Representation dimension: An invariant under stable equivalence (Q4671902)

Artin algebras of finite representation type were studied in many different ways. One of them was the investigation of M. Auslander. He proved that an Artin algebra \(\Lambda\) is of finite representation type if and only if it is the endomorphism algebra of a projective-injective module over an Artin algebra with global dimension at most two and dominant dimension at least two. This motivated Auslander to introduce the representation dimension in 1971 in his notes [\textit{M. Auslander}, Representation dimension of Artin algebras. Queen Mary College Mathematics Notes (1971; Zbl 0331.16026)]. This notion was hoped to measure how far away an Artin algebra is from being representation-finite. To understand the representation dimension and the difficulty in studying it, now let us recall shortly the history of the developments and advances in representation dimension. In his notes in 1971, Auslander introduced and studied the notion in detail, and obtained many fundamental results. The definition of representation dimension is as follows: Let \(A\) be a given Artin algebra. Consider an Artin algebra \(\Lambda\) of dominant dimension at least two such that \(\text{End}_\Lambda(I_0(\Lambda))\) is Morita equivalent to \(A\), where \(I_0(\Lambda)\) is the injective envelope of \(_\Lambda\Lambda\). Then the representation dimension of \(A\) is defined to be the minimum of the global dimensions of all possible \(\Lambda\), and denoted by \(\text{rep.dim}(A)\). In fact, Auslander also proved in the notes that the above definition is equivalent to the following definition: \[ \text{rep.dim}(A)=\inf\{\text{gl.dim(End}_A(M))\mid M\text{ is a generator-cogenerator}\}. \] Note that an \(A\)-module \(M\) is called a generator-cogenerator if every indecomposable projective module and also every indecomposable injective module is isomorphic to a summand of \(M\). By definition, for a semisimple algebra its representation dimension is zero. It is easy to see that there is no algebra whose representation dimension is \(1\). The results on representation dimension in the Queen Mary College Mathematics Notes can be collected as follows: Let \(A\) be a non-semisimple Artin algebra. Then: (1) \(\text{rep.dim}(A)=2\) if and only if \(A\) is representation-finite. (2) If \(A\) is a self-injective algebra, then \(\text{rep.dim}(A)\leqslant\text{LL}(A)\), where \(\text{LL}(A)\) stands for the Loewy length of \(A\). (3) If \(\text{gl.dim}(A)\leqslant 1\), then \(\text{rep.dim}(A)\leqslant 3\). (4) If the radical-square of \(A\) is zero, then \(\text{rep.dim}(A)\leqslant 3\). (5) Let \(n\) be the nilpotency index of the Jacobson radical \(\text{rad}(A)\) of \(A\). If \(A/\text{rad}^{n-1}(A)\) is representation-finite, then \(\text{rep.dim}(A)\leq 3\). Note that the bound \(\text{LL}(A)\) in (2) follows from a result in another paper of Auslander's. In the notes the bound was given by \(\text{LL}(A)+1\). In the notes in 1971, Auslander mentioned also several questions on representation dimension. For example, he asked if the representation dimension of an Artin algebra is always finite? or bounded above by \(3\)? These were all results at the time we knew on the representation dimension. 4 years later, in 1975, \textit{R. M. Fossum, P. A. Griffith} and \textit{I. Reiten} proved the following results [Trivial extensions of Abelian categories. Homological algebra of trivial extensions of Abelian categories with applications to ring theory. Lect. Notes Math. 456 (1975; Zbl 0303.18006), p. 114-115]: Let \(A\) be an Artin algebra and let \(T_2(A)\) denote the \(2\times 2\) triangular matrix algebra over \(A\). Then: (6) \(\text{rep.dim}(T_2(A))\leqslant\text{rep.dim}(A)+2\). (7) If \(\text{gl.dim}(A)<\infty\) and if \(A\) is weakly right \(T\)-stable, then \(\text{rep.dim}(A)\leq 1+2\text{gl.dim}(A)\). In 1976, \textit{M. Auslander} and \textit{I. Reiten} extended (3) to algebras stably equivalent to hereditary algebras [in Adv. Math. 17, 167-195 (1975; Zbl 0312.16018)]. (8) If an Artin algebra is stably equivalent to a hereditary algebra, then its representation dimension is bounded above by \(3\). As we saw, there were not so many results on the representation dimension. In 1998, Idun Reiten gave a talk in the international conference on representation theory of algebras in Bielefeld. As one of her questions in the representation theory of Artin algebras she proposed to study the representation dimension. Since then, the reviewer has begun to study and understand the notion of representation dimension. In 1998, the reviewer obtained the following result which was published in 2000 [in J. Algebra 226, No. 1, 332-346 (2000; Zbl 0960.16016)]: (9) If \(A_i\) is a finite-dimensional algebra over a perfect field \(k\) for \(i=1,\dots,n\), then \[ \text{rep.dim}(A_1\otimes_k\cdots\otimes_kA_n)\leq\sum_{i=1}^n\text{rep.dim}(A_i). \] Two years later, among other results, the reviewer proved [in Adv. Math. 168, No. 2, 193-212 (2002; Zbl 1050.16004)] the following results: (10) The representation dimension is invarant under stable equivalences of Morita type. (11) If \(A\) is a stably hereditary algebra, then \(\text{rep.dim}(A)\leq 3\). For the definition of stable equivalence of Morita type we refer the reader to \textit{M. Broué}'s paper [in: Finite dimensional algebras and related topics. NATO ASI Ser., Ser. C, Math. Phys. Sci. 424, 1-26 (1994; Zbl 0827.20007)]; and for more information on recent advances we refer to [\textit{Y. M. Liu} and \textit{C. C. Xi}, ``Constructions of stable equivalences of Morita type for finite dimensional algebras I-III''. Preprints are available at http://math.bnu.edu.cn/\(^{\sim}\)ccxi/]. Here the notion of stably hereditary algebra was introduced in order to generalize the notion of ``stably equivalent to hereditary algebra''. Moreover, in this paper of the reviewer it was conjectured that (a) representation dimension is invariant under stable equivalences (see Section 4, p. 200). (b) Representation dimension is always finite. (c) For each finitely generated module \(M\) over an Artin algebra \(A\), there is always a finitely generated \(A\)-module \(M'\) such that the endomorphism algebra of \(M\oplus M'\) is quasi-hereditary. In 2003, the conjecture (c), and therefore the conjecture (b), was solved positively by \textit{O. Iyama} [in Proc. Am. Math. Soc. 131, No. 4, 1011-1014 (2003; Zbl 1018.16010)]. The paper under review extends the result (10) to stable equivalence, and thus solves the conjecture (a). Since the year 2000, many authors have been interested in studying representation dimension because the notion is related to many subjects in mathematics: to quasi-hereditary algebras, to dimension of triangulated categories, to the finitistic dimension conjecture, to algebraic geometry via coherent functors and so on. Let us just mention the recent result of Igusa-Todorov: if \(\text{rep.dim}(A)\leq 3\), then the finitistic dimension of \(A\) is finite. Note that Rouquier proved that the representation dimension can take any value \(n\geq 2\). From the history of the notion of representation dimension and its connection with other subjects the reviewer feels that the representation dimension is still a mysterious subject in the representation theory, and deserves more attention. For recent developments on the representation dimension we mention some authors and papers: Th. Holm (also jointly with others) has determined the representation dimensions of many tame blocks of group algebras and related algebras. His aim is to show that tame algebras would have representation dimension \(3\). K. Erdmann, Th. Holm, O. Iyama and J. Schröer proved that if \(B\) is a subalgebra of an algebra \(A\) with the same radical and if \(A\) is representation-finite, then \(\text{rep.dim}(B)\leq 3\). In 2004, F. U. Coelho and M. I. Platzeck showed that if \(A\) is a glued algebra or the trivial extension of an iterated tilted algebra, then the representation dimension of \(A\) is bounded above by \(3\). For some more recent results on representation dimension one may see the preprints [\textit{C. C. Xi}, On the finitistic dimension conjecture. II: Related to finite global dimension. Adv. Math. (to appear)], [\textit{I. Assem, M. I. Platzeck} and \textit{S. Trepode}, On the representation dimension of tilted and laura algebras. Preprint, (2004)] and [\textit{C. C. Xi}, Adjoint functors and representation dimensions. Preprint available at http://math.bnu.edu.cn/\(^{\sim}\)ccxi/].