betaFeedback

Reading v1View latest version

WorksMathematics

Igusa–Todorov Algebras and the Auslander–Reiten Conjecture

Abstract

We prove that every Igusa–Todorov Artin algebra satisfies the Auslander–Reiten conjecture. More precisely, let VV be an nn-Igusa–Todorov witness, and let tt be the number of isomorphism classes of nonprojective indecomposable summands of VV. If a finitely generated module MM satisfies \Ext_A^i(M,A)=0 for every i>0i>0 and \Ext_A^q(M,M)=0 for 1≤q≤2t+11\leq q\leq 2t+1, then MM is projective. As applications, algebras of representation dimension at most three and algebras satisfying J2m+1=0J^{2m+1}=0 for which A/JmA/J^m has finite representation type satisfy the Auslander–Reiten conjecture.

Introduction

Let AA be an Artin algebra over a commutative Artinian ring RR. We write AA-mod for the category of finitely generated left AA-modules, and all modules considered below belong to this category. For a module VV, let add⁡V\operatorname{add} V denote the full subcategory of direct summands of finite direct sums of copies of VV. The Auslander–Reiten conjecture (ARC), formulated by Auslander and Reiten [2], asserts that

Ext⁡Ai(M,M⊕A)=0 for every i>0⟹M is projective.(1)\operatorname{Ext}^{i}_{A}(M,M\oplus A)=0\text{ for every }i>0 \quad\Longrightarrow\quad M\text{ is projective}. \tag*{(1)}

A module MM is self-orthogonal if Ext⁡Ai(M,M)=0\operatorname{Ext}^{i}_{A}(M,M)=0 for every i>0i>0. It is a generator if A∈add⁡MA\in\operatorname{add} M. Thus (1) is equivalent to the statement that every self-orthogonal generator is projective. The generalized Nakayama conjecture asserts that every indecomposable injective left AA-module occurs as a direct summand of some term in a minimal injective resolution of the left regular module $A_{A}. Auslander and Reiten [2] proved that ARC holds for all Artin algebras if and only if the generalized Nakayama conjecture does. The finitistic dimension conjecture, formulated by Bass [3], states that the supremum of the finite projective dimensions of modules in AA-mod is finite. Chen, Hu, Qin and Wang [6 Introduction] recall that its validity for all Artin algebras would imply ARC for all Artin algebras. They also explain why this global implication does not give the corresponding implication for a fixed algebra. The generalized Nakayama conjecture, the finitistic dimension conjecture and ARC remain open in general. Recent developments concerning ARC include the work of Chen and Xi [7] and Zhang and Zhou [13].

Igusa and Todorov [9] introduced their functions to study the little finitistic dimension. Wei [11 Theorem 2.3] subsequently introduced Igusa–Todorov algebras and proved that they have finite finitistic dimension. Their defining property is the existence of short presentations of syzygies by modules from one fixed additive subcategory. This class contains syzygy-finite algebras, as discussed by Wei [11] and Wei [12 Section 5]. Wei [10 Lemma 2.1] also showed that an algebra of representation dimension at most three is 0-Igusa–Todorov. In Proposition 4.2, we construct a 1-Igusa–Todorov witness under a condition on radical powers. In particular, this construction covers algebras with radical cube zero.

Further developments include the broader class of Lat-Igusa–Todorov algebras introduced by Bravo, Lanzilotta, Mendoza and Vivero [5]. Barrios, Lanzilotta and Mata [4] survey the corresponding homological functions and related classes. These results do not deduce ARC from the Igusa–Todorov property alone. In particular, finite finitistic dimension for one algebra does not imply ARC for that algebra by any known general result. To the best of our knowledge, Theorem I is the first result that proves ARC for every Igusa–Todorov Artin algebra. It simultaneously covers syzygy-finite algebras and algebras of representation dimension at most three. Our argument uses the witness presentations themselves and gives a finite bound for the degree of a nonzero self-extension. Remark 3.5 explains why the proof does not extend directly to Lat-Igusa–Todorov algebras.

The nonprojective support of a module VV is the set of isomorphism classes of its nonprojective indecomposable direct summands. We write t(V)t(V) for the cardinality of this set. Thus multiplicities do not enter t(V)t(V). We also write

⊥A={X∈A-mod∣Ext⁡Ai(X,A)=0 for every i>0}.{}^{\perp} A = \{X \in A\text{-mod} \mid\operatorname{Ext}^{i}_{A}(X,A)=0 \text{ for every } i>0\}.

Recall that AA is nn-Igusa–Todorov if there is a fixed module VV such that each ΩnY\Omega^{n}Y admits an exact sequence 0→V1→V0→ΩnY→00 \to V_{1} \to V_{0} \to\Omega^{n}Y \to0 with V0,V1∈add⁡VV_{0}, V_{1} \in\operatorname{add} V, see Definition 2.1. The module VV is called a witness. Our first result is a finite test for projectivity.

Theorem I. (see Theorem 3.2 for details) Let AA be an nn-Igusa–Todorov Artin algebra with witness VV, and put t=t(V)t=t(V). If M∈⊥AM \in{}^{\perp} A and

Ext⁡Aq(M,M)=0(1≤q≤2t+1),\operatorname{Ext}^{q}_{A}(M,M)=0 \qquad(1 \leq q \leq2t+1),

then MM is projective. Consequently, every Igusa–Todorov Artin algebra satisfies the Auslander–Reiten conjecture.

The bound depends only on the nonprojective support of the chosen witness and not on nn. No optimality is asserted. The hypothesis M∈⊥AM \in{}^{\perp} A still requires vanishing in every positive degree. The finite test concerns self-extensions only.

In Proposition 3.1, we prove the underlying count for a finite segment of syzygies. We pull back each projective cover extension along a presentation in add⁡V\operatorname{add} V. Vanishing pullbacks force the next syzygies into add⁡V\operatorname{add} V. Orthogonality in stable Hom then makes their nonprojective supports disjoint. Nonvanishing pullbacks give a triangular matrix of Ext lengths with positive diagonal. Its factorization through ZtZ^{t} bounds its rank by tt. Each of the two classes therefore contains at most tt indices, contradicting the existence of 2t+12t+1 such extensions. The matrix rank argument is related to that of Zhang and Zhou [13 Lemma 3.8]. The entries used here are differences of Ext lengths attached to witness presentations.

Wei [10 Lemma 2.1] proved that an algebra of representation dimension at most three is 0-Igusa–Todorov. Theorem I supplies the new step from this presentation property to ARC and gives the following consequence.

Theorem II. (see Proposition 4.1 for details) Every Artin algebra of representation dimension at most three satisfies the Auslander–Reiten conjecture.

The second application concerns a condition on powers of the radical. We construct an explicit 1-Igusa–Todorov witness under this condition.

Theorem III. (see Proposition 4.2 and Corollary 4.3) *Let AA be an Artin algebra with radical JJ. If J2m+1=0J^{2m+1}=0 and A/JmA/J^{m} has finite representation type for some m≥1m \geq1, then AA satisfies the

Auslander–Reiten conjecture. In particular, every Artin algebra with radical cube zero satisfies the conjecture.

The same condition on radical powers appears in the work of Dräxler and Happel [8] on the generalized Nakayama conjecture. Here we obtain ARC for arbitrary Artin algebras satisfying that condition. For split algebras of finite dimension with radical cube zero, Zhang and Zhou [13 Theorem 1.1] obtain a bound of 3s+13s+1, where ss is the number of simple modules. That bound uses a different parameter from 2t(V)+12t(V)+1 and is not uniformly smaller for every choice of witness.

The paper is structured as follows. In § 2, we recall the definition of Igusa–Todorov algebras and establish the syzygy identities and a lifting result used in the proof. In § 3, we prove the estimate for a finite segment, deduce Theorem I and give a version restricted to self-orthogonal modules. In § 4, we prove Theorems II and III.

Preliminaries

For a module XX, choose a projective cover PX↠XP_X \twoheadrightarrow X and define ΩX\Omega X by the exact sequence

0⟶ΩX⟶PX⟶X⟶0.0 \longrightarrow\Omega X \longrightarrow P_X \longrightarrow X \longrightarrow0.

Higher syzygies are defined inductively, with Ω0X=X\Omega^0X=X. We write Hom⁡‾A(X,Y)\underline{\operatorname{Hom}}_A(X,Y) for Hom⁡A(X,Y)\operatorname{Hom}_A(X,Y) modulo morphisms that factor through projective modules. If EE is an RR-module of finite length, then ℓRE\ell_RE denotes its composition length. In particular, all Ext groups appearing below have finite RR-length. A module is called basic if its indecomposable direct summands are pairwise nonisomorphic.

We use the syzygy formulation of the definition given by Wei [11], as recalled by Wei [12 Section 5].

Definition 2.1. Let n≥0n\geq0 be an integer. The algebra AA is nn-Igusa–Todorov if there is a module VV such that every module YY admits an exact sequence

0⟶V1⟶V0⟶ΩnY⟶0,V0,V1∈add⁡V.0 \longrightarrow V_1 \longrightarrow V_0 \longrightarrow\Omega^nY \longrightarrow0,\qquad V_0,V_1\in\operatorname{add}V.

We call VV an nn-Igusa–Todorov witness. An algebra is Igusa–Todorov if it is nn-Igusa–Todorov for some nn.

Adjoining AA to a witness preserves the Igusa–Todorov property and does not change t(V)t(V).

Lemma 2.2. Let X∈⊥AX\in{}^\perp A.

  1. ΩaX∈⊥A\Omega^aX\in{}^\perp A for every a≥0a\geq0.

  1. For a≥0a\geq0 and q≥1q\geq1,

Ext⁡Aq(ΩaX,ΩaX)≅Ext⁡Aq(X,X).\operatorname{Ext}_A^q(\Omega^aX,\Omega^aX)\cong\operatorname{Ext}_A^q(X,X).
  1. For a≥1a\geq1 and every module YY,

Hom⁡‾A(ΩaX,Y)≅Ext⁡Aa(X,Y).\underline{\operatorname{Hom}}_A(\Omega^aX,Y)\cong\operatorname{Ext}_A^a(X,Y).
  1. If pd⁡AX<∞\operatorname{pd}_A X<\infty, then XX is projective.

Proof. The first assertion follows by dimension shifting. Let

0⟶ΩX⟶P⟶X⟶00\longrightarrow\Omega X\longrightarrow P\longrightarrow X\longrightarrow0

be a projective cover sequence. Every finitely generated projective module belongs to add⁡A\operatorname{add}A. Hence X∈⊥AX\in{}^\perp A gives Ext⁡A>0(X,P)=0\operatorname{Ext}_A^{>0}(X,P)=0. Dimension shifting in the two variables now gives, for q≥1q\geq1,

Ext⁡Aq(ΩX,ΩX)≅Ext⁡Aq+1(X,ΩX)≅Ext⁡Aq(X,X).\operatorname{Ext}_A^q(\Omega X,\Omega X)\cong\operatorname{Ext}_A^{q+1}(X,\Omega X)\cong\operatorname{Ext}_A^q(X,X).

Iteration proves (2).

For (3), the long exact sequence associated to the projective cover of Ωa−1X\Omega^{a-1}X identifies Ext⁡A1(Ωa−1X,Y)\operatorname{Ext}^{1}_{A}(\Omega^{a-1}X,Y) with the quotient of Hom⁡A(ΩaX,Y)\operatorname{Hom}_{A}(\Omega^{a}X,Y) by the morphisms that extend to that projective cover. These morphisms factor through a projective module. Conversely, if a morphism factors as ΩaX→Q→Y\Omega^{a}X \to Q \to Y with QQ projective, its first factor extends to the cover because

Ext⁡A1(Ωa−1X,Q)≃Ext⁡Aa(X,Q)=0.\operatorname{Ext}^{1}_{A}(\Omega^{a-1}X,Q) \simeq\operatorname{Ext}^{a}_{A}(X,Q)=0.

The quotient is therefore stable Hom, and dimension shifting proves (3).

For (4), suppose that pd⁡AX=d>0\operatorname{pd}_{A}X=d>0. Then ΩdX\Omega^{d}X is projective and

Ext⁡A1(Ωd−1X,ΩdX)≃Ext⁡Ad(X,ΩdX)=0.\operatorname{Ext}^{1}_{A}(\Omega^{d-1}X,\Omega^{d}X) \simeq\operatorname{Ext}^{d}_{A}(X,\Omega^{d}X)=0.

The projective cover sequence of Ωd−1X\Omega^{d-1}X splits, making Ωd−1X\Omega^{d-1}X projective, contrary to the choice of dd. □

Lemma 2.3. If X∈⊥AX \in{}^{\perp}A, then ΩX\Omega X has no nonzero projective direct summand.

Proof. Write $0 \to \Omega X \xrightarrow{u} P \to X \to 0$ for its projective cover sequence. If a projective module QQ is a direct summand of ΩX\Omega X, the projection π:ΩX→Q\pi:\Omega X \to Q extends to π~:P→Q\widetilde{\pi}:P \to Q, because Ext⁡A1(X,Q)=0\operatorname{Ext}^{1}_{A}(X,Q)=0. The composite of the inclusion Q→ΩX→uPQ \to\Omega X \xrightarrow{u} P with π~\widetilde{\pi} is $1_Q$. Thus the image of QQ is a direct summand of PP contained in rad⁡P\operatorname{rad}P. If P=Q⊕P′P=Q\oplus P' with this copy of QQ, then rad⁡P=rad⁡Q⊕rad⁡P′\operatorname{rad}P=\operatorname{rad}Q\oplus\operatorname{rad}P', and hence Q=rad⁡QQ=\operatorname{rad}Q. Nakayama’s lemma gives Q=0Q=0. □

Lemma 2.4. Let M∈⊥AM \in{}^{\perp}A, let N≥1N \ge1 be an integer, and suppose that Ext⁡Aq(M,M)=0\operatorname{Ext}^{q}_{A}(M,M)=0 for 1≤q≤N1 \le q \le N. Fix an integer d≥1d \ge1 and set Xi=Ωd+iMX_i=\Omega^{d+i}M for i≥0i \ge0. Then

  1. Ext⁡Aq(Xi,Xi)=0\operatorname{Ext}^{q}_{A}(X_i,X_i)=0 for 1≤q≤N1 \le q \le N;

  1. Hom⁡A(Xa,Xb)=0\operatorname{Hom}_{A}(X_a,X_b)=0 if 1≤a−b≤N1 \le a-b \le N;

  1. Ext⁡Aq(Xa,Xb)=0\operatorname{Ext}^{q}_{A}(X_a,X_b)=0 if a≥ba \ge b, q≥1q \ge1, and q+a−b≤Nq+a-b \le N.

Proof. Part (1) is Theorem 2.2(2). For a>ba>b, part (3) of that lemma gives

Hom⁡A(Xa,Xb)≃Ext⁡Aa−b(Xb,Xb).\operatorname{Hom}_{A}(X_a,X_b) \simeq\operatorname{Ext}^{a-b}_{A}(X_b,X_b).

For a≥ba \ge b and q≥1q \ge1, dimension shifting gives

Ext⁡Aq(Xa,Xb)≃Ext⁡Aq+a−b(Xb,Xb).\operatorname{Ext}^{q}_{A}(X_a,X_b) \simeq\operatorname{Ext}^{q+a-b}_{A}(X_b,X_b).

These identities prove the remaining assertions. □

Lemma 2.5. Consider exact sequences

0⟶K→vL→pX⟶0,ε:0⟶Y→uP→πX⟶0,0 \longrightarrow K \xrightarrow{v} L \xrightarrow{p} X \longrightarrow0,\qquad\varepsilon: 0 \longrightarrow Y \xrightarrow{u} P \xrightarrow{\pi} X \longrightarrow0,

where PP is projective. If the image of [ε][\varepsilon] in Ext⁡A1(L,Y)\operatorname{Ext}^{1}_{A}(L,Y) under pullback along pp is zero, then

L⊕Y≃K⊕P.L \oplus Y \simeq K \oplus P.

In particular, if K,L∈add⁡VK,L \in\operatorname{add}V and A∈add⁡VA \in\operatorname{add}V, then Y∈add⁡VY \in\operatorname{add}V.

Proof. The pullback splits if and only if pp lifts to a morphism h:L→Ph:L \to P with πh=p\pi h=p. There is then a morphism g:K→Yg:K \to Y such that ug=hvug=hv. The sequence

0⟶K→(v,−g)L⊕Y→(h,u)P⟶00 \longrightarrow K \xrightarrow{(v,-g)} L \oplus Y \xrightarrow{(h,u)} P \longrightarrow0

is exact. Surjectivity follows by first lifting the image in XX through pp. The description of the kernel follows from ker⁡p=Im⁡v\ker p=\operatorname{Im}v and ker⁡π=Im⁡u\ker\pi=\operatorname{Im}u. It splits because PP is projective. The last assertion follows since add⁡V\operatorname{add}V is closed under direct summands. □

The finite self-extension test

The following estimate contains the counting argument. It requires presentations only for the specified syzygies.

Proposition 3.1 (A bound for a finite segment of syzygies). Let M∈⊥AM \in{}^{\perp} A be nonprojective, let VV be a module, and let d,N≥1d,N \ge1 be integers. Suppose that

Ext⁡Aq(M,M)=0(1≤q≤N).\operatorname{Ext}_{A}^{q}(M,M)=0 \quad(1 \le q \le N).

Set Xi=Ωd+iMX_i=\Omega^{d+i}M for 0≤i≤N0 \le i \le N. If, for every 0≤i<N0 \le i < N, there is an exact sequence

0⟶Ki⟶Li→piXi⟶0,Ki,Li∈add⁡V,(2)0 \longrightarrow K_i \longrightarrow L_i \xrightarrow{p_i} X_i \longrightarrow0,\qquad K_i,L_i \in\operatorname{add} V, \tag*{(2)}

then N≤2t(V)N \le2t(V).

Proof. Replace VV by V⊕AV \oplus A and then by a basic module with the same additive closure. Write t=t(V)t=t(V). Adding AA introduces only projective indecomposable summands, so this replacement preserves tt. It also preserves the assumed presentations. Since MM is nonprojective, Theorem 2.2(4) gives pd⁡AM=∞\operatorname{pd}_{A} M=\infty. Every XiX_i is consequently nonzero. Moreover, d≥1d \ge1 ensures that each XiX_i is a positive syzygy of MM. It follows from Theorems 2.2 and 2.3 that Xi∈⊥AX_i \in{}^{\perp} A and that XiX_i has no nonzero projective direct summand. In particular, none of the XiX_i is projective.

For 0≤i<N0 \le i < N, write the projective cover extension as

εi:0⟶Xi+1⟶Pi⟶Xi⟶0.(3)\varepsilon_i: 0 \longrightarrow X_{i+1} \longrightarrow P_i \longrightarrow X_i \longrightarrow0. \tag*{(3)}

Since XiX_i is not projective, [εi]≠0[\varepsilon_i] \ne0. Denote the pullback map by

ρi:Ext⁡A1(Xi,Xi+1)⟶Ext⁡A1(Li,Xi+1).\rho_i:\operatorname{Ext}_{A}^{1}(X_i,X_{i+1}) \longrightarrow\operatorname{Ext}_{A}^{1}(L_i,X_{i+1}).

Partition the index set {0,…,N−1}\{0,\ldots,N-1\} as

I={i∣ρi([εi])=0},II={i∣ρi([εi])≠0}.I=\{i\mid\rho_i([\varepsilon_i])=0\},\qquad II=\{i\mid\rho_i([\varepsilon_i])\ne0\}.

The indices in II. For i∈Ii\in I, Theorem 2.5 gives

Li⊕Xi+1≃Ki⊕Pi,and hence Xi+1∈add⁡V.(4)L_i\oplus X_{i+1}\simeq K_i\oplus P_i,\qquad\text{and hence }X_{i+1}\in\operatorname{add} V. \tag*{(4)}

If i<ji<j lie in II and Xi+1X_{i+1} and Xj+1X_{j+1} have a common indecomposable summand UU, the composite

X_{j+1}\twoheadrightarrow U\lhook\joinrel\longrightarrow X_{i+1}

has nonzero stable class. Indeed, a factorization of this composite through a projective module would, after composing with the corresponding inclusion and projection, factor 1U1_U through a projective module. This would make UU projective, a contradiction. But Theorem 2.4(2) gives

Hom⁡A(Xj+1,Xi+1)=0,1≤j−i≤N−1.\operatorname{Hom}_{A}(X_{j+1},X_{i+1})=0,\qquad1\le j-i\le N-1.

Thus the nonprojective supports of the modules Xi+1X_{i+1}, i∈Ii\in I, are nonempty and pairwise disjoint. By (4), they lie among the tt nonprojective summands of VV. Therefore ∣I∣≤t|I|\le t.

The indices in IIII. Write II={i1<⋯<iq}II=\{i_1<\cdots<i_q\}. If q=0q=0, then ∣II∣≤t|II|\le t is immediate, so suppose q>0q>0. Define the integer matrix C=(cab)1≤a,b≤qC=(c_{ab})_{1\le a,b\le q} by

cab=ℓRExt⁡A1(Lia,Xib+1)−ℓRExt⁡A1(Kia,Xib+1).c_{ab}=\ell_R\operatorname{Ext}_{A}^{1}(L_{i_a},X_{i_b+1})-\ell_R\operatorname{Ext}_{A}^{1}(K_{i_a},X_{i_b+1}).

For later use, applying Hom⁡A(−,Y)\operatorname{Hom}_{A}(-,Y) to (2) gives the exact segment

Ext⁡A1(Xi,Y)⟶Ext⁡A1(Li,Y)⟶Ext⁡A1(Ki,Y)⟶Ext⁡A2(Xi,Y).(5)\operatorname{Ext}_{A}^{1}(X_i,Y)\longrightarrow\operatorname{Ext}_{A}^{1}(L_i,Y)\longrightarrow\operatorname{Ext}_{A}^{1}(K_i,Y) \longrightarrow\operatorname{Ext}_{A}^{2}(X_i,Y). \tag*{(5)}

If a>ba>b, then ia≥ib+1i_a\ge i_b+1. For r=1,2r=1,2, we have

1≤r+ia−(ib+1)≤N.1\le r+i_a-(i_b+1)\le N.

Indeed, r+ia−(ib+1)≤2+(N−1)−1=Nr+i_a-(i_b+1)\le2+(N-1)-1=N. Dimension shifting and Theorem 2.2(2) give

Ext⁡Ar(Xia,Xib+1)≃Ext⁡Ar+ia−ib−1(M,M)=0(r=1,2).\operatorname{Ext}^{r}_{A}(X_{i_a},X_{i_b+1})\simeq\operatorname{Ext}^{r+i_a-i_b-1}_{A}(M,M)=0 \qquad(r=1,2).

The middle map in (5), with i=iai=i_a and Y=Xib+1Y=X_{i_b+1}, is therefore an isomorphism. This proves cab=0c_{ab}=0.

For the diagonal, apply Hom⁡A(Xi,−)\operatorname{Hom}_{A}(X_i,-) to (3). Since Xi∈⊥AX_i\in{}^{\perp}A, we obtain

Ext⁡A2(Xi,Xi+1)≃Ext⁡A1(Xi,Xi)=0.\operatorname{Ext}^{2}_{A}(X_i,X_{i+1})\simeq\operatorname{Ext}^{1}_{A}(X_i,X_i)=0.

Consequently, (5) gives the short exact sequence

0⟶Im⁡ρi⟶Ext⁡A1(Li,Xi+1)⟶Ext⁡A1(Ki,Xi+1)⟶0.0\longrightarrow\operatorname{Im}\rho_i\longrightarrow\operatorname{Ext}^{1}_{A}(L_i,X_{i+1})\longrightarrow\operatorname{Ext}^{1}_{A}(K_i,X_{i+1})\longrightarrow0.

Additivity of composition length yields

caa=ℓR(Im⁡ρia)>0,c_{aa}=\ell_R(\operatorname{Im}\rho_{i_a})>0,

where strict positivity follows from ia∈IIi_a\in II. Thus CC is upper triangular with positive diagonal, so

rank⁡QC=q.(6)\operatorname{rank}_{\mathbb{Q}}C=q. \tag*{(6)}

Let U1,⋯ ,UtU_1,\cdots,U_t be representatives of the nonprojective indecomposable summands of VV. By Krull–Schmidt, we may write

Li≃Lipr⁡⊕⨁α=1tUαℓiα,Ki≃Kipr⁡⊕⨁α=1tUαkiα,L_i\simeq L_i^{\operatorname{pr}}\oplus\bigoplus_{\alpha=1}^{t}U_\alpha^{\ell_{i\alpha}},\qquad K_i\simeq K_i^{\operatorname{pr}}\oplus\bigoplus_{\alpha=1}^{t}U_\alpha^{k_{i\alpha}},

where Lipr⁡L_i^{\operatorname{pr}} and Kipr⁡K_i^{\operatorname{pr}} are projective. Since first Ext vanishes when its first argument is projective,

cab=∑α=1t(ℓiaα−kiaα)ℓRExt⁡A1(Uα,Xib+1).c_{ab}=\sum_{\alpha=1}^{t}(\ell_{i_a\alpha}-k_{i_a\alpha})\ell_R\operatorname{Ext}^{1}_{A}(U_\alpha,X_{i_b+1}).

Hence C=DEC=DE, where

D=(ℓiaα−kiaα)a,α,E=(ℓRExt⁡A1(Uα,Xib+1))α,bD=(\ell_{i_a\alpha}-k_{i_a\alpha})_{a,\alpha},\qquad E=(\ell_R\operatorname{Ext}^{1}_{A}(U_\alpha,X_{i_b+1}))_{\alpha,b}

have sizes q×tq\times t and t×qt\times q, respectively. It follows that rank⁡QC≤t\operatorname{rank}_{\mathbb{Q}}C\le t. Together with (6), this gives ∣II∣=q≤t|II|=q\le t. The same conclusion covers t=0t=0, when the factorization forces C=0C=0.

Combining the two counts proves

N=∣I∣+∣II∣≤2t,N=|I|+|II|\le2t,

as required. □

We now apply the estimate to an Igusa–Todorov witness.

Theorem 3.2 (Finite self-extension test). Let AA be an nn-Igusa–Todorov Artin algebra with witness VV, and put t=t(V)t=t(V). If M∈⊥AM\in{}^{\perp}A and

Ext⁡Aq(M,M)=0(1≤q≤2t+1),\operatorname{Ext}^{q}_{A}(M,M)=0\qquad(1\le q\le2t+1),

then MM is projective. Consequently, every Igusa–Todorov Artin algebra satisfies ARC.

Proof. Suppose that MM is nonprojective and set N=2t+1N=2t+1. For each 0≤i<N0\le i<N, apply the nn-Igusa–Todorov property to Y=Ωi+1MY=\Omega^{i+1}M. This gives a presentation in add⁡V\operatorname{add}V of ΩnY≃Ωn+1+iM\Omega^nY\simeq\Omega^{n+1+i}M. Thus Theorem 3.1 applies with d=n+1d=n+1 and yields 2t+1=N≤2t2t+1=N\le2t, a contradiction. The case t=0t = 0 is included, since then N=1N = 1. Finally, the vanishing in (1) implies the hypotheses of the finite test.

Corollary 3.3. Let AA be a self-injective Igusa–Todorov Artin algebra with witness VV. Every nonprojective module has a nonzero self-extension in some degree between 11 and 2t(V)+12t(V) + 1. In particular, every self-orthogonal AA-module is projective.

Proof. Since AA is injective as a left module over itself, ⊥A=A-mod{}^{\perp}A = A\text{-mod}. Apply Theorem 3.2.

The estimate for a finite segment also gives a criterion restricted to self-orthogonal modules.

Proposition 3.4 (A criterion on self-orthogonal modules). Suppose that there are an integer n≥0n \ge0 and a module VV such that every module YY satisfying Ext⁡Ai(Y,Y⊕A)=0\operatorname{Ext}^{i}_{A}(Y,Y \oplus A) = 0 for all i>0i > 0 admits an exact sequence

0⟶V1⟶V0⟶ΩnY⟶0,V0,V1∈add⁡V.0 \longrightarrow V_{1} \longrightarrow V_{0} \longrightarrow\Omega^{n}Y \longrightarrow0,\qquad V_{0}, V_{1} \in\operatorname{add} V.

Then AA satisfies ARC.

Proof. Put t=t(V)t = t(V). If a nonprojective module MM satisfied the vanishing in (1), then every syzygy of MM would satisfy the same vanishing by Theorem 2.2. In particular, the assumed presentation is available for Y=Ωi+1MY = \Omega^{i+1}M, 0≤i≤2t0 \le i \le2t. Applying Theorem 3.1 with d=n+1d = n + 1 and N=2t+1N = 2t + 1 gives the contradiction 2t+1≤2t2t + 1 \le2t.

Remark 3.5. The finite number of indecomposable isomorphism classes in add⁡V\operatorname{add} V is essential to both counts in the proof. Bravo, Lanzilotta, Mendoza and Vivero [5 Definition 5.1] introduced Lat-Igusa–Todorov algebras, for which the additional summands may range over a subcategory with infinitely many indecomposable isomorphism classes. The argument therefore does not establish ARC for that larger class. In particular, Theorem 3.3 retains the Igusa–Todorov hypothesis.

Applications

Representation dimension at most three

A module GG is a generator-cogenerator if every indecomposable projective module and every indecomposable injective module belongs to add⁡G\operatorname{add} G. Auslander [1] defined the representation dimension of AA as the infimum of gl.dim⁡End⁡A(G)op⁡\operatorname{gl.dim}\operatorname{End}_{A}(G)^{\operatorname{op}} over all generator-cogenerators GG. The following known consequence will be used.

Proposition 4.1. If rep.dim⁡A≤3\operatorname{rep.dim} A \le3, then AA is 00-Igusa–Todorov. Consequently, AA satisfies ARC.

Proof. Choose a generator-cogenerator GG such that gl.dim⁡End⁡A(G)op⁡≤3\operatorname{gl.dim}\operatorname{End}_{A}(G)^{\operatorname{op}} \le3. Apply the characterization of Wei [10 Lemma 2.1] with parameter 1. It gives an exact sequence

0⟶G1⟶G0⟶X⟶00 \longrightarrow G_{1} \longrightarrow G_{0} \longrightarrow X \longrightarrow0

for every X∈A-modX \in A\text{-mod}, where G0,G1∈add⁡GG_{0}, G_{1} \in\operatorname{add} G. Hence GG is a 00-Igusa–Todorov witness. The second assertion follows from Theorem 3.2.

A criterion using radical powers

An Artin algebra CC has finite representation type if C-modC\text{-mod} has only finitely many isomorphism classes of indecomposable modules. A module VV is an additive generator of C-modC\text{-mod} if C-mod=add⁡VC\text{-mod} = \operatorname{add} V.

Proposition 4.2. Let J=rad⁡AJ=\operatorname{rad} A and let m≥1m \ge1. Assume that J2m+1=0J^{2m+1}=0 and C=A/JmC=A/J^m has finite representation type. If VV is an additive generator of CC-mod, viewed as an AA-module, then

W=A⊕V⊕ΩAVW=A\oplus V\oplus\Omega_A V

is a 1-Igusa–Todorov witness for AA.

Proof. Let N∈AN\in A-mod and set X=ΩANX=\Omega_A N. If P↠NP\twoheadrightarrow N is its projective cover, then X⊆JPX\subseteq JP. Consequently,

J2mX⊆J2m+1P=0.J^{2m}X\subseteq J^{2m+1}P=0.

Put U=JmXU=J^mX and Q=X/UQ=X/U. Both UU and QQ are annihilated by JmJ^m, so they are CC-modules and hence belong to add⁡V\operatorname{add} V.

Form the pullback of a projective cover PQ↠QP_Q\twoheadrightarrow Q along X↠QX\twoheadrightarrow Q. Its middle module EE fits into exact sequences

0⟶ΩAQ⟶E⟶X⟶0,0⟶U⟶E⟶PQ⟶0.0\longrightarrow\Omega_A Q\longrightarrow E\longrightarrow X\longrightarrow0,\qquad0\longrightarrow U\longrightarrow E\longrightarrow P_Q\longrightarrow0.

The second sequence splits, so the first becomes

0⟶ΩAQ⟶U⊕PQ⟶ΩAN⟶0.0\longrightarrow\Omega_A Q\longrightarrow U\oplus P_Q\longrightarrow\Omega_A N\longrightarrow0.

Since Q∈add⁡VQ\in\operatorname{add} V, there is a module Q′Q' and an integer r≥1r\ge1 such that Q⊕Q′≃VrQ\oplus Q'\simeq V^r. Taking projective covers gives

ΩAQ⊕ΩAQ′≃ΩA(Vr)≃(ΩAV)r.\Omega_A Q\oplus\Omega_A Q'\simeq\Omega_A(V^r)\simeq(\Omega_A V)^r.

Thus ΩAQ∈add⁡(ΩAV)\Omega_A Q\in\operatorname{add}(\Omega_A V). The two left terms therefore lie in add⁡W\operatorname{add} W, as required. □

Corollary 4.3. Under the hypotheses of Theorem 4.2, let t=t(A⊕V⊕ΩAV)t=t(A\oplus V\oplus\Omega_A V). If M∈⊥AM\in{}^{\perp}A and Ext⁡Aq(M,M)=0\operatorname{Ext}^{q}_{A}(M,M)=0 for 1≤q≤2t+11\le q\le2t+1, then MM is projective. In particular, AA satisfies ARC.

Proof. Apply Theorem 3.2 to the witness in Theorem 4.2. □

Corollary 4.4. Every Artin algebra with radical cube zero satisfies ARC.

Proof. Take m=1m=1 in Theorem 4.3. The algebra A/JA/J is semisimple and therefore has finite representation type. □

These results prove Theorems II and III. For J3=0J^3=0, one may take VV to be the direct sum of representatives of the simple modules. The support of the witness A⊕V⊕ΩAVA\oplus V\oplus\Omega_A V also involves the radicals of indecomposable projective modules. This explains why t(V⊕ΩAV)t(V\oplus\Omega_A V) differs from the parameter counting simple modules in the work of Zhang and Zhou [13].

Remark 4.5. The generalized Nakayama conjecture and ARC are equivalent as assertions over all Artin algebras. This does not give a pointwise implication from the former to the latter, as explained by Chen, Hu, Qin and Wang [6 Introduction]. The application concerning radical powers instead follows directly from the explicit Igusa–Todorov witness.

Disclosure of computational assistance. Chatgpt was used for language editing and improving the clarity of the manuscript. All mathematical results, arguments, and proofs were developed and verified by the authors.

Acknowledgements. Xiaojin Zhang is supported by the National Natural Science Foundation of China (Grant Nos. 12171207 and 12371038). Panyue Zhou is supported by the National Natural Science Foundation of China (Grant No. 12371034).

Data Availability Data sharing is not applicable to this article, as no datasets were generated or analysed during the current study.

Conflict of Interests The authors declare that they have no conflicts of interest.

References

  1. [1]Auslander, M. (1971). Representation dimension of Artin algebras. In Queen Mary College Mathematics Notes.
  2. [2]Auslander, M., & Reiten, I. (1975). On a generalized version of the Nakayama conjecture. Proc. Amer. Math. Soc., 52, 69–74.
  3. [3]Bass, H. (1960). Finitistic dimension and a homological generalization of semi-primary rings. Trans. Amer. Math. Soc., 95, 466–488.
  4. [4]M. Barrios, M. Lanzilotta and G. Mata, A survey on Igusa–Todorov functions, Rev. R. Acad. Cienc. Exactas Fís. Nat. Ser. A Mat. 118 (2024), article 146.arxiv.org/abs/2310.09283
  5. [5]D. Bravo, M. Lanzilotta, O. Mendoza and J. Vivero, Generalised Igusa–Todorov functions and Lat-Igusa–Todorov algebras, J. Algebra 580 (2021), 63–83.arxiv.org/abs/2002.07866
  6. [6]Y. Chen, W. Hu, Y. Qin and R. Wang, Singular equivalences and Auslander–Reiten conjecture, J. Algebra 623 (2023), 42–63.arxiv.org/abs/2011.02729
  7. [7]H. Chen and C. Xi, Homological theory of self-orthogonal modules, Trans. Amer. Math. Soc. 378 (2025), no. 10, 7287–7335.DOI
  8. [8]Dräxler, P., & Happel, D. (1992). A proof of the generalized Nakayama conjecture for algebras with $J^{2\ell+1}=0$ and $A/J^\ell$ representation finite. J. Pure Appl. Algebra, 78(2), 161–164.
  9. [9]K. Igusa and G. Todorov, On the finitistic global dimension conjecture for Artin algebras, in: Representations of Algebras and Related Topics, Fields Inst. Commun. 45, Amer. Math. Soc., Providence, RI, 2005, 201–204.DOI
  10. [10]J. Wei, Finitistic and representation dimensions, preprint (2008), arXiv: 0803.3364v3.arXiv
  11. [11]J. Wei, Finitistic dimension and Igusa–Todorov algebras, Adv. Math. 222 (2009), no. 6, 2215–2226.DOI
  12. [12]J. Wei, Derived categories and syzygies, preprint (2011), arXiv: 1109.6226.arXiv
  13. [13]X. Zhang and P. Zhou, The Auslander–Reiten conjecture for algebras with radical cube zero, preprint (2026), arXiv: 2609.08679.arXiv

Paper details

vlnc:W_8d9cqtr5
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