}^{\mathrm{fi}}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} F : S X , ω → A ∞ fi Cat R un , ∞ which maps F : D ↦ F ( X , ω ) D F:\mathcal{D}\mapsto\mathcal{F}(X,\omega)_{\mathcal{D}} F : D ↦ F ( X , ω ) D on 0-simplices.
To see why this might be reasonable, consider the (oversimplified) case in which D \mathcal{D} D is just a choice of an arbitrary almost complex structure J J J on X X X compatible with ω \omega ω . Then we take S X , ω \mathcal{S}_{X,\omega} S X , ω to be the simplicial set whose N N N -simplices { J x : x ∈ Δ t o p N } \{J_x:x\in\Delta_{\mathrm{top}}^{N}\} { J x : x ∈ Δ top N } are smooth families of almost complex structures J x J_x J x on X X X compatible with ω \omega ω parametrized by points x x x in the topological N N N -simplex Δ t o p N \Delta_{\mathrm{top}}^{N} Δ top N of (2.4), thought of as a smooth manifold with corners. Equivalently, N N N -simplices are smooth maps Δ t o p N → J X , ω \Delta_{\mathrm{top}}^{N}\to\mathcal{J}_{X,\omega} Δ top N → J X , ω , where J X , ω \mathcal{J}_{X,\omega} J X , ω is the infinite-dimensional manifold of almost complex structures on X X X compatible with ω \omega ω . Since J X , ω \mathcal{J}_{X,\omega} J X , ω is contractible, S X , ω \mathcal{S}_{X,\omega} S X , ω is a contractible Kan complex.
It is important here that F F F maps to our explicit ∞ \infty ∞ -category A ∞ f i C a t R u n , ∞ \mathbf{A}^{\mathrm{fi}}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} A ∞ fi Cat R un , ∞ , whose N N N -simplices involve collections of maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k satisfying (8.4)–(8.5), and not to a non-explicit localized ∞ \infty ∞ -category such as A ∞ C a t R u n [ H E u n − 1 ] ∞ \mathbf{A}_{\infty}\mathbf{Cat}^{\mathrm{un}}_{R}[\mathcal{H}E^{-1}_{\mathrm{un}}]_{\infty} A ∞ Cat R un [ H E un − 1 ] ∞ . The definition of F F F will be a generalization of the definition of F ( X , ω ) D \mathcal{F}(X,\omega)_{\mathcal{D}} F ( X , ω ) D in (a), which reduces to this for 0-simplices. For each N N N -simplex in S X , ω \mathcal{S}_{X,\omega} S X , ω , we arrange to define collections of maps m k I ; x 0 , … , x k m^{I;x_{0},\ldots,x_{k}}_{k} m k I ; x 0 , … , x k by defining virtual chains for moduli spaces of J x J_x J x -holomorphic curves for x ∈ Δ t o p N x\in\Delta_{\mathrm{top}}^{N} x ∈ Δ top N , and prove they satisfy equations (8.4)–(8.5). Guillem Cazassus, Alexander Ritter and the author are working on a construction of this kind [16].
(d) We have an ∞ \infty ∞ -functor Der R f i ∘ F : S X , ω → C a t ∞ s t \operatorname{Der}^{\mathrm{fi}}_{R}\circ F:\mathcal{S}_{X,\omega}\to\mathbf{Cat}^{\mathrm{st}}_{\infty} Der R fi ∘ F : S X , ω → Cat ∞ st which maps D ↦ D b F ( X , ω ) D \mathcal{D}\mapsto D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D ↦ D b F ( X , ω ) D on 0-simplices. As S X , ω \mathcal{S}_{X,\omega} S X , ω is a connected Kan complex, if D 1 , D 2 \mathcal{D}_{1},\mathcal{D}_{2} D 1 , D 2 are choices of auxiliary data we have a Joyal equivalence D b F ( X , ω ) D 1 ≃ D b F ( X , ω ) D 2 D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}_{1}}\simeq D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}_{2}} D b F ( X , ω ) D 1 ≃ D b F ( X , ω ) D 2 . As S X , ω \mathcal{S}_{X,\omega} S X , ω is a contractible Kan complex, these equivalences are coherent up to a contractible choice in an ∞ \infty ∞ -categorical sense.
Note that the ∞ \infty ∞ -category formalism gives a very clean way to say how D b F ( X , ω ) D D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D depends on the auxiliary data D \mathcal{D} D .
(e) An alternative way to define D b F ( X , ω ) D D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} D b F ( X , ω ) D for auxiliary data D \mathcal{D} D is by the Cartesian diagram of simplicial sets
D b F ( X , ω ) D → N Δ ∞ C a t R u n , ∞ ↓ ↓ p Δ ∗ → Tw ∞ ∘ Π A ∞ ∞ ∘ Str ∞ ∘ F ∘ D A ∞ Δ C a t R u n , ∞ \begin{CD}
D^{b}\mathcal{F}(X,\omega)_{\mathcal{D}} @>>> N\Delta_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R}\\
@VVV @VV p^{\Delta} V\\
* @>{\operatorname{Tw}^{\infty}\circ\Pi_{A^{\infty}_{\infty}}\circ\operatorname{Str}^{\infty}\circ F\circ\mathcal{D}}>> A^{\Delta}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R}
\end{CD} D b F ( X , ω ) D ↓ ⏐ ∗ Tw ∞ ∘ Π A ∞ ∞ ∘ Str ∞ ∘ F ∘ D N Δ ∞ Cat R un , ∞ ↓ ⏐ p Δ A ∞ Δ Cat R un , ∞
Here p Δ : N Δ ∞ C a t R u n , ∞ → A ∞ Δ C a t R u n , ∞ p^{\Delta}:N\Delta_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R}\to A^{\Delta}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} p Δ : N Δ ∞ Cat R un , ∞ → A ∞ Δ Cat R un , ∞ is the restriction to A ∞ Δ C a t R u n , ∞ ⊂ A ∞ C a t R u n , ∞ A^{\Delta}_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R}\subset A_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} A ∞ Δ Cat R un , ∞ ⊂ A ∞ Cat R un , ∞ of the cocartesian fibration p R ∞ : N ∞ C a t R u n , ∞ → A ∞ C a t R u n , ∞ p^{\infty}_{R}:N_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R}\to A_{\infty}\mathbf{Cat}^{\mathrm{un},\infty}_{R} p R ∞ : N ∞ Cat R un , ∞ → A ∞ Cat R un , ∞ from Definition 9.11 and Theorem 9.12, and ∗ = Δ 0 *=\Delta^{0} ∗ = Δ 0 is the point simplicial set, and we regard D \mathcal{D} D as a morphism D : ∗ → S X , ω \mathcal{D}:*\to\mathcal{S}_{X,\omega} D : ∗ → S X , ω . The idea here is that by Theorem 9.12(b), the final morphism N A ∞ ∞ \mathcal{N}^{\infty}_{A_{\infty}} N A ∞ ∞ in (9.43) maps an object A \mathcal{A} A to the fibre of p Δ p^{\Delta} p Δ over A \mathcal{A} A .
Define the big derived Fukaya category D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) big by the Cartesian diagram of simplicial sets
D b F ( X , ω ) b i g → N ∞ Cat R u n , ∞ Π b i g ↓ ↓ p Δ S X , ω → Tw ∞ ∘ Π A ∞ A ∞ ∘ Str ∞ ∘ F A ∞ Δ Cat R u n , ∞ (282) \begin{CD}
D^b\mathcal{F}(X,\omega)^{\mathrm{big}} @>>> N\infty\operatorname{Cat}_{R}^{\mathrm{un},\infty} \\
@V{\Pi^{\mathrm{big}}}VV @VV{p^\Delta}V \\
\mathcal{S}_{X,\omega} @>{\operatorname{Tw}^{\infty}\circ\Pi_{\mathcal{A}_\infty}^{\mathcal{A}_\infty}\circ\operatorname{Str}^{\infty}\circ F}>> \mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} \tag*{(282)}
\end{CD} D b F ( X , ω ) big Π big ↓ ⏐ S X , ω Tw ∞ ∘ Π A ∞ A ∞ ∘ Str ∞ ∘ F N ∞ Cat R un , ∞ ↓ ⏐ p Δ A ∞ Δ Cat R un , ∞ (282)
As p Δ p^\Delta p Δ is a cocartesian fibration and S X , ω \mathcal{S}_{X,\omega} S X , ω is a Kan complex, D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) big is an ∞ \infty ∞ -category. Since S X , ω \mathcal{S}_{X,\omega} S X , ω is contractible, D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) big is unchanged up to equivalence by replacing S X , ω \mathcal{S}_{X,\omega} S X , ω by a point D \mathcal{D} D of S X , ω \mathcal{S}_{X,\omega} S X , ω . That is, for any D \mathcal{D} D in S X , ω \mathcal{S}_{X,\omega} S X , ω , there is an embedding of ∞ \infty ∞ -categories D b F ( X , ω ) D ↪ D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)_{\mathcal{D}}\hookrightarrow D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) D ↪ D b F ( X , ω ) big which is a Joyal equivalence.
Thus D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) big is an alternative model for the derived Fukaya category of ( X , ω ) (X,\omega) ( X , ω ) which has the advantage that it is independent of choices D \mathcal{D} D . It is constructed from not one, but a huge number, of equivalent A ∞ A_\infty A ∞ -categories.
(f) Here is a (speculative) application of these ideas. Suppose G G G is a Lie group with an action on X X X by symplectomorphisms which admits a moment map μ : X → g ∗ \mu:X\to\mathfrak{g}^* μ : X → g ∗ . Define a simplicial group G Δ G^\Delta G Δ – a group in simplicial sets – such that N N N -simplices of G Δ G^\Delta G Δ are smooth maps S : Δ t o p N → G S:\Delta^N_{\mathrm{top}}\to G S : Δ top N → G , regarding Δ t o p N \Delta^N_{\mathrm{top}} Δ top N as a smooth manifold with corners. Then we should write down a simplicial action of G Δ G^\Delta G Δ on S X , ω \mathcal{S}_{X,\omega} S X , ω . The ∞ \infty ∞ -functor Tw ∞ ∘ Π A ∞ A ∞ ∘ Str ∞ ∘ F : S X , ω → A ∞ Δ Cat R u n , ∞ \operatorname{Tw}^{\infty}\circ\Pi_{\mathcal{A}_\infty}^{\mathcal{A}_\infty}\circ\operatorname{Str}^{\infty}\circ F:\mathcal{S}_{X,\omega}\to\mathcal{A}^{\Delta}_{\infty}\operatorname{Cat}_{R}^{\mathrm{un},\infty} Tw ∞ ∘ Π A ∞ A ∞ ∘ Str ∞ ∘ F : S X , ω → A ∞ Δ Cat R un , ∞ should be invariant under G Δ G^\Delta G Δ up to coherent isomorphisms. Thus there is a simplicial action of G Δ G^\Delta G Δ on D b F ( X , ω ) b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}} D b F ( X , ω ) big such that Π b i g \Pi^{\mathrm{big}} Π big in (282) is G Δ G^\Delta G Δ -equivariant.
Define the big G G G -equivariant derived Fukaya category D b F ( X , ω ) G b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}}_G D b F ( X , ω ) G big to be the simplicial set of G Δ G^\Delta G Δ -equivariant sections s : S X , ω → D b F ( X , ω ) b i g s:\mathcal{S}_{X,\omega}\to D^b\mathcal{F}(X,\omega)^{\mathrm{big}} s : S X , ω → D b F ( X , ω ) big of Π b i g \Pi^{\mathrm{big}} Π big , that is, Π b i g ∘ s = id S X , ω \Pi^{\mathrm{big}}\circ s=\operatorname{id}_{\mathcal{S}_{X,\omega}} Π big ∘ s = id S X , ω . We claim that this is a stable ∞ \infty ∞ -category. We expect that under good conditions, D b F ( X , ω ) G b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}}_G D b F ( X , ω ) G big will roughly be a direct sum of stable ∞ \infty ∞ -subcategories D b F ( X , ω ) G , η b i g D^b\mathcal{F}(X,\omega)^{\mathrm{big}}_{G,\eta} D b F ( X , ω ) G , η big for each η ∈ Z ( g ∗ ) ⊂ g ∗ \eta\in Z(\mathfrak{g}^*)\subset\mathfrak{g}^* η ∈ Z ( g ∗ ) ⊂ g ∗ , which should roughly coincide with the derived Fukaya category of the symplectic quotient X / / η G = μ − 1 ( η ) / G X//_{\eta}G=\mu^{-1}(\eta)/G X / / η G = μ − 1 ( η ) / G when this is nonsingular.
Note that this construction would be impossible without ∞ \infty ∞ -categories.
Proofs of results in Chapters 6, 7 and 8
Proofs of Theorems 6.15 and 7.6
We prove Theorem 7.6(a), which implies Theorem 6.15(a) by restricting to A ∞ A_\infty A ∞ -categories A \mathcal{A} A with O = { ∗ } \mathcal{O}=\{\ast\} O = { ∗ } . If n < k n<k n < k then by Definition 7.2(b) both sides of (204) are zero. If n ≥ k n\geq k n ≥ k we have
d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = ∑ 0 ≤ j < k , y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) = ( − 1 ) k ′ ′ ( k ′ + n ) . \begin{aligned}
d^{\{i_0\};x_0,\ldots,x_n}_{n,k;y_0,\ldots,y_k}
=
\sum_{\substack{0\leq j<k,\\ y_i=x_i,\ 0\leq i\leq j,\\ y_i=x_{n-k+i},\ j<i\leq k}}
(-1)^{j+(n-k+1)(k-j-1)}
\operatorname{id}_{\mathcal{H}^{i_0}(x_0,x_1)}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j-1},x_j)}\otimes
m^{\{i_0\};x_j,x_{j+1},\ldots,x_{j+n-k+1}}_{n-k+1}
\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{j+n-k+1},x_{j+n-k+2})}\otimes\cdots\otimes
\operatorname{id}_{\mathcal{H}^{i_0}(x_{n-1},x_n)} \\
&= (-1)^{k''(k'+n)}.
\end{aligned} d n , k ; y 0 , … , y k { i 0 } ; x 0 , … , x n = 0 ≤ j < k , y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ∑ ( − 1 ) j + ( n − k + 1 ) ( k − j − 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m n − k + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) = ( − 1 ) k ′′ ( k ′ + n ) .
( ∑ 0 ≤ j < k ′ y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ( − 1 ) j + ( k ′ − j − 1 ) ( ( n − k ′ ′ ) − k ′ + 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m ( n − k ′ ′ ) − k ′ + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − k ′ ′ − 1 , x n − k ′ ′ ) ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) ) + ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ ( ∑ 0 ≤ j ~ < k ′ ′ y i = x i , 0 ≤ i ≤ k ′ + j ~ , y i = x n − k + i , k ′ + j ~ < i ≤ k ( − 1 ) j ~ + ( k ′ ′ − j ~ − 1 ) ( ( n − k ′ ) − k ′ ′ + 1 ) id H i 0 ( x k ′ , x k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( x k ′ + j ~ − 1 , x k ′ + j ~ ) ⊗ m ( n − k ′ ) − k ′ ′ + 1 { i 0 } ; x k ′ + j ~ , x k ′ + j ~ + 1 , … , x j ~ + n − k ′ ′ + 1 ⊗ id H i 0 ( x j ~ + n − k ′ ′ + 1 , x j ~ + n − k ′ ′ + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) ) \begin{aligned}
&\left(\sum_{\substack{0\le j<k'\\ y_i=x_i,\ 0\le i\le j,\\ y_i=x_{n-k+i},\ j<i\le k}}
(-1)^{j+(k'-j-1)((n-k'')-k'+1)}
\operatorname{id}_{\mathcal H^{i_0}}(x_0,x_1)\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(x_{j-1},x_j)\right.\\
&\qquad\otimes m^{\{i_0\};x_j,x_{j+1},\ldots,x_{j+n-k+1}}_{(n-k'')-k'+1}
\otimes\operatorname{id}_{\mathcal H^{i_0}}(x_{j+n-k+1},x_{j+n-k+2})\\
&\qquad\left.\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(x_{n-k''-1},x_{n-k''})
\otimes\operatorname{id}_{\mathcal H^{i_0}}(y_{k'},y_{k'+1})\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(y_{k-1},y_k)\right)\\
&\quad+(-1)^{k'}\operatorname{id}_{\mathcal H^{i_0}}(y_0,y_1)\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(y_{k'-1},y_{k'})\\
&\quad\otimes\left(\sum_{\substack{0\le\widetilde j<k''\\ y_i=x_i,\ 0\le i\le k'+\widetilde j,\\ y_i=x_{n-k+i},\ k'+\widetilde j<i\le k}}
(-1)^{\widetilde j+(k''-\widetilde j-1)((n-k')-k''+1)}
\operatorname{id}_{\mathcal H^{i_0}}(x_{k'},x_{k'+1})\otimes\cdots\right.\\
&\qquad\otimes\operatorname{id}_{\mathcal H^{i_0}}(x_{k'+\widetilde j-1},x_{k'+\widetilde j})
\otimes m^{\{i_0\};x_{k'+\widetilde j},x_{k'+\widetilde j+1},\ldots,x_{\widetilde j+n-k''+1}}_{(n-k')-k''+1}\\
&\qquad\left.\otimes\operatorname{id}_{\mathcal H^{i_0}}(x_{\widetilde j+n-k''+1},x_{\widetilde j+n-k''+2})\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(x_{n-1},x_n)\right)
\end{aligned} 0 ≤ j < k ′ y i = x i , 0 ≤ i ≤ j , y i = x n − k + i , j < i ≤ k ∑ ( − 1 ) j + ( k ′ − j − 1 ) (( n − k ′′ ) − k ′ + 1 ) id H i 0 ( x 0 , x 1 ) ⊗ ⋯ ⊗ id H i 0 ( x j − 1 , x j ) ⊗ m ( n − k ′′ ) − k ′ + 1 { i 0 } ; x j , x j + 1 , … , x j + n − k + 1 ⊗ id H i 0 ( x j + n − k + 1 , x j + n − k + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − k ′′ − 1 , x n − k ′′ ) ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) ) + ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ 0 ≤ j < k ′′ y i = x i , 0 ≤ i ≤ k ′ + j , y i = x n − k + i , k ′ + j < i ≤ k ∑ ( − 1 ) j + ( k ′′ − j − 1 ) (( n − k ′ ) − k ′′ + 1 ) id H i 0 ( x k ′ , x k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( x k ′ + j − 1 , x k ′ + j ) ⊗ m ( n − k ′ ) − k ′′ + 1 { i 0 } ; x k ′ + j , x k ′ + j + 1 , … , x j + n − k ′′ + 1 ⊗ id H i 0 ( x j + n − k ′′ + 1 , x j + n − k ′′ + 2 ) ⊗ ⋯ ⊗ id H i 0 ( x n − 1 , x n ) )
= { ( − 1 ) k ′ ′ ( k ′ + n ) d n − k ′ ′ , k ′ ; y 0 , … , y k ′ { i 0 } ; x 0 , … , x n − k ′ ′ ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) , if y i = x n − k + i , k ′ ≤ i ≤ k , 0 , otherwise + { ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ d n − k ′ , k ′ ′ ; y k ′ , … , y k { i 0 } ; x k ′ , … , x n , if y i = x i , 0 ≤ i ≤ k ′ , 0 , otherwise. (283) \begin{aligned}
={}&
\begin{cases}
(-1)^{k''(k'+n)}d^{\{i_0\};x_0,\ldots,x_{n-k''}}_{n-k'',k';y_0,\ldots,y_{k'}}
\otimes\operatorname{id}_{\mathcal H^{i_0}}(y_{k'},y_{k'+1})\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(y_{k-1},y_k),
&\text{if }y_i=x_{n-k+i},\ k'\le i\le k,\\
0,&\text{otherwise}
\end{cases}\\
&+
\begin{cases}
(-1)^{k'}\operatorname{id}_{\mathcal H^{i_0}}(y_0,y_1)\otimes\cdots\otimes
\operatorname{id}_{\mathcal H^{i_0}}(y_{k'-1},y_{k'})
\otimes d^{\{i_0\};x_{k'},\ldots,x_n}_{n-k',k'';y_{k'},\ldots,y_k},
&\text{if }y_i=x_i,\ 0\le i\le k',\\
0,&\text{otherwise.}
\end{cases} \tag*{(283)}
\end{aligned} = { ( − 1 ) k ′′ ( k ′ + n ) d n − k ′′ , k ′ ; y 0 , … , y k ′ { i 0 } ; x 0 , … , x n − k ′′ ⊗ id H i 0 ( y k ′ , y k ′ + 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k − 1 , y k ) , 0 , if y i = x n − k + i , k ′ ≤ i ≤ k , otherwise + { ( − 1 ) k ′ id H i 0 ( y 0 , y 1 ) ⊗ ⋯ ⊗ id H i 0 ( y k ′ − 1 , y k ′ ) ⊗ d n − k ′ , k ′′ ; y k ′ , … , y k { i 0 } ; x k ′ , … , x n , 0 , if y i = x i , 0 ≤ i ≤ k ′ , otherwise. (283)
where in the first and third steps we use (7.8), and in the second step we split the sum ∑ j = 0 k − 1 \sum_{j=0}^{k-1} ∑ j = 0 k − 1 into two sums ∑ j = 0 k ′ − 1 \sum_{j=0}^{k'-1} ∑ j = 0 k ′ − 1 and ∑ j = k ′ k ′ + k ′ ′ − 1 \sum_{j=k'}^{k'+k''-1} ∑ j = k ′ k ′ + k ′′ − 1 , replace j j j by j ~ = j − k ′ \widetilde j=j-k' j = j − k ′ in the second sum, and rewrite the signs.
For Theorems 6.15(b) and 7.6(b), for simplicity we will prove equation (6.16) in Theorem 6.15 rather than equation (7.13) in Theorem 7.6, that is, we will omit objects x 0 , … , x n , y 0 , … , y k x_0,\ldots,x_n,y_0,\ldots,y_k x 0 , … , x n , y 0 , … , y k from operators d^I_{n,k;y_0,\ldots,y_k}^{x_0,\ldots,x_n} . We could include objects as in equation (10.1), but this adds notational complexity without changing the important ideas. With I = { i 0 < ⋯ < i M } I=\{i_0<\cdots<i_M\} I = { i 0 < ⋯ < i M } for M > 0 M>0 M > 0 , consider the long equation
d n , k I = ∑ isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with n i ≥ 0 leaves, 1 ≤ i ≤ k , n = n 1 + ⋯ + n k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 ( − 1 ) ( k + 1 ) ( M + n + 1 ) ( ∏ v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ and v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) ) ⋅ C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k ) = ∑ 0 < l ≤ M , n = n ′ + n ′ ′ , n ′ , n ′ ′ ≥ 0 , 0 = j 0 < j 1 < ⋯ < j l = M ( − 1 ) ( k + 1 ) ( M + n + 1 ) ⋅ ( − 1 ) ∑ a = 1 l − 1 ( j a − a ) ⋅ ( − 1 ) n ′ ( n ′ ′ − k ′ ′ ) + ( n ′ − k ′ ) ( k ′ ′ − n ′ ′ + 1 − l ) ⋅ ( − 1 ) ( k ′ ′ + 1 ) ( l + n ′ ′ + 1 ) \begin{aligned}
d^I_{n,k}
={}&\sum_{\substack{\text{isomorphism classes}\\\text{of }\Gamma_1,\ldots,\Gamma_k:\\
\Gamma_i\text{ is a balanced}\\
I\text{-labelled tree}\\
\text{with }n_i\ge0\text{ leaves,}\\
1\le i\le k,\\
n=n_1+\cdots+n_k,\\
L(\Gamma_1)=I,\\
R(\Gamma_k)=\{i_0,i_M\},\\
R(\Gamma_i)=L(\Gamma_{i+1}),\\
i=1,\ldots,k-1}}
(-1)^{(k+1)(M+n+1)}
\left(\prod_{\substack{v\text{ internal vertex or bud of }\Gamma_i,\ 1\le i\le k,\\
w\text{ internal vertex or bud of }\Gamma_{i'},\ 1\le i'\le k,\\
\text{either }i<i',\text{ or }i=i'\text{ and }v\prec w\text{ in order }\prec\text{ of }\Gamma_i,\\
v\text{ has }p\text{ incoming edges including }q\text{ leaves},\ w\text{ has }r\text{ incoming edges}}}
(-1)^{N(v,w)+q(r+1)+(p+1)(|I_w|+r+1)}
\right)\\
&\qquad\cdot C(\Gamma_1)\otimes C(\Gamma_2)\otimes\cdots\otimes C(\Gamma_k)\\
={}&\sum_{\substack{0<l\le M,\ n=n'+n'',\ n',n''\ge0,\\
0=j_0<j_1<\cdots<j_l=M}}
(-1)^{(k+1)(M+n+1)}\cdot(-1)^{\sum_{a=1}^{l-1}(j_a-a)}
\cdot(-1)^{n'(n''-k'')+(n'-k')(k''-n''+1-l)}\\
&\qquad\cdot(-1)^{(k''+1)(l+n''+1)}
\end{aligned} d n , k I = = isomorphism classes of Γ 1 , … , Γ k : Γ i is a balanced I -labelled tree with n i ≥ 0 leaves, 1 ≤ i ≤ k , n = n 1 + ⋯ + n k , L ( Γ 1 ) = I , R ( Γ k ) = { i 0 , i M } , R ( Γ i ) = L ( Γ i + 1 ) , i = 1 , … , k − 1 ∑ ( − 1 ) ( k + 1 ) ( M + n + 1 ) v internal vertex or bud of Γ i , 1 ≤ i ≤ k , w internal vertex or bud of Γ i ′ , 1 ≤ i ′ ≤ k , either i < i ′ , or i = i ′ and v ≺ w in order ≺ of Γ i , v has p incoming edges including q leaves , w has r incoming edges ∏ ( − 1 ) N ( v , w ) + q ( r + 1 ) + ( p + 1 ) ( ∣ I w ∣ + r + 1 ) ⋅ C ( Γ 1 ) ⊗ C ( Γ 2 ) ⊗ ⋯ ⊗ C ( Γ k ) 0 < l ≤ M , n = n ′ + n ′′ , n ′ , n ′′ ≥ 0 , 0 = j 0 < j 1 < ⋯ < j l = M ∑ ( − 1 ) ( k + 1 ) ( M + n + 1 ) ⋅ ( − 1 ) ∑ a = 1 l − 1 ( j a − a ) ⋅ ( − 1 ) n ′ ( n ′′ − k ′′ ) + ( n ′ − k ′ ) ( k ′′ − n ′′ + 1 − l ) ⋅ ( − 1 ) ( k ′′ + 1 ) ( l + n ′′ + 1 )