betaFeedback

WorksMathematics

Quasi-F-split heights and derived Morita invariance

Abstract

In this paper, we study quasi-F-splitting from the viewpoint of non-commutative algebraic geometry. Let k be a perfect field of characteristic p. For a proper dg-category over k with a Serre functor, we define quasi-F^e-split heights for e\geq 1 by using finite topological restriction homology with coefficients in a shift of the Serre functor. Using the functoriality of polygonic topological Hochschild homology for laced categories due to Harpaz–Nikolaus–Saunier, we show that these heights are invariant under derived Morita equivalence. For a proper Gorenstein scheme of pure dimension over k, the height for e=1 of its dg-category of perfect complexes coincides with the quasi-F-split height in the sense of Yobuko. For a complete Noetherian Gorenstein local kk-algebra with residue field k, the dg-category of perfect complexes supported at the closed point recovers the n-quasi-F^e-splitting of Tanaka–Witaszek–Yobuko for all n,e\geq 1. The proofs use the computation of the zeroth homotopy groups of TR with coefficients due to Dotto–Krause–Nikolaus–Patchkoria, Zariski descent, and Serre duality or local duality.

Introduction

The aim of this paper is to construct an analogue of the quasi-FF-split height for non-commutative algebraic varieties, and to compare it with the classical invariants of proper Gorenstein schemes and complete Gorenstein local rings.

Let kk be a perfect field of characteristic p>0p > 0. Quasi-FF-splitting was introduced by Yobuko [21] as a generalization of FF-splitting, in which a splitting of the Frobenius map is replaced by a splitting involving Witt vectors of finite length. It has applications in birational geometry [11], and the variant with higher Frobenius exponents, quasi-FeF^e-splitting, was introduced by Tanaka–Witaszek–Yobuko [20]. Fedder-type criteria make these notions computable for hypersurfaces [12, 22].

The definition of quasi-FF-splitting involves the structure sheaf, the module structure on Witt vectors and a specific map ΦX,n\Phi_{X,n} (see § 2), whereas a general non-commutative algebra has no ring endomorphism a↦apa \mapsto a^p.

In this paper, we use finite topological restriction homology TRn\mathrm{TR}^n, which has Frobenius and restriction maps, in place of Witt vectors. In order to obtain the splitting condition for a proper scheme from its dg-category of perfect complexes, we take coefficients in the Serre functor. For a Gorenstein scheme, this coefficient corresponds to the canonical bundle. Then the lowest homotopy group of a certain pushout of TR\mathrm{TR} spectra is identified with the top cohomology of the classical pushout QX,nQ_{X,n} tensored with the canonical bundle, and Serre duality translates the injectivity of the induced map into the existence of the required splitting.

The categorical construction

Throughout this paper, all dg-categories are assumed to be small, idempotent-complete and kk-linear, and to admit a classical generator. Equivalently, such a dg-category is of the form Perf⁡(A)\operatorname{Perf}(A) for a dg-algebra AA over kk. Recall that a dg-category T\mathcal{T} is proper if the total cohomology of Hom⁡T(x,y)\operatorname{Hom}_{\mathcal{T}}(x,y) is finite-dimensional for all objects x,yx,y. Let T≃Perf⁡(A)\mathcal{T} \simeq\operatorname{Perf}(A) be a proper dg-category. We say that T\mathcal{T} admits a Serre functor if the dual bimodule

A∨:=RHom⁡k(A,k)A^\vee:= \operatorname{RHom}_k(A,k)

is invertible with respect to the derived tensor product. In this case, ST:=−⊗ALA∨S_{\mathcal{T}} := - \otimes_A^{\mathrm{L}} A^\vee is a Serre functor of T\mathcal{T}, that is, there are natural equivalences RHom⁡T(x,y)∨≃RHom⁡T(y,STx)\operatorname{RHom}_{\mathcal{T}}(x,y)^\vee\simeq\operatorname{RHom}_{\mathcal{T}}(y,S_{\mathcal{T}}x), and we always regard STS_{\mathcal{T}} as a dg-functor in this way. These notions do not depend on the choice of AA.

Fix an integer dd, and put PT:=ST[−d]P_{\mathcal{T}} := S_{\mathcal{T}}[-d], which we regard as a coefficient bimodule. We use the polygonic spectra of Krause–McCandless–Nikolaus [9], which refine topological Hochschild homology with coefficients. For an exact endofunctor PP of T\mathcal{T}, we denote by TRm(T;P)\mathrm{TR}^m(\mathcal{T};P) the finite TR\mathrm{TR} with respect to the truncation set {1,p,…,pm−1}\{1,p,\ldots,p^{m-1}\}; as usual, we have TR1(T;P)=THH(T;P)\mathrm{TR}^1(\mathcal{T};P)=\mathrm{THH}(\mathcal{T};P). There are restriction and Frobenius maps

res⁡ ⁣:TRm+1(T;P)⟶TRm(T;P),F ⁣:TRm+1(T;P)⟶TRm(T;P⊗p),\operatorname{res}\colon\mathrm{TR}^{m+1}(\mathcal{T};P)\longrightarrow\mathrm{TR}^{m}(\mathcal{T};P), \qquad F\colon\mathrm{TR}^{m+1}(\mathcal{T};P)\longrightarrow\mathrm{TR}^{m}(\mathcal{T};P^{\otimes p}),

where the Frobenius changes the coefficient from PP to P⊗pP^{\otimes p}.

For n,e≥1n,e \ge1, we define

Qne(T;P):=cofib⁡[TRn+e(T;P)→(Fe, −res⁡n+e−1)TRn(T;P⊗pe)⊕THH(T;P)],(1)Q^e_n(\mathcal{T};P):=\operatorname{cofib}\left[\mathrm{TR}^{n+e}(\mathcal{T};P)\xrightarrow{(F^e,\,-\operatorname{res}^{n+e-1})}\mathrm{TR}^{n}(\mathcal{T};P^{\otimes p^e})\oplus\mathrm{THH}(\mathcal{T};P)\right], \tag*{(1)}

and we denote by ηne(T;P) ⁣:THH(T;P)→Qne(T;P)\eta^e_n(\mathcal{T};P)\colon\mathrm{THH}(\mathcal{T};P)\to Q^e_n(\mathcal{T};P) the induced map. We define

ht⁡de(T):=inf⁡{n≥1∣π−d(ηne(T;PT)) is injective},\operatorname{ht}^e_d(\mathcal{T}):=\inf\{n\ge1\mid\pi_{-d}(\eta^e_n(\mathcal{T};P_{\mathcal{T}}))\text{ is injective}\},

where we put inf⁡∅=∞\inf\varnothing=\infty. The integer dd is part of the data.

Theorem A (Morita invariance; Subsection 4.2). *Fix an integer dd. Let T\mathcal{T} and U\mathcal{U} be proper dg-categories admitting Serre functors. Then a kk-linear Morita equivalence T≃U\mathcal{T}\simeq\mathcal{U} induces an equivalence between the diagrams defining ηne(T;PT)\eta^e_n(\mathcal{T};P_{\mathcal{T}}) and ηne(U;PU)\eta^e_n(\mathcal{U};P_{\mathcal{U}}) for all n,e≥1n,e\ge1, including the Frobenius and restriction maps. In particular, we have

ht⁡de(T)=ht⁡de(U)(e≥1).\operatorname{ht}^e_d(\mathcal{T})=\operatorname{ht}^e_d(\mathcal{U}) \qquad(e\ge1).

Moreover, if π−d(ηne)\pi_{-d}(\eta_n^e) is injective, then so is π−d(ηn+1e)\pi_{-d}(\eta_{n+1}^e).

Theorems A and B rely on the functoriality of polygonic THH with respect to laced functors due to Harpaz–Nikolaus–Saunier (Theorem 3.2). A construction is given in the thesis of Saunier [17 Theorem 6.30] and in his lecture notes [18 Corollary 7.3.14], and complete proofs are announced in [7, 6]; see Remark 3.3. Theorem C below does not depend on it.

Comparison with the classical invariants

For a scheme XX over Fp\mathbf{F}_p, we denote by ht⁡(X)\operatorname{ht}(X) the quasi-FF-split height of XX in the sense of Yobuko (Definition 2.3). For a Noetherian FF-finite Gorenstein local Fp\mathbf{F}_p-algebra RR and e≥1e \ge1, we denote by ht⁡de(R)\operatorname{ht}_d^e(R) the smallest integer nn such that RR is nn-quasi-FeF^e-split in the sense of [20], in the form given in [22 Definition 2.1] (Definition 2.4).

Theorem B (Proper Gorenstein schemes; Subsection 7.1 and Corollary 7.4). Let XX be a nonempty proper Gorenstein scheme over kk of pure dimension dd, and put PX:=SPerf⁡(X)[−d]P_X := \mathcal{S}_{\operatorname{Perf}(X)}[-d]. Then, for every n≥1n \ge1, the scheme XX is nn-quasi-FF-split if and only if the map

π−d(ηne(Perf⁡(X);PX))\pi_{-d}\left(\eta_n^e(\operatorname{Perf}(X);P_X)\right)

is injective. In particular, we have

ht⁡d1(Perf⁡(X))=ht⁡(X).\operatorname{ht}_d^1(\operatorname{Perf}(X))=\operatorname{ht}(X).

Moreover, if YY is another such scheme and Perf⁡(X)≃Perf⁡(Y)\operatorname{Perf}(X) \simeq\operatorname{Perf}(Y) as kk-linear dg-categories, then dim⁡X=dim⁡Y\dim X = \dim Y and ht⁡(X)=ht⁡(Y)\operatorname{ht}(X) = \operatorname{ht}(Y).

Theorem B applies to singular Gorenstein schemes and to smooth proper varieties with nontrivial canonical bundle. The Serre functor cannot be replaced by the identity coefficient: for an ordinary smooth projective curve XX of genus at least two, the Frobenius acts bijectively on H1(X,OX)H^1(X,\mathcal{O}_X), so that the construction with the identity coefficient gives an injective map, whereas XX is not quasi-FF-split (Remark 7.5).

Theorem C (Complete local Gorenstein rings; Subsection 8.4). Let (R,m,k)(R,\mathfrak{m},k) be a nonzero complete Noetherian Gorenstein local kk-algebra with residue field kk, and put d=dim⁡Rd = \dim R. Let TR:=Perf⁡m(R)\mathcal{T}_R := \operatorname{Perf}_{\mathfrak{m}}(R) be the dg-category of perfect RR-complexes supported at m\mathfrak{m}. Then TR\mathcal{T}_R is proper, and its Serre functor is [d][d]; in particular, PTR≃id⁡P_{\mathcal{T}_R} \simeq\operatorname{id}. Moreover, for all n,e≥1n,e \ge1, the ring RR is nn-quasi-FeF^e-split if and only if the map

π−d(ηne(TR;id⁡))\pi_{-d}\left(\eta_n^e(\mathcal{T}_R;\operatorname{id})\right)

is injective. In particular, we have

ht⁡de(Perf⁡m(R))=ht⁡e(R).\operatorname{ht}_d^e(\operatorname{Perf}_{\mathfrak{m}}(R))=\operatorname{ht}^e(R).

The ring RR in Theorem C is automatically FF-finite, since it is a quotient of k[[x1,…,xN]]k[[x_1,\ldots,x_N]] by the Cohen structure theorem. Theorem C applies to complete hypersurfaces R=k[[x1,…,xN]]/(f)R=k[[x_1,\ldots,x_N]]/(f), where ff is a nonzero element of the maximal ideal. In this case, we have ht⁡de(R)=ht⁡N−1e(Perf⁡m(R))\operatorname{ht}_d^e(R)=\operatorname{ht}_{N-1}^e(\operatorname{Perf}_{\mathfrak{m}}(R)), and the Fedder-type criterion of Yoshikawa [22 Theorem A] computes this categorical invariant. For example, if p=2p=2, kk is algebraically closed and R=k[[x,y,z]]/(z2+x2y+xyn)R=k[[x,y,z]]/(z^2+x^2y+xy^n) with n≥2n \ge2, then ht⁡21(Perf⁡m(R))=⌈log⁡2n⌉\operatorname{ht}_2^1(\operatorname{Perf}_{\mathfrak{m}}(R))=\lceil\log_2 n\rceil by Theorem C and [13 Theorem 4.31]; see Example 8.8. The dg-category TR\mathcal{T}_R consists of perfect complexes with finite length cohomology; it is not the singularity category.

Relation to previous work

Yobuko [21 Theorem 4.5] shows that the quasi-FF-split height of a Calabi–Yau variety coincides with the height of its Artin–Mazur formal group. Antieau–Bragg [1 Proposition 1.1 and Theorem 5.1] prove that this formal group, and hence its height, is a derived invariant. Therefore the derived invariance of the quasi-FF-split height of Calabi–Yau varieties, including K3 surfaces, is already known.

For smooth projective varieties, the consequence ht⁡(X)=ht⁡(Y)\operatorname{ht}(X)=\operatorname{ht}(Y) of Theorem B is new only in a limited range. If ωX\omega_X is ample or anti-ample, then X≃YX \simeq Y by Bondal–Orlov [BO01]. If the Kodaira dimension of XX is positive, then so is that of YY (see [8 Section 6.1]), and neither XX nor YY is quasi-FF-split by [11 Proposition 3.14]. Thus Theorem B gives new information for smooth projective varieties of nonpositive Kodaira dimension which are neither Calabi–Yau nor determined by their derived categories, and for singular Gorenstein schemes. In a related direction, Krah–Vial [KV25] prove that global FF-regularity is a derived invariant for normal Gorenstein projective varieties.

Our computation relies on the theory of Witt vectors with coefficients due to Dotto–Krause–Nikolaus–Patchkoria [4] and on the polygonic spectra of [9]. The localization of THH with coefficients, which we need for schemes, is provided by the theory of laced categories due to Harpaz–Nikolaus–Saunier [7], and for THH and TR without coefficients we use the localization theorems of Blumberg–Mandell [2]. Liu–Liu [15 Theorems 3.9 and 3.11] prove that coefficient Morita equivalences induce equivalences of THH with coefficients and of the associated genuine restriction systems; Theorem 3.2, due to Harpaz–Nikolaus–Saunier, provides the corresponding functoriality for polygonic spectra. On the classical side, Lemma 7.2 and Proposition 8.5 are versions of the cohomological criteria in [11 Lemmas 2.10 and 3.13], and dualizing complexes on Witt rings are studied in [10 Section 9].

The global comparison (Theorem B) concerns only the case e=1e = 1, whereas the local comparison (Theorem C) holds for all e≥1e \ge1 and uses Matlis duality over Wn(R)W_n(R). If e>1e > 1, then the classical pushout QX,neQ^e_{X,n} is only a WnOXW_n\mathcal{O}_X-module, and a global comparison would require duality for WnOXW_n\mathcal{O}_X-modules. This leads to the following question.

Question 1.1. Does Theorem B extend to all e≥1e \ge1, with respect to the definition of quasi-FF-splitting for schemes in [20]?

We do not discuss quasi-FF-regularity in this paper. Its definition [20] involves a multiplier cc which runs over all non-zero-divisors, and Theorem A does not provide the corresponding structure on dg-categories.

Organization of the paper

In § 2, we fix conventions and recall the definitions of quasi-FF-splitting and quasi-FeF_e-splitting. In § 3, we recall finite TR with coefficients and its functoriality for laced categories, and we deduce its invariance under derived Morita equivalence (Corollary 3.4). In § 4, we define the invariants htde\mathrm{htd}^e and prove Theorem A. In § 5, we compute the zeroth homotopy groups of TR with coefficients in a line bundle. In § 6, we prove Zariski descent for TR with coefficients in a line bundle, and we compute the lowest homotopy group of derived global sections. Using these results, we prove Theorem B in § 7 and Theorem C in § 8, where we also give examples.

Use of AI. The proof strategy of this paper is due to the author. In working out the details of the arguments, the author benefited greatly from ChatGPT Astra 6. Claude Opus 5.5 was of great help in writing this paper, in particular with its English and grammar. The author takes full responsibility for all the contents of this paper.

Conventions and classical definitions

Conventions

Throughout this paper, we fix a prime pp and a perfect field kk of characteristic pp. All tensor products of bimodules are derived, and we often omit L\mathbf{L} from the notation. We regard a dg-algebra as an associative ring spectrum via the Eilenberg–MacLane functor. We always consider THH relative to the sphere spectrum.

For a complex KK, we use the convention πi(K)=H−i(K)\pi_i(K)=H^{-i}(K), so that the shift [1][1] corresponds to the suspension. A spectrum is connective if its negative homotopy groups vanish. For a finite group GG, we denote by (−)hG(-)^{hG}, (−)hG(-)_{hG} and (−)tG(-)^{tG} the homotopy fixed points, the homotopy orbits and the Tate construction, respectively.

For an exact endofunctor PP of a stable ∞\infty-category C\mathcal{C}, we consider the associated bimodule

MP(x,y):=Map⁡C(x,P(y)).M_P(x,y):=\operatorname{Map}_{\mathcal{C}}(x,P(y)).

With this convention, the bimodule associated with −⊗L-\otimes L for a line bundle LL is LL. If C≃Perf⁡(A)\mathcal{C}\simeq\operatorname{Perf}(A), we use the same symbol for an endofunctor and the corresponding AA-bimodule. Following [7], we call a pair (C,M)(\mathcal{C},M) of a stable ∞\infty-category C\mathcal{C} and a C\mathcal{C}-bimodule MM a laced category.

Witt vectors and quasi-FF-splitting

Let AA be an Fp\mathbb{F}_p-algebra. We denote by Wn(A)W_n(A) the ring of pp-typical Witt vectors of length nn, by [a]∈Wn(A)[a]\in W_n(A) the Teichmüller lift of a∈Aa\in A, and by

res⁡:Wn(A)⟶Wn−1(A),V:Wn−1(A)⟶Wn(A)\operatorname{res}: W_n(A)\longrightarrow W_{n-1}(A),\qquad V: W_{n-1}(A)\longrightarrow W_n(A)

the restriction and the Verschiebung. Let FA:A→AF_A:A\to A be the absolute Frobenius. The induced ring endomorphism Wn(FA)W_n(F_A) of Wn(A)W_n(A) is given by (a0,…,an−1)↦(a0p,…,an−1p)(a_0,\ldots,a_{n-1})\mapsto(a_0^p,\ldots,a_{n-1}^p) and preserves the length. We distinguish it from the Frobenius F:Wn+1(A)→Wn(A)F:W_{n+1}(A)\to W_n(A), which decreases the length and corresponds to the Frobenius of finite TR (Proposition 3.5). For a Wn(A)W_n(A)-module MM and e≥1e\geq1, we denote by F∗eMF_*^eM the Wn(A)W_n(A)-module obtained by restriction of scalars along Wn(FA)eW_n(F_A)^e.

Definition 2.1. Let n,e≥1n,e\geq1. We define a Wn(A)W_n(A)-module QA,neQ^e_{A,n} and a Wn(A)W_n(A)-linear map ΦA,ne:A→QA,ne\Phi^e_{A,n}:A\to Q^e_{A,n} by the pushout diagram of Wn(A)W_n(A)-modules

Wn(A)→Wn(FA)eF∗eWn(A)res⁡n−1↓↓A→ΦA,neQA,ne\begin{CD} W_n(A) @>{W_n(F_A)^e}>> F_*^eW_n(A) \\ @V{\operatorname{res}^{n-1}}VV @VVV \\ A @>{\Phi^e_{A,n}}>> Q^e_{A,n} \end{CD}

where AA is regarded as a Wn(A)W_n(A)-module via res⁡n−1\operatorname{res}^{n-1}. We put QA,n:=QA,n1Q_{A,n}:=Q^1_{A,n} and ΦA,n:=ΦA,n1\Phi_{A,n}:=\Phi^1_{A,n}.

If e≥2e\geq2, then QA,neQ^e_{A,n} is not an AA-module in general; for e=1e=1, we have the following description.

Lemma 2.2. The Wn(A)W_n(A)-module structure on QA,nQ_{A,n} factors through res⁡n−1:Wn(A)→A\operatorname{res}^{n-1}:W_n(A)\to A, and there is a natural isomorphism

QA,n≃F∗(Wn(A)/pWn(A)).Q_{A,n}\simeq F_* (W_n(A)/pW_n(A)).

Under this isomorphism, the AA-module structure on QA,nQ_{A,n} and the map ΦA,n\Phi_{A,n} are given by

a⋅w‾=[a]pw‾,ΦA,n(a)=[a]p‾.a\cdot\overline{w}=[a]^p\overline{w},\qquad\Phi_{A,n}(a)=\overline{[a]^p}.

Proof. Since res⁡n−1:Wn(A)→A\operatorname{res}^{n-1}:W_n(A)\to A is surjective with kernel VWn−1(A)VW_{n-1}(A), the pushout is the quotient of F∗Wn(A)F_*W_n(A) by the image of VWn−1(A)VW_{n-1}(A) under Wn(FA)W_n(F_A). Since AA has characteristic pp, we have

Wn(FA)(VWn−1(A))=pWn(A).W_n(F_A)(VW_{n-1}(A))=pW_n(A).

Hence the elements of VWn−1(A)VW_{n-1}(A) act on the quotient by zero, and the module structure factors through AA. The formulas follow by evaluating on Teichmüller lifts. □\square

This construction is compatible with localization, and hence it sheafifies on Fp\mathbb{F}_p-schemes. For an Fp\mathbb{F}_p-scheme XX, we denote by QX,nQ_{X,n} and ΦX,n:OX→QX,n\Phi_{X,n}:\mathcal{O}_X\to Q_{X,n} the resulting OX\mathcal{O}_X-module and map.

Definition 2.3. Let XX be an Fp\mathbb{F}_p-scheme and n≥1n\geq1. We say that XX is nn-quasi-FF-split if ΦX,n\Phi_{X,n} admits an OX\mathcal{O}_X-linear retraction. The quasi-FF-split height of XX is defined by

ht⁡(X):=inf⁡{n≥1∣X is n-quasi-F-split}.\operatorname{ht}(X):=\inf\{n\geq1\mid X\text{ is }n\text{-quasi-}F\text{-split}\}.

This definition agrees with [21 Definition 4.1].

Definition 2.4. Let RR be a Noetherian FF-finite Gorenstein local Fp\mathbf{F}_p-algebra, and let WnωRW_n\omega_R be the canonical (dualizing) module of Wn(R)W_n(R) (see [10 Section 9]). Following [20 Sections 3.1 and 3.5] in the form of [22 Definition 2.1], we say that RR is nn-quasi-FeF^e-split if the map

(ΦR,ne)∗:Hom⁡Wn(R)(QR,ne,WnωR)⟶Hom⁡Wn(R)(R,WnωR)(2)(\Phi^e_{R,n})^*: \operatorname{Hom}_{W_n(R)}(Q^e_{R,n},W_n\omega_R)\longrightarrow\operatorname{Hom}_{W_n(R)}(R,W_n\omega_R) \tag*{(2)}

is surjective. We denote by ht⁡e(R)\operatorname{ht}^e(R) the smallest such integer nn, where we put ht⁡e(R)=∞\operatorname{ht}^e(R)=\infty if no such nn exists.

Finite TR with coefficients

Polygonic spectra and finite TR

3.1. Polygonic spectra and finite TR. For a ring spectrum AA and an AA-bimodule PP, we put

X(A;P)r:=THH⁡(A;P⊗Ar)(r≥1).\mathcal{X}(A;P)_r:=\operatorname{THH}(A;P^{\otimes_A r})\qquad(r\geq1).

The cyclic bar construction gives a CrC_r-action on this spectrum, and the Tate diagonal gives natural maps

φp,r:X(A;P)r⟶X(A;P⊗Ap)prtCp,\varphi_{p,r}:\mathcal{X}(A;P)_r\longrightarrow\mathcal{X}(A;P^{\otimes_A p})_{pr}^{tC_p},

which are equivariant with respect to the residual CrC_r-actions. By [9 Theorem 6.31], these data form a polygonic spectrum X(A;P)\mathcal{X}(A;P), functorially in the AA-bimodule PP. We only use the part indexed by powers of pp.

Definition 3.1. For m≥1m\geq1, we put Tm:={1,p,…,pm−1}T_m:=\{1,p,\ldots,p^{m-1}\} and define

TR⁡m(A;P):=TR⁡Tm(X(A;P)),\operatorname{TR}^m(A;P):=\operatorname{TR}_{T_m}(\mathcal{X}(A;P)),

where the right-hand side is the finite TR of polygonic spectra with respect to the truncation set TmT_m, which is defined as a mapping spectrum in [9 Section 3].

Explicitly, writing Xr=X(A;P)r\mathcal{X}_r=\mathcal{X}(A;P)_r, the spectrum TR⁡m(A;P)\operatorname{TR}^m(A;P) is the equalizer of two maps

∏0≤i<mXpihCpi⇉∏1≤i<m(XpitCpi)hCpi−1,\prod_{0\leq i<m}\mathcal{X}_{p^i}^{hC_{p^i}}\rightrightarrows\prod_{1\leq i<m}(\mathcal{X}_{p^i}^{tC_{p^i}})^{hC_{p^{i-1}}},

one of which is induced by the canonical maps and the other by the maps φp\varphi_p. In particular, finite TR preserves finite limits of polygonic spectra.

The inclusion Tm⊂Tm+1T_m\subset T_{m+1} induces the restriction map res⁡:TR⁡m+1(A;P)⟶TR⁡m(A;P)\operatorname{res}:\operatorname{TR}^{m+1}(A;P)\longrightarrow\operatorname{TR}^m(A;P). The shift of X(A;P)\mathcal{X}(A;P) by pp is identified with X(A;P⊗Ap)\mathcal{X}(A;P^{\otimes_A p}), and the Frobenius of finite TR gives a map

F:TR⁡m+1(A;P)⟶TR⁡m(A;P⊗Ap).(3)F:\operatorname{TR}^{m+1}(A;P)\longrightarrow\operatorname{TR}^m(A;P^{\otimes_A p}). \tag*{(3)}

These maps commute with the restriction maps and can be iterated.

Functoriality and Morita invariance

3.2. Functoriality and Morita invariance. We use the following functoriality of polygonic THH for laced categories, due to Harpaz–Nikolaus–Saunier.

Theorem 3.2 (Harpaz–Nikolaus–Saunier). The following assertions hold.

(i) The construction (C,M)↦(THH⁡(C;M∘r))r≥1(\mathcal{C},M)\mapsto(\operatorname{THH}(\mathcal{C};M^{\circ r}))_{r\geq1}, together with the cyclic actions and the Tate diagonals, extends to a functor X\mathcal{X} from the ∞\infty-category of laced categories to the ∞\infty-category of polygonic spectra in the sense of [9]; see [17 Theorem 6.30] and [18 Corollary 7.3.14]. For M=id⁡M=\operatorname{id}, it recovers the cyclotomic structure on THH⁡(C)\operatorname{THH}(\mathcal{C}).

(ii) If AA is a connective ring spectrum and PP is a connective AA-bimodule, then X(Perf⁡(A),P)\mathcal{X}(\operatorname{Perf}(A),P) is naturally equivalent to the polygonic spectrum X(A;P)\mathcal{X}(A;P) of [9 Theorem 6.31]; see [17 Section 7.2].

(iii) The structure maps satisfy φp,r(C,M)=φp,1(C,M∘r)\varphi_{p,r}(\mathcal{C},M)=\varphi_{p,1}(\mathcal{C},M^{\circ r}), naturally in (C,M)(\mathcal{C},M); see [17 Section 6.4] and [18 Section 7.2]. In particular, the shift of X(C,M)\mathcal{X}(\mathcal{C},M) by pp is identified with X(C,M∘p)\mathcal{X}(\mathcal{C},M^{\circ p}), naturally in (C,M)(\mathcal{C},M).

Remark 3.3. The construction in [17 Section 6.4] is presented as a sketch, statement (ii) is stated without proof in [17 Section 7.2] (see also [17 Proposition 7.16]), and [18] consists of lecture notes. A genuine refinement of (i) is stated in [17 Theorem 7.18], and complete proofs are announced in [7 Introduction] and [6]. Theorems A and B depend on Theorem 3.2. Theorem C does not: its coefficient is equivalent to the identity, and if TRm(TR;id)\mathrm{TR}^{m}(\mathrm{TR};\mathrm{id}) is defined via the cyclotomic spectrum THH(TR)\mathrm{THH}(\mathrm{TR}), then its proof only uses the functoriality of X(A;P)\mathcal{X}(A;P) in PP for a fixed dg-algebra AA [9 Theorem 6.31], the comparison of TR of cyclotomic spectra with the classical TR [9], and the invariance and localization theorems for THH and TR [2].

For a laced category (C,P)(C,P), we write THH(C;P)\mathrm{THH}(C;P) for the first level of X(C,P)\mathcal{X}(C,P) and TRm(C;P):=TRTm(X(C,P))\mathrm{TR}^{m}(C;P):=\mathrm{TR}_{T_m}(\mathcal{X}(C,P)). By Theorem 3.2, the restriction and Frobenius maps of finite TR are natural with respect to laced functors.

Corollary 3.4. Let AA and BB be dg-algebras, and let MM be an AA-BB-bimodule and NN a BB-AA-bimodule such that

M⊗BN≃A,N⊗AM≃BM \otimes_B N \simeq A,\qquad N \otimes_A M \simeq B

as bimodules. For an AA-bimodule PP, we put P′:=N⊗AP⊗AMP' := N \otimes_A P \otimes_A M. Then there is an equivalence of polygonic spectra

X(Perf⁡(A),P)≃X(Perf⁡(B),P′),\mathcal{X}(\operatorname{Perf}(A),P)\simeq\mathcal{X}(\operatorname{Perf}(B),P'),

which induces equivalences between the finite TR spectra compatible with the Frobenius and restriction maps.

Proof. The functor −⊗AM:Perf⁡(A)→Perf⁡(B)-\otimes_A M:\operatorname{Perf}(A)\to\operatorname{Perf}(B) is an equivalence, and we have a natural equivalence (−⊗AP)⊗AM≃(−⊗AM)⊗BP′(-\otimes_A P)\otimes_A M\simeq(-\otimes_A M)\otimes_B P' since M⊗BN≃AM\otimes_B N\simeq A. Hence it underlies an equivalence of laced categories (Perf⁡(A),P)≃(Perf⁡(B),P′)(\operatorname{Perf}(A),P)\simeq(\operatorname{Perf}(B),P'). By Theorem 3.2(i), it induces an equivalence of polygonic spectra, which is compatible with the identifications of the shifts by pp by Theorem 3.2(iii). Since finite TR and its restriction and Frobenius maps are natural in polygonic spectra, we obtain the claim. □

In particular, for a dg-category T\mathcal{T} and an exact endofunctor PP of T\mathcal{T}, the spectra THH(T;P)\mathrm{THH}(\mathcal{T};P) and TRm(T;P)\mathrm{TR}^{m}(\mathcal{T};P), together with their restriction and Frobenius maps, are invariant under derived Morita equivalence. If T≃Perf⁡(A)\mathcal{T}\simeq\operatorname{Perf}(A) with AA and PP connective, then they are computed by X(A;P)\mathcal{X}(A;P) by Theorem 3.2(ii).

Connectivity and the zeroth homotopy groups.

Proposition 3.5. Let AA be a connective ring spectrum and PP a connective AA-bimodule. Then TRm(A;P)\mathrm{TR}^{m}(A;P) is connective, and there is a natural isomorphism

π0TRm(A;P)≃Wm(π0A;π0P)\pi_0\mathrm{TR}^{m}(A;P)\simeq W_m(\pi_0A;\pi_0P)

compatible with restriction, Frobenius and Verschiebung. Here Wm(π0A;π0P)W_m(\pi_0A;\pi_0P) denotes the Witt vectors of length mm with coefficients in π0P\pi_0P in the sense of [4].

Proof. The fiber of res:TRm+1(A;P)→TRm(A;P)\mathrm{res}:\mathrm{TR}^{m+1}(A;P)\to\mathrm{TR}^{m}(A;P) is given by the homotopy orbits of the corresponding cyclic tensor product. It is connective, since all tensor factors are connective and taking homotopy orbits preserves connectivity. Hence the first assertion follows by induction on mm; compare [9 Lemma 3.14 and the proof of Proposition 5.2].

The second assertion is [4 Theorem 2.5], and the compatibility with the operators follows from [4 Proposition 2.13] together with the compatibility with restriction in its construction. Our index mm is the number of elements of the truncation set; thus WmW_m in this paper is denoted by W⟨pm−1⟩W_{\langle p^{m-1}\rangle} in [4]. □

The invariants and their Morita invariance

4.1. Definition. Let T\mathcal{T} be a dg-category and PP an exact endofunctor of T\mathcal{T}. We define Qne(T;P)Q_n^e(\mathcal{T};P) by (1), and we denote by

ηne(T;P):THH⁡(T;P)⟶Qne(T;P)\eta_n^e(\mathcal{T};P): \operatorname{THH}(\mathcal{T};P) \longrightarrow Q_n^e(\mathcal{T};P)

the composite of the inclusion of the second summand and the canonical map to the cofiber. Thus we have a pushout square of spectra

TR⁡n+e(T;P)→FeTR⁡n(T;P⊗Pe)res⁡n+e−1↓↓THH⁡(T;P)→ηne(T;P)Qne(T;P).\begin{CD} \operatorname{TR}^{n+e}(\mathcal{T};P) @>{F^e}>> \operatorname{TR}^{n}(\mathcal{T};P^{\otimes P^e}) \\ @V{\operatorname{res}^{n+e-1}}VV @VVV \\ \operatorname{THH}(\mathcal{T};P) @>{\eta_n^e(\mathcal{T};P)}>> Q_n^e(\mathcal{T};P). \end{CD}

Definition 4.1. Let dd be an integer and e≥1e \ge1. We define

ht⁡de(T;P):=inf⁡{n≥1∣π−d(ηne(T;P)) is injective}.\operatorname{ht}_d^e(\mathcal{T};P) := \inf\{n \ge1 \mid\pi_{-d}(\eta_n^e(\mathcal{T};P))\text{ is injective}\}.

If T\mathcal{T} is proper and admits a Serre functor STS_{\mathcal{T}}, we put

PT:=ST[−d],ht⁡de(T):=ht⁡de(T;PT).P_{\mathcal{T}} := S_{\mathcal{T}}[-d], \qquad\operatorname{ht}_d^e(\mathcal{T}) := \operatorname{ht}_d^e(\mathcal{T};P_{\mathcal{T}}).

Lemma 4.2. There are maps

Qn+1e(T;P)⟶Qne(T;P)Q_{n+1}^e(\mathcal{T};P) \longrightarrow Q_n^e(\mathcal{T};P)

induced by the restriction maps, whose composite with ηn+1e(T;P)\eta_{n+1}^e(\mathcal{T};P) is ηne(T;P)\eta_n^e(\mathcal{T};P). In particular, if π−d(ηne(T;P))\pi_{-d}(\eta_n^e(\mathcal{T};P)) is injective, then π−d(ηn′e(T;P))\pi_{-d}(\eta_{n'}^e(\mathcal{T};P)) is injective for every n′≥nn' \ge n.

Proof. We apply the restriction maps to the two TR terms and the identity to the THH term. Since Fe∘res⁡=res⁡∘FeF^e \circ\operatorname{res} = \operatorname{res} \circ F^e and the two composites of restriction maps to THH agree, we obtain a map of pushout diagrams, and hence the required map. If π−d(ηne)\pi_{-d}(\eta_n^e) is injective, then any element killed by π−d(ηn+1e)\pi_{-d}(\eta_{n+1}^e) is also killed by the composite π−d(ηne)\pi_{-d}(\eta_n^e), and hence it is zero. □

Proposition 4.3. A derived Morita equivalence T≃U\mathcal{T} \simeq\mathcal{U} which carries PP to P′P' induces an equivalence of arrows

ηne(T;P)≃ηne(U;P′).\eta_n^e(\mathcal{T};P) \simeq\eta_n^e(\mathcal{U};P').

In particular, we have ht⁡de(T;P)=ht⁡de(U;P′)\operatorname{ht}_d^e(\mathcal{T};P) = \operatorname{ht}_d^e(\mathcal{U};P') for all ee and $d.

Proof. By Corollary 3.4, both arrows in the diagram defining QneQ_n^e are identified. Since pushouts are functorial, the pushouts and the maps from the THH terms are also identified. Hence the two maps on π−d\pi_{-d} are conjugate by isomorphisms, and one of them is injective if and only if the other is. □

4.2. Proof of Theorem A. Let E:T≃UE:\mathcal{T}\simeq\mathcal{U} be the given equivalence, and choose dg-algebras AA and BB with T≃Perf⁡(A)\mathcal{T}\simeq\operatorname{Perf}(A) and U≃Perf⁡(B)\mathcal{U}\simeq\operatorname{Perf}(B). Then EE is given by an invertible bimodule, and it carries the AA-bimodule A∨A^\vee to the BB-bimodule B∨B^\vee, since both are the kk-linear duals of the corresponding mapping bimodules. Hence E∘ST≃SU∘EE\circ S_{\mathcal{T}}\simeq S_{\mathcal{U}}\circ E and E∘PT≃PU∘EE\circ P_{\mathcal{T}}\simeq P_{\mathcal{U}}\circ E, and the first assertion follows from Proposition 4.3. The last assertion is Lemma 4.2. □

Remark 4.4. The definition of ht⁡de\operatorname{ht}_d^e does not use a distinguished unit element in HH⁡0\operatorname{HH}_0 or in TR. Such an element is not invariant under Morita equivalence: under the Morita equivalence HH⁡0(Mr(k))≃HH⁡0(k)=k\operatorname{HH}_0(M_r(k))\simeq\operatorname{HH}_0(k)=k, the class of the identity matrix corresponds to rr.

Witt vectors with coefficients in a line bundle

Let XX be a Noetherian Fp\mathbb{F}_p-scheme and LL an invertible OX\mathcal{O}_X-module. We denote by

Wm(OX;L)W_m(\mathcal{O}_X;L)

the sheaf of Witt vectors of length mm with coefficients in LL, regarded as a symmetric bimodule, which is obtained by gluing the local constructions of [4].

Lemma 5.1. The sheaf Wm(OX;L)W_m(\mathcal{O}_X;L) is an invertible WmOXW_m\mathcal{O}_X-module. More precisely, let ℓ\ell and ℓ′=gℓ\ell'=g\ell be local generators of LL, where gg is a unit. Then the transition map from the trivialization given by ℓ′\ell' to the trivialization given by ℓ\ell is

(a0,…,am−1)⟼(ga0,gpa1,…,gpm−1am−1),(a_0,\ldots,a_{m-1})\longmapsto(ga_0,g^pa_1,\ldots,g^{p^{m-1}}a_{m-1}),

that is, the multiplication by [g][g].

Proof. If LL is trivialized by ℓ\ell, then the tensor powers appearing in the construction of Witt vectors with coefficients are identified with the corresponding powers of ℓ\ell, and hence Wm(OX;L)W_m(\mathcal{O}_X;L) is identified with WmOXW_m\mathcal{O}_X. The multiplication by gg on LL acts on the ii-th coordinate, which lies in L⊗piL^{\otimes p^i}, by gpig^{p^i}; this is the multiplication by the Teichmüller lift [g][g]. Since [gh]=[g][h][gh]=[g][h], these transition maps satisfy the cocycle condition, and the local modules glue. □

Remark 5.2. By Lemma 5.1, Wm(OX;L)W_m(\mathcal{O}_X;L) is the Teichmüller lift of LL. In particular, if XX is normal and L=OX(D)L=\mathcal{O}_X(D) for a Cartier divisor DD, then Wm(OX;L)W_m(\mathcal{O}_X;L) is isomorphic to the Witt divisorial sheaf WmOX(D)W_m\mathcal{O}_X(D) of [11 Subsection 2.6], which is locally generated by [f]−1[f]^{-1} if D=div⁡(f)D=\operatorname{div}(f) locally. For L=ωXL=\omega_X, this sheaf should be distinguished from the dualizing sheaf WmωXW_m\omega_X of WmOXW_m\mathcal{O}_X.

We have the maps

res⁡n ⁣:Wn+1(OX;L)⟶L,F ⁣:Wn+1(OX;L)⟶Wn(OX;L⊗p).\operatorname{res}^n\colon W_{n+1}(\mathcal{O}_X;L)\longrightarrow L,\qquad F\colon W_{n+1}(\mathcal{O}_X;L)\longrightarrow W_n(\mathcal{O}_X;L^{\otimes p}).

Proposition 5.3. The pushout of the above two maps in the category of sheaves of abelian groups is naturally isomorphic to

QX,n⊗OXL,Q_{X,n}\otimes_{\mathcal{O}_X}L,

where QX,nQ_{X,n} is equipped with the OX\mathcal{O}_X-module structure of Lemma 2.2. Under this isomorphism, the induced map from LL is $\Phi_{X,n}\otimes L.

Proof. The map res⁡n\operatorname{res}^n is surjective with kernel

VWn(OX;L⊗p).V W_n(\mathcal{O}_X;L^{\otimes p}).

For general coefficients, the composite FVFV is the norm map of the cyclic action on tensor powers. Since this action is trivial on tensor powers of an invertible module regarded as a symmetric bimodule, we have FV=pFV=p on Wn(OX;L⊗p)W_n(\mathcal{O}_X;L^{\otimes p}). Hence the pushout is the sheaf of abelian groups

Wn(OX;L⊗p)/p.W_n(\mathcal{O}_X;L^{\otimes p})/p.

We equip this quotient with the OX\mathcal{O}_X-module structure given by a⋅w=[a]pwa\cdot w=[a]^{p}w. If ℓ\ell is a local generator of LL, then this quotient is identified with QX,nQ_{X,n} by Lemma 2.2. If ℓ′=gℓ\ell'=g\ell is another local generator, then the transition map is the multiplication by [g]p[g]^p by Lemma 5.1, which is exactly the transition map of QX,n⊗LQ_{X,n}\otimes L with respect to the OX\mathcal{O}_X-module structure of Lemma 2.2.

Moreover, with respect to the generator ℓ\ell, an element aℓa\ell of LL is sent to the class of [a]pℓ⊗p[a]^p\ell^{\otimes p}, which corresponds to ΦX,n(a)⊗ℓ\Phi_{X,n}(a)\otimes\ell. Hence these identifications glue, and the induced map is ΦX,n⊗L\Phi_{X,n}\otimes L. □

Corollary 5.4. Let U=Spec⁡AU=\operatorname{Spec} A be an affine open subscheme of XX, and regard L(U)L(U) as a symmetric AA-bimodule. Then the spectra THH⁡(A;L(U))\operatorname{THH}(A;L(U)) and Qn1(A;L(U))Q^1_n(A;L(U)) are connective, and there are natural isomorphisms

π0THH⁡(A;L(U))≃L(U),\pi_0\operatorname{THH}(A;L(U))\simeq L(U),
π0Qn1(A;L(U))≃(QX,n⊗L)(U),\pi_0Q^1_n(A;L(U))\simeq(Q_{X,n}\otimes L)(U),

compatible with the restriction to smaller affine open subschemes. Under these isomorphisms, the map π0(ηn1)\pi_0(\eta^1_n) corresponds to ΦX,n⊗L\Phi_{X,n}\otimes L.

Proof. By Proposition 3.5, all the terms in the diagram defining Qn1(A;L(U))Q^1_n(A;L(U)) are connective, and their zeroth homotopy groups are given by Witt vectors with coefficients. Since a cofiber of a map of connective spectra is connective and its π0\pi_0 is the cokernel of the induced map on π0\pi_0, the assertion follows from Proposition 5.3. □

Zariski descent and the lowest homotopy group

Zariski descent

Let XX be a quasi-compact separated scheme and LL a line bundle on XX. For an open subscheme U⊂XU \subset X, we denote by PUP_U the endofunctor −⊗L∣U-\otimes L|_U of Perf⁡(U)\operatorname{Perf}(U).

Proposition 6.1. The polygonic spectra X(Perf⁡(−),P(−))\mathcal{X}(\operatorname{Perf}(-),P_{(-)}), the spectra TR⁡m(Perf⁡(−);P(−))\operatorname{TR}^m(\operatorname{Perf}(-);P_{(-)}) and the spectra Qne(Perf⁡(−);P(−))Q^e_n(\operatorname{Perf}(-);P_{(-)}) satisfy Zariski descent; that is, they send a cover of a quasi-compact open subscheme by two quasi-compact open subschemes to a Mayer–Vietoris square. Moreover, if XX is Noetherian of finite Krull dimension, then these spectra for XX are identified with the derived global sections of the associated Zariski sheaves of spectra.

Proof. For quasi-compact open subschemes V⊂UV \subset U, the restriction functor Perf⁡(U)→Perf⁡(V)\operatorname{Perf}(U) \to\operatorname{Perf}(V) commutes with −⊗L-\otimes L, and hence it underlies a laced functor (Perf⁡(U),PU)→(Perf⁡(V),PV)(\operatorname{Perf}(U),P_U) \to(\operatorname{Perf}(V),P_V). By Theorem 3.2, we obtain a functor from quasi-compact open subschemes of XX to polygonic spectra. Since the Tate construction is exact, finite limits of polygonic spectra are computed levelwise, and hence it suffices to show that THH⁡(Perf⁡(−);Pr)\operatorname{THH}(\operatorname{Perf}(-);P^r) satisfies Mayer–Vietoris for every r≥1r \ge1.

Let j:U↪Xj: U \hookrightarrow X be a quasi-compact open immersion, and put Z=X∖UZ = X \setminus U. For every r≥1r \ge1, the endofunctor PXr=−⊗L⊗rP^r_X = -\otimes L^{\otimes r} preserves Perf⁡Z(X)\operatorname{Perf}_Z(X), and the endofunctor induced on Perf⁡(U)\operatorname{Perf}(U) is PUrP^r_U; on Ind-categories, this follows from

j∗(j∗E⊗L⊗r)≃E⊗j∗L⊗r.j^*(j_*E \otimes L^{\otimes r}) \simeq E \otimes j^*L^{\otimes r}.

Thus the localization sequence Perf⁡Z(X)→Perf⁡(X)→Perf⁡(U)\operatorname{Perf}_Z(X) \to\operatorname{Perf}(X) \to\operatorname{Perf}(U) gives, for each rr, a laced Verdier sequence up to idempotent completion. By [7 Proposition 5.18 and Corollary 5.19], THH⁡\operatorname{THH} with these coefficients sends it to a fiber sequence. If X=U∪VX = U \cup V, then the dg-categories of perfect complexes supported on X∖UX \setminus U in XX and in VV are equivalent, compatibly with the coefficients, by excision. Comparing the localization sequences, we obtain a Mayer–Vietoris square for THH⁡(Perf⁡(−);Pr)\operatorname{THH}(\operatorname{Perf}(-);P^r).

Since finite TR⁡\operatorname{TR} preserves finite limits of polygonic spectra, and since the pushout defining QneQ^e_n is exact and its Frobenius term is natural by Theorem 3.2(iii), the spectra TR⁡m\operatorname{TR}^m and QneQ^e_n also satisfy Mayer–Vietoris. Finally, if XX is Noetherian of finite Krull dimension, a finite affine open cover and its intersections compute the associated sheaf by iterated Mayer–Vietoris squares, and the convergence of the descent spectral sequence (Lemma 6.2) identifies the constructions for XX with the derived global sections. □

The lowest homotopy group

Lemma 6.2. Let XX be a Noetherian scheme of dimension dd, and let E\mathcal{E} be a sheaf of spectra on the Zariski site of XX whose homotopy sheaves πtE\pi_t\mathcal{E} vanish for t<0t < 0. Then we have

πiR⁡Γ(X,E)=0(i<−d),π−dR⁡Γ(X,E)≃Hd(X,π0E),\pi_i\operatorname{R}\Gamma(X,\mathcal{E}) = 0 \quad(i < -d), \qquad\pi_{-d}\operatorname{R}\Gamma(X,\mathcal{E}) \simeq H^d(X,\pi_0\mathcal{E}),

and the isomorphism is natural in E\mathcal{E}.

Proof. Since the underlying topological space of XX is Noetherian of Krull dimension dd, the ∞\infty-topos of sheaves on XX is locally of homotopy dimension at most dd by [16 Proposition 7.2.4.7 and Corollary 7.2.4.17], and hence it is hypercomplete by [16 Corollary 7.2.1.12]. Moreover, the Zariski cohomological dimension of XX is at most dd by [19 Tag 02UZ]. Hence the descent spectral sequence

E2s,t=Hs(X,πtE)⟹πt−sR⁡Γ(X,E)E_2^{s,t} = H^s(X,\pi_t\mathcal{E}) \Longrightarrow\pi_{t-s}\operatorname{R}\Gamma(X,\mathcal{E})

converges. Since E2s,t=0E_2^{s,t}=0 unless 0≤s≤d0 \le s \le d and t≥0t \ge0, there are no nonzero terms of total degree t−s<−dt-s < -d, and the only term of total degree −d-d is E2d,0E_2^{d,0}. No nonzero differential enters this term, since its source would have t<0t < 0, and no nonzero differential leaves it, since its target would have s>ds > d. Hence we obtain the claim. □

Lemma 6.3. Let RR be a Noetherian ring, let t1,…,td∈Rt_1,\ldots,t_d \in R, and put I=(t1,…,td)I=(t_1,\ldots,t_d) and tJ=∏i∈Jtit_J=\prod_{i\in J}t_i for a subset J⊂{1,…,d}J \subset\{1,\ldots,d\}. Let J↦EJJ \mapsto E_J be a cubical diagram of connective spectra indexed by the subsets of {1,…,d}\{1,\ldots,d\}, and let EtotE^{\mathrm{tot}} be its total fiber. Then πiEtot=0\pi_iE^{\mathrm{tot}}=0 for i<−di<-d. Moreover, if the diagram J↦π0EJJ \mapsto\pi_0E_J is isomorphic to the diagram J↦MtJJ \mapsto M_{t_J} of localizations of an RR-module MM, then there is a natural isomorphism

π−dEtot≃HId(M).\pi_{-d}E^{\mathrm{tot}} \simeq H_I^d(M).

Proof. Filtering the total fiber by the cardinality of JJ, we obtain a spectral sequence

E1s,t=⨁∣J∣=sπtEJ⟹πt−sEtot.E_1^{s,t}=\bigoplus_{|J|=s}\pi_tE_J \Longrightarrow\pi_{t-s}E^{\mathrm{tot}}.

where t≥0t \ge0 and 0≤s≤d0 \le s \le d. Hence πiEtot=0\pi_iE^{\mathrm{tot}}=0 for i<−di<-d, and π−dEtot\pi_{-d}E^{\mathrm{tot}} is the dd-th cohomology of the complex E∙,0E^{\bullet,0}. Under the assumption, this complex is the Čech complex of MM with respect to t1,…,tdt_1,\ldots,t_d, whose dd-th cohomology is HId(M)H_I^d(M). □

Proper Gorenstein schemes

Let XX be a nonempty proper Gorenstein scheme over kk of pure dimension dd. Then ωX\omega_X is invertible, and ωX[d]\omega_X[d] is a dualizing complex of XX. Since XX is proper, Perf⁡(X)\operatorname{Perf}(X) is proper. By [19 Tag 09IS], there is a perfect complex which generates DQcoh(OX)D_{\mathrm{Qcoh}}(\mathcal{O}_X); since the compact objects of DQcoh(OX)D_{\mathrm{Qcoh}}(\mathcal{O}_X) are the perfect complexes [19 Tag 09M1], it is a classical generator of Perf⁡(X)\operatorname{Perf}(X). The Serre functor of Perf⁡(X)\operatorname{Perf}(X) is

SX=−⊗ωX[d];S_X=-\otimes\omega_X[d];

see [19 Tag 0FY8]. Hence the coefficient in Definition 4.1 is PX=−⊗ωXP_X=-\otimes\omega_X.

Proof of Theorem B

Proposition 7.1. The map π−d(ηn1(Perf⁡(X);PX))\pi_{-d}(\eta_n^1(\operatorname{Perf}(X);P_X)) is naturally identified with the map

Hd(X,ωX)→Hd(ΦX,n⊗ωX)Hd(X,QX,n⊗ωX).(4)H^d(X,\omega_X)\xrightarrow{H^d(\Phi_{X,n}\otimes\omega_X)}H^d(X,Q_{X,n}\otimes\omega_X). \tag*{(4)}

Proof. By Proposition 6.1, the source and the target of ηn1(Perf⁡(X);PX)\eta_n^1(\operatorname{Perf}(X);P_X), together with the map between them, are identified with the derived global sections of the corresponding Zariski sheaves of spectra. By Theorem 3.2(ii) on affine open subschemes and Corollary 5.4, the homotopy sheaves of these sheaves vanish in negative degrees, and their zeroth homotopy sheaves and the map between them are identified with

ωX→ΦX,n⊗ωXQX,n⊗ωX.\omega_X\xrightarrow{\Phi_{X,n}\otimes\omega_X}Q_{X,n}\otimes\omega_X.

Hence the assertion follows from Lemma 6.2 and its naturality. □

Lemma 7.2. Let QQ be a coherent OX\mathcal{O}_X-module and Φ:OX→Q\Phi:\mathcal{O}_X\to Q an OX\mathcal{O}_X-linear map. Then the induced map

Hd(X,ωX)⟶Hd(X,Q⊗ωX)H^d(X,\omega_X)\longrightarrow H^d(X,Q\otimes\omega_X)

is injective if and only if Φ\Phi admits an OX\mathcal{O}_X-linear retraction.

Proof. Since XX is an equidimensional proper Cohen–Macaulay scheme, Serre duality gives a natural isomorphism

Hd(X,F)∨≃Hom⁡X(F,ωX)H^d(X,\mathcal{F})^\vee\simeq\operatorname{Hom}_X(\mathcal{F},\omega_X)

for every coherent OX\mathcal{O}_X-module F\mathcal{F}; see [19 Tag 0F VZ]. Applying it to F=Q⊗ωX\mathcal{F}=Q\otimes\omega_X and F=ωX\mathcal{F}=\omega_X, and using the invertibility of ωX\omega_X, we identify the dual of the above map with

Φ∗:Hom⁡X(Q,OX)⟶Hom⁡X(OX,OX).\Phi^*:\operatorname{Hom}_X(Q,\mathcal{O}_X)\longrightarrow\operatorname{Hom}_X(\mathcal{O}_X,\mathcal{O}_X).

Since all the groups are finite-dimensional over kk, the above map is injective if and only if Φ∗\Phi^* is surjective. If Φ∗\Phi^* is surjective, then a preimage of id⁡OX\operatorname{id}_{O_X} is a retraction of Φ\Phi. Conversely, if ψ\psi is a retraction of Φ\Phi, then Φ∗(u∘ψ)=u\Phi^*(u \circ\psi)=u for every endomorphism uu of OXO_X. □

Proof of the first assertion of Theorem B. Since XX is of finite type over a perfect field, the Frobenius of XX is finite, and hence QX,nQ_{X,n} is a coherent OXO_X-module. By Proposition 7.1 and Lemma 7.2, the map π−d(η11(Perf⁡(X);PX))\pi_{-d}(\eta^1_1(\operatorname{Perf}(X);P_X)) is injective if and only if ΦX,n\Phi_{X,n} admits an OXO_X-linear retraction, that is, XX is nn-quasi-FF-split. Since this holds for every nn, we obtain ht⁡d1(Perf⁡(X))=ht⁡(X)\operatorname{ht}_d^1(\operatorname{Perf}(X))=\operatorname{ht}(X). □

Dimension and derived invariance

Let T\mathcal{T} be a nonzero proper dg-category with a Serre functor SS and a classical generator GG. Following Elagin–Lunts [5 Definition 5.3], we put

e−(G,G′):=inf⁡{i∣Hom⁡T(G,G′[i])≠0},e+(G,G′):=sup⁡{i∣Hom⁡T(G,G′[i])≠0}.e_-(G,G'):=\inf\{i\mid\operatorname{Hom}_{\mathcal{T}}(G,G'[i])\ne0\},\qquad e_+(G,G'):=\sup\{i\mid\operatorname{Hom}_{\mathcal{T}}(G,G'[i])\ne0\}.

and define the upper and lower Serre dimensions by

Sdim⁡‾(T):=lim sup⁡m→∞−e−(G,SmG)m,Sdim⁡‾(T):=lim inf⁡m→∞−e+(G,SmG)m.\overline{\operatorname{Sdim}}(\mathcal{T}):=\limsup_{m\to\infty}\frac{-e_-(G,S^mG)}{m},\qquad\underline{\operatorname{Sdim}}(\mathcal{T}):=\liminf_{m\to\infty}\frac{-e_+(G,S^mG)}{m}.

They do not depend on the choice of GG [5], and they are invariant under kk-linear equivalences, which commute with Serre functors.

Proposition 7.3. Let XX be as above. Then Sdim⁡‾(Perf⁡(X))=Sdim⁡‾(Perf⁡(X))=d\overline{\operatorname{Sdim}}(\operatorname{Perf}(X))=\underline{\operatorname{Sdim}}(\operatorname{Perf}(X))=d. In particular, dd is invariant under kk-linear equivalences of these dg-categories.

Proof. For smooth projective XX, this is [5 Lemma 5.6]; the same argument applies. Let GG be a classical generator of Perf⁡(X)\operatorname{Perf}(X). Since G∨⊗GG^\vee\otimes G is perfect, its cohomology sheaves are concentrated in a bounded range of degrees, and this range does not change after tensoring with ωX⊗m\omega_X^{\otimes m}. Since coherent cohomology on XX vanishes in degrees greater than dd, there are integers a≤ba\leq b, independent of mm, such that

Hj(X,G∨⊗G⊗ωX⊗m)=0unless a≤j≤b.H^j(X,G^\vee\otimes G\otimes\omega_X^{\otimes m})=0\quad\text{unless }a\leq j\leq b.

Since

Hom⁡X(G,SXmG[i])≅Hmd+i(X,G∨⊗G⊗ωX⊗m),\operatorname{Hom}_X(G,S_X^mG[i])\cong H^{md+i}(X,G^\vee\otimes G\otimes\omega_X^{\otimes m}),

this group vanishes unless a−md≤i≤b−mda-md\leq i\leq b-md. On the other hand, at least one of these groups is nonzero, since otherwise we would have SXmG=0S_X^mG=0. Hence a−md≤e−(G,SXmG)≤e+(G,SXmG)≤b−mda-md\leq e_-(G,S_X^mG)\leq e_+(G,S_X^mG)\leq b-md, and dividing by mm, we obtain the claim. □

Corollary 7.4. Let XX and YY be nonempty proper Gorenstein schemes over kk of pure dimension. If there is a kk-linear derived Morita equivalence Perf⁡(X)≃Perf⁡(Y)\operatorname{Perf}(X)\simeq\operatorname{Perf}(Y), then

dim⁡X=dim⁡Y,ht⁡(X)=ht⁡(Y).\dim X=\dim Y,\qquad\operatorname{ht}(X)=\operatorname{ht}(Y).

In particular, quasi-FF-splitting and FF-splitting are preserved by such equivalences.

Proof. The assertion follows from Proposition 7.3, Theorem A and the first assertion of Theorem B, since XX is FF-split if and only if ht⁡(X)=1\operatorname{ht}(X)=1. □

The role of the Serre functor

Remark 7.5. Assume that kk is algebraically closed, and let XX be an ordinary smooth projective curve over kk of genus at least two. Since the Kodaira dimension of XX is one, XX is not quasi-FF-split by [11 Proposition 3.14], and hence ht⁡11(Perf⁡(X))=∞\operatorname{ht}_1^1(\operatorname{Perf}(X))=\infty by Theorem B. On the other hand, the Frobenius acts bijectively on H1(X,OX)H^1(X,O_X). The argument of Proposition 7.1 with L=OXL=O_X identifies π−1(η11(Perf⁡(X);id⁡))\pi_{-1}(\eta^1_1(\operatorname{Perf}(X);\operatorname{id})) with the Frobenius on H1(X,OX)H^1(X,O_X), which is injective. Hence the identity coefficient does not recover the quasi-FF-split height, and the shift of the Serre functor is necessary in Theorem B.

Complete local Gorenstein rings

Throughout this section, let (R,m,k)(R,\mathfrak{m},k) be a nonzero complete Noetherian Gorenstein local kk-algebra with residue field kk, and put d=dim⁡Rd=\dim R. As mentioned in the introduction, RR is FF-finite.

The dg-category of supported perfect complexes.

Proposition 8.1. The dg-category TR=Perf⁡m(R)\mathcal{T}_R=\operatorname{Perf}_{\mathfrak{m}}(R) is proper, has a classical generator, and admits the Serre functor [d][d].

Proof. For K,L∈TRK,L\in\mathcal{T}_R, the complex RHom⁡R(K,L)\operatorname{RHom}_R(K,L) is a perfect RR-complex with finite length cohomology, and hence TR\mathcal{T}_R is proper. A Koszul complex K0K_0 on a system of parameters of RR is a classical generator of TR\mathcal{T}_R; see [19 Tag 0A9A, Tag 0A9B]. We put A:=RHom⁡R(K0,K0)A:=\operatorname{RHom}_R(K_0,K_0), so that TR≃Perf⁡(A)\mathcal{T}_R\simeq\operatorname{Perf}(A).

Since RR is Gorenstein, R[d]R[d] is a normalized dualizing complex of RR. For an RR-module MM of finite length, we have Ext⁡Ri(M,R)=0\operatorname{Ext}^i_R(M,R)=0 for i≠di\ne d and Ext⁡Rd(M,R)≃Hom⁡R(M,ER(k))\operatorname{Ext}^d_R(M,R)\simeq\operatorname{Hom}_R(M,E_R(k)) by local duality [19 Tag 0A81]. Since k⊂Rk\subset R is a coefficient field, we have ER(k)≃colim⁡qHom⁡k(R/mq,k)E_R(k)\simeq\operatorname*{colim}_q\operatorname{Hom}_k(R/\mathfrak{m}^q,k), and the evaluation at 11 identifies Hom⁡R(M,ER(k))\operatorname{Hom}_R(M,E_R(k)) with Hom⁡k(M,k)\operatorname{Hom}_k(M,k); see [19 Tag 0A7Q]. Hence we obtain a natural equivalence

RHom⁡k(C,k)≃RHom⁡R(C,R[d])\operatorname{RHom}_k(C,k)\simeq\operatorname{RHom}_R(C,R[d])

of functors on RR-complexes CC with finite length cohomology. Since the AA-bimodule structure on AA is RR-linear, applying this to C=AC=A gives an equivalence of AA-bimodules A∨≃RHom⁡R(A,R)[d]A^\vee\simeq\operatorname{RHom}_R(A,R)[d]. Since K0K_0 is perfect, the trace pairing A⊗RA→RA\otimes_R A\to R, (u,v)↦tr⁡(uv)(u,v)\mapsto\operatorname{tr}(uv), induces an equivalence of AA-bimodules A≃RHom⁡R(A,R)A\simeq\operatorname{RHom}_R(A,R). Hence A∨≃A[d]A^\vee\simeq A[d] as AA-bimodules, that is, the Serre functor of TR\mathcal{T}_R is [d][d].

Consequently, we have PTR=STR[−d]≃id⁡P_{\mathcal{T}_R}=S_{\mathcal{T}_R}[-d]\simeq\operatorname{id}.

The zeroth homotopy groups.

For a discrete commutative Fp\mathbb{F}_p-algebra AA, we put Qne(A):=Qne(Perf⁡(A);id⁡)\mathcal{Q}^e_n(A):=\mathcal{Q}^e_n(\operatorname{Perf}(A);\operatorname{id}).

Proposition 8.2. The spectrum Qne(A)\mathcal{Q}^e_n(A) is connective, and there is a natural isomorphism

π0Qne(A)≃QA,ne\pi_0\mathcal{Q}^e_n(A)\simeq Q^e_{A,n}

under which the map π0(ηne)\pi_0(\eta^e_n) corresponds to ΦA,ne\Phi^e_{A,n}.

Proof. By Proposition 3.5, the three spectra TR⁡∙n+e(A)\operatorname{TR}^{n+e}_\bullet(A), TR⁡∙n(A)\operatorname{TR}^n_\bullet(A) and THH⁡(A)\operatorname{THH}(A) in the diagram defining Qne(A)\mathcal{Q}^e_n(A) are connective, and their zeroth homotopy groups are Wn+e(A)W_{n+e}(A), Wn(A)W_n(A) and AA, respectively.

Since AA has characteristic pp, the Frobenius Fe ⁣:Wn+e(A)→Wn(A)F^e\colon W_{n+e}(A)\to W_n(A) is given by

(a0,…,an+e−1)⟼(a0pe,…,an−1pe).(a_0,\ldots,a_{n+e-1})\longmapsto(a_0^{p^e},\ldots,a_{n-1}^{p^e}).

Hence it factors as the surjection res⁡e ⁣:Wn+e(A)→Wn(A)\operatorname{res}^e\colon W_{n+e}(A)\to W_n(A) followed by Wn(FA)eW_n(F_A)^e. The other map res⁡n+e−1\operatorname{res}^{n+e-1} factors as res⁡n−1∘res⁡e\operatorname{res}^{n-1}\circ\operatorname{res}^e. In particular, both maps vanish on ker⁡(res⁡e)\ker(\operatorname{res}^e).

Since all the spectra are connective, π0\pi_0 of the pushout is the pushout of abelian groups of the zeroth homotopy groups. Dividing the source by the common kernel ker⁡(res⁡e)\ker(\operatorname{res}^e), we identify it with the underlying abelian group of the pushout of Wn(A)W_n(A)-modules in Definition 2.1, compatibly with the maps from AA. This also equips π0Qne(A)\pi_0\mathcal{Q}^e_n(A) with the Wn(A)W_n(A)-module structure of QA,neQ^e_{A,n}.

Local cohomology.

Proposition 8.3. The map π−d(ηne(TR;id⁡))\pi_{-d}(\eta^e_n(\mathcal{T}_R;\operatorname{id})) is naturally identified with the map

Hmd(R)→Hmd(ΦR,ne)Hmd(QR,ne).(5)H^d_{\mathfrak{m}}(R)\xrightarrow{H^d_{\mathfrak{m}}(\Phi^e_{R,n})}H^d_{\mathfrak{m}}(Q^e_{R,n}). \tag*{(5)}

Proof. Since the coefficient is the identity, the polygonic spectra here come from cyclotomic spectra, and finite TR agrees with the classical TR [9]. Put U=Spec⁡R∖{m}U = \operatorname{Spec} R \setminus\{\mathfrak{m}\}. By the localization theorem [2 Theorems 1.1 and 1.3], THH⁡(TR)\operatorname{THH}(\mathcal{T}_R) is identified with the fiber of THH⁡(R)→THH⁡(U)\operatorname{THH}(R) \to\operatorname{THH}(U), compatibly with the cyclotomic structures. Since finite TR and the cofiber defining QneQ_n^e are exact, Qne(TR)Q_n^e(\mathcal{T}_R) is also identified with the fiber of Qne(R)→Qne(U)Q_n^e(R) \to Q_n^e(U).

Choose a system of parameters t1,…,tdt_1,\ldots,t_d of RR. Then the principal open subschemes D(ti)D(t_i) cover UU. By Zariski descent [2], THH⁡(TR)\operatorname{THH}(\mathcal{T}_R) and Qne(TR)Q_n^e(\mathcal{T}_R) are identified with the total fibers of the cubes J↦THH⁡(RtJ)J \mapsto\operatorname{THH}(R_{t_J}) and J↦Qne(RtJ)J \mapsto Q_n^e(R_{t_J}), which consist of connective spectra. For THH, the diagram of zeroth homotopy groups is the diagram J↦RtJJ \mapsto R_{t_J} of localizations of RR.

For QneQ_n^e, by Proposition 8.2, the diagram of zeroth homotopy groups is J↦QRtJ,neJ \mapsto Q_{R_{t_J},n}^e, which is the diagram of localizations of the Wn(R)W_n(R)-module QR,neQ_{R,n}^e at the Teichmüller lifts [tJ][t_J]. Indeed, we have Wn(R)[t]≃Wn(Rt)W_n(R)[t] \simeq W_n(R_t), and localizing F∗eWn(R)F_*^eW_n(R) at [t][t] is the same as localizing it at [t]pe[t]^{p^e}. Since the kernel of Wn(R)→RW_n(R) \to R is nilpotent, the ideal ([t1],…,[td])([t_1],\ldots,[t_d]) of Wn(R)W_n(R) has the maximal ideal as its radical.

Applying Lemma 6.3 to the Noetherian ring Wn(R)W_n(R) (Lemma 8.4) and the elements [t1],…,[td][t_1],\ldots,[t_d], we obtain

π−dTHH⁡(TR)≃Hmd(R),π−dQne(TR)≃Hmd(QR,ne).\pi_{-d}\operatorname{THH}(\mathcal{T}_R) \simeq H^d_{\mathfrak{m}}(R), \qquad\pi_{-d}Q_n^e(\mathcal{T}_R) \simeq H^d_{\mathfrak{m}}(Q_{R,n}^e).

where both local cohomology groups are taken over Wn(R)W_n(R) with respect to its maximal ideal; the first one agrees with the usual local cohomology of RR. By naturality, we obtain the claim. □

Duality over Wn(R)W_n(R) and the proof of Theorem C

For the convenience of the reader, we include a proof of the following ring-theoretic facts.

Lemma 8.4. The ring Wn(R)W_n(R) is a complete Noetherian Cohen–Macaulay local ring of dimension dd. Moreover, QR,neQ_{R,n}^e is a finite Wn(R)W_n(R)-module.

Proof. Put W0(R)=0W_0(R) = 0. The Verschiebung filtration of Wn(R)W_n(R) has graded pieces

ViWn−i(R)/Vi+1Wn−i−1(R)≃F∗iR(0≤i≤n−1).V^iW_{n-i}(R)/V^{i+1}W_{n-i-1}(R) \simeq F_*^iR \qquad(0 \leq i \leq n-1).

On the ii-th graded piece, [a][a] acts as the multiplication by apia^{p^i}; in particular, the action of Wn(R)W_n(R) on each graded piece factors through the surjection Wn(R)→RW_n(R) \to R.

Firstly, we prove that Wn(R)W_n(R) is Noetherian by induction on nn, using the exact sequence

0⟶Vn−1R⟶Wn(R)⟶Wn−1(R)⟶0.0 \longrightarrow V^{n-1}R \longrightarrow W_n(R) \longrightarrow W_{n-1}(R) \longrightarrow0.

The kernel is isomorphic to F∗n−1RF_*^{n-1}R, which is a finite RR-module since RR is FF-finite. Hence it is a Noetherian Wn(R)W_n(R)-module. Since Wn−1(R)W_{n-1}(R) is a Noetherian ring by the induction hypothesis, it is also a Noetherian Wn(R)W_n(R)-module, and hence Wn(R)W_n(R) is Noetherian.

Secondly, since the kernel of Wn(R)→RW_n(R) \to R is nilpotent, Wn(R)W_n(R) is local, its spectrum is homeomorphic to Spec⁡R\operatorname{Spec} R, and its dimension is dd. On each graded piece F∗iRF_*^iR, the topology defined by the maximal ideal of Wn(R)W_n(R) coincides with the m\mathfrak{m}-adic topology, since the ideal generated by the pip^i-th powers of the elements of m\mathfrak{m} is m\mathfrak{m}-primary. These finite RR-modules are complete. Since completion is exact for finite modules over a Noetherian ring, the above exact sequence and induction on nn show that Wn(R)W_n(R) is complete.

Thirdly, let t1,…,tdt_1,\ldots,t_d be a system of parameters of RR, which is a regular sequence since RR is Cohen–Macaulay. On the ii-th graded piece, [t1],…,[td][t_1],\ldots,[t_d] act as t1pi,…,tdpit_1^{p^i},\ldots,t_d^{p^i}, which form a regular sequence on F∗iRF_*^iR. Since a sequence which is regular on modules M′M' and M′′M'' is regular on every extension of M′M' by M′′M'', we see by induction on the filtration that [t1],…,[td][t_1],\ldots,[t_d] is a regular sequence on Wn(R)W_n(R). Hence Wn(R)W_n(R) is Cohen–Macaulay.

Finally, F∗eWn(R)F_*^eW_n(R) has a finite filtration whose graded pieces are F∗e+iRF_*^{e+i}R, which are finite Wn(R)W_n(R)-modules. Hence F∗eWn(R)F_*^eW_n(R) is a finite Wn(R)W_n(R)-module. Since RR is also a finite Wn(R)W_n(R)-module, so is the pushout QR,neQ_{R,n}^e. □

Proposition 8.5. The map (5) is injective if and only if RR is nn-quasi-FeF^e-split.

:::

Proof. Let ωWn(R)=WnωR\omega_{W_n(R)} = W_n\omega_R be a canonical module of Wn(R)W_n(R), and let

Dn(−):=Hom⁡Wn(R)(−,EWn(R)(k))D_n(-) := \operatorname{Hom}_{W_n(R)}(-, E_{W_n(R)}(k))

be the Matlis duality over Wn(R)W_n(R). By Lemma 8.4 and local duality [19 Tag 0A81], we have a natural isomorphism

DnHmd(M)≃Hom⁡Wn(R)(M,ωWn(R))D_n H^d_{\mathfrak m}(M) \simeq\operatorname{Hom}_{W_n(R)}(M,\omega_{W_n(R)})

for every finite Wn(R)W_n(R)-module MM. The local cohomology modules Hmd(M)H^d_{\mathfrak m}(M) are Artinian, and Matlis duality is exact and faithful on Artinian modules. Hence, applying DnD_n to (5), we see that (5) is injective if and only if the map

Hom⁡Wn(R)(QR,ne,WnωR)⟶Hom⁡Wn(R)(R,WnωR)\operatorname{Hom}_{W_n(R)}(Q^e_{R,n},W_n\omega_R) \longrightarrow\operatorname{Hom}_{W_n(R)}(R,W_n\omega_R)

is surjective. This is the condition in Definition 2.4. □

Proof of Theorem C. By Proposition 8.1, the dg-category TR\mathcal{T}_R is proper and its Serre functor is [d][d], and hence PTR≃id⁡\mathcal{P}_{\mathcal{T}_R} \simeq\operatorname{id}. By Proposition 8.3, the map π−d(ηe(TR;id⁡))\pi_{-d}(\eta^e(\mathcal{T}_R;\operatorname{id})) is identified with the map (5) of local cohomology groups. By Proposition 8.5, this map is injective if and only if RR is nn-quasi-FeF^e-split. Since this holds for every nn, we obtain ht⁡de(Perf⁡m(R))=ht⁡e(R)\operatorname{ht}^e_d(\operatorname{Perf}_{\mathfrak m}(R)) = \operatorname{ht}^e(R). □

For e=1e = 1, Proposition 8.5 and [11 Lemma 3.13] show that ht⁡1(R)\operatorname{ht}^1(R) coincides with the quasi-FF-split height ht⁡(R)\operatorname{ht}(R) of Spec⁡R\operatorname{Spec} R in the sense of Definition 2.3.

Remark 8.6 (A dg-algebra model of a singularity). For a Koszul complex K0K_0 as in the proof of Proposition 8.1, the dg-algebra AR=RHom⁡R(K0,K0)A_R = \operatorname{RHom}_R(K_0,K_0) is proper, and we have Perf⁡(AR)≃Perf⁡m(R)\operatorname{Perf}(A_R) \simeq\operatorname{Perf}_{\mathfrak m}(R). Hence Theorem C expresses the classical heights in terms of a proper dg-algebra model of RR, and the result does not depend on the choice of K0K_0.

Examples

Example 8.7. Let A=k[ϵ]/(ϵ2)A = k[\epsilon]/(\epsilon^2). Then AA is a zero-dimensional complete Gorenstein local ring, and Perf⁡(A)=Perf⁡(ϵ)(A)\operatorname{Perf}(A) = \operatorname{Perf}_{(\epsilon)}(A) is a proper dg-category which is not smooth and whose Serre functor is the identity. For all nn, e≥1e \ge1, we have ΦA,ne(ϵ)=[ϵpe]=0\Phi^e_{A,n}(\epsilon) = [\epsilon^{p^e}] = 0 by the definition of the pushout. Since d=0d = 0, the map (5) is ΦA,ne\Phi^e_{A,n} itself, and hence ht⁡0e(Perf⁡(A))=∞\operatorname{ht}^e_0(\operatorname{Perf}(A)) = \infty for all e≥1e \ge1 by Theorem C. The same argument applies to every zero-dimensional Gorenstein local kk-algebra AA which is not a field: a nonzero element ss of the socle satisfies s2=0s^2 = 0, and hence ΦA,ne(s)=0\Phi^e_{A,n}(s) = 0. Thus finite heights greater than one only appear in positive dimension.

Example 8.8. Assume that kk is algebraically closed of characteristic 2. For n≥2n \ge2, let

R=k[[x,y,z]]/(z2+x2y+xyn),R = k[[x,y,z]]/(z^2+x^2y+xy^n),

which is a rational double point. By [13 Theorem 4.31], the quasi-FF-split height of RR is ⌈log⁡2n⌉\lceil\log_2 n\rceil. Hence, by Theorem C and the comparison for e=1e = 1 above, we have

ht⁡21(Perf⁡m(R))=⌈log⁡2n⌉.\operatorname{ht}^1_2(\operatorname{Perf}_{\mathfrak m}(R)) = \lceil\log_2 n\rceil.

In particular, the invariant ht⁡21\operatorname{ht}^1_2 takes every positive integer as a value on proper dg-categories of the form Perf⁡m(R)\operatorname{Perf}_{\mathfrak m}(R).

References

  1. [1]Benjamin Antieau and Daniel Bragg, Derived invariants from topological Hochschild homology, Algebr. Geom. 9 (2022), 364–399, doi:10.14231/AG-2022-011; arXiv:1906.12267.DOI
  2. [2]Andrew J. Blumberg and Michael A. Mandell, Localization theorems in topological Hochschild homology and topological cyclic homology, Geom. Topol. 16 (2012), 1053–1120, doi:10.2140/gt.2012.16.1053; arXiv:0802.3938.DOI
  3. [4]Emanuele Dotto, Achim Krause, Thomas Nikolaus, and Irakli Patchkoria, Witt vectors with coefficients and TR, Proc. Lond. Math. Soc. (3) 130 (2025), Paper No. e70047, doi:10.1112/plms.70047; arXiv:2312.12971.DOI
  4. [5]Alexey Elagin and Valery A. Lunts, Three notions of dimension for triangulated categories, J. Algebra 569 (2021), 334–376, doi:10.1016/j.jalgebra.2020.10.027; arXiv:1901.09461.DOI
  5. [6]Yonatan Harpaz, Thomas Nikolaus, Maxime Ramzi, and Victor Saunier, Trace methods for stable categories II: The cyclotomic trace map, In preparation, 2026.
  6. [7]Yonatan Harpaz, Thomas Nikolaus, and Victor Saunier, Trace methods for stable categories I: The linear approximation of algebraic K-theory, arXiv:2411.04743; numbering as in arXiv:2411.04743v2, 2024.arxiv.org/abs/2411.04743
  7. [8]Daniel Huybrechts, Fourier–Mukai transforms in algebraic geometry, Oxford Mathematical Monographs, The Clarendon Press, Oxford University Press, Oxford, 2006.DOI
  8. [9]Achim Krause, Jonas McCandless, and Thomas Nikolaus, Polygonic spectra and TR with coefficients, arXiv:2302.07686, 2023.arxiv.org/abs/2302.07686
  9. [10]Tatsuro Kawakami, Teppei Takamatsu, Hiromu Tanaka, Jakub Witaszek, Fuetaro Yobuko, and Shou Yoshikawa, Quasi-F-splittings in birational geometry III, arXiv:2408.01921, 2024.arxiv.org/abs/2408.01921
  10. [11]———, Quasi-F-splittings in birational geometry, Ann. Sci. Éc. Norm. Supér. (4) 58 (2025), 665–748, doi:10.24033/asens.2614; arXiv:2208.08016.DOI
  11. [12]Tatsuro Kawakami, Teppei Takamatsu, and Shou Yoshikawa, Fedder type criteria for quasi-F-splitting I, arXiv:2204.10076, 2022.arxiv.org/abs/2204.10076
  12. [13]———, Fedder type criteria for quasi-F-splitting II, arXiv:2511.17270, 2025.arxiv.org/abs/2511.17270
  13. [15]Haoyang Liu and Tianle Liu, Intrinsic restriction traces and toric dynamics, arXiv:2609.08129, 2026.arxiv.org/abs/2609.08129
  14. [16]Jacob Lurie, Higher Topos Theory, Annals of Mathematics Studies, vol. 170, Princeton University Press, Princeton, NJ, 2009.
  15. [17]Victor Saunier, Trace methods for stable ∞-categories, Ph.D. thesis, Université Sorbonne Paris Nord, 2025.
  16. [18]———, Topological Hochschild homology, traces and higher categories, Lecture notes, Universität Bielefeld, version of October 2026, available at the author’s webpage, 2026.
  17. [19]The Stacks Project Authors, The Stacks Project, https://stacks.math.columbia.edu.
  18. [20]Hiromu Tanaka, Jakub Witaszek, and Fuetaro Yobuko, Quasi-Fᵉ-splittings and quasi-F-regularity, arXiv:2404.06788, 2024.arxiv.org/abs/2404.06788
  19. [21]Fuetaro Yobuko, Quasi-Frobenius splitting and lifting of Calabi–Yau varieties in characteristic p, Math. Z. 292 (2019), 307–316, doi:10.1007/s00209-018-2198-7; arXiv:1704.05604.DOI
  20. [22]Shou Yoshikawa, Fedder-type criterion for quasi-Fᵉ-splitting and quasi-F-regularity, To appear in Nagoya Math. J.; arXiv:2505.09015, 2025.

Paper details

vlnc:W_fag2g7sv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents