Weak cotorsion, $\tau$-tilting and two-term categories
A. B. Buan,Yu Zhou
2021-11-22
Abstract:Motivated by its links to τ -tilting theory, we introduce a generalization of cotorsion pairs in module categories. Such pairs are also linked to co-t-structures in corresponding triangulated categories, and to cotorsion pairs in certain extension closed (and hence extriangulated) subcategories, which we call two-term categories. Introduction and main results We introduce the notion of weak cotorsion pairs in module categories, and show that support τ -tilting modules, as defined by Adachi-Iyama-Reiten [AIR], give rise to such pairs. In particular, this allows us to generalize a theorem of [BBOS], from a tilting to a support τ -tilting setting, which can be considered as a characterization of the (left) weak cotorsion pairs which come from τ -tilting. Our main result concerns modules over a finite dimensional algebra A, but we will need to study cotorsion in two-term categories, as was recently introduced and studied by Paukzstello and Zvonareva [PZ], in order to prove our main results. The motivating, and for us also the most central, example of a two-term category, is the category whose objects are maps between finite-dimensional projective A-modules, and whose morphisms are chain maps up to homotopy. Note, that this is an extension closed subcategory of a triangulated category, the homotopy category of complexes of projectives. Hence it is in particular equipped with a natural structure of an extriangulated category [NP]. In [PZ], they prove a version of the HRS-tilting theorem [HRS] in the setting of complete cotorsion pairs in two-term categories. Moreover, they also prove a correspondence between such complete cotorsion pairs and functorially finite torsion pairs in a corresponding module category. Both these results are crucial for proving our main theorems. We provide independent proofs of these results, adapted to our setting. In addition, we give a version of the Brenner-Butler theorem [BB] for two-term categories, which will also be needed for our main results. In order to state precisely our main results, we need some notation. Let A be a finite dimensional algebra, and let modA be the category of finite dimensional left A-modules. Let GenT denote the full subcategory of modules which are generated by direct sums of copies of a module T in modA. For a full subcategory X , let X = {Y | Hom(X , Y ) = 0}, and for a module X, let X = {X}. Now recall the following important theorem from [AIR]. Theorem 0.1. The map T 7→ (Gen T, T) gives a bijection between support τ -tilting modules and functorially finite torsion pairs in modA. This work was supported by grant number FRINAT 301375 from the Norwegian Research Council and by grant number 12031007 from National Natural Science Foundation of China. The first named author would like to thank the Isaac Newton Institute for Mathematical Sciences, Cambridge, for support and hospitality during the programme Cluster Algebras and Representation Theory where work on this paper was undertaken. This work was supported by EPSRC grant no EP/K032208/1. 1 2 ASLAK BAKKE BUAN AND YU ZHOU Our aim is to give a cotorsion version of this theorem, for this we introduce the notion of left weak cotorsion pair: Definition 0.2. A pair (C,T ) of subcategories of modA is called a left weak cotorsion pair (or lw-cotorsion pair for short) if (1) Ext(C,T ) = 0; (2) for any M ∈ modA, there are exact sequences 0 → YM → XM fM −−→ M → 0 and M g −−→ Y M → X → 0 with XM ,X M ∈ C, YM , Y M ∈ T , fM a right C-approximation of M , and g M a left T -approximation of M . Let 1X = {Y | Ext(Y,X ) = 0} for a subcategory X . Our first main theorem is the following. Theorem 0.3. Let T be a support τ -tilting A-module. Then (1GenT ,Gen T ) is a lwcotorsion pair in modA. Note that there are in general lw-cotorsion pairs (C,T ), such that T is not closed under factors. Indeed, for any algebra A, it is easily verified that (modA,I(A)), where I(A) denotes the category of all injective objects in modA, is a lw-cotorsion pair (in fact it is a cotorsion pair). And for any non-hereditary algebra A, we have that I(A) is not closed under factors. A triple of subcategories (C,T ,F) is called a lw-cotorsion-torsion triple if (C,T ) is a lw-cotorsion pair, and (T ,F) is a torsion pair. We prove the following strengthening of the above theorem, where the tilting part is [BBOS, Theorem 2.29]. Here addT denotes the additive closure of a module T . Theorem 0.4. The map T 7→ (1GenT ,Gen T, T) is a bijection between basic support τ -tilting modules and lw-cotorsion-torsion triples, with inverse (C,T ,F) 7→ T , where addT = C ∩ T . The map specializes to a bijection between tilting modules and cotorsion-torsion triples. Note that a support τ -tilting module T is tilting if and only if it is faithful (i.e. there is a monomorphism A → T t for some positive integer t (see [AIR, Proposition 2.2]), and that a lw-cotorsion pair is a (complete) cotorsion pair if the map g in Definition 0.2 can always be chosen to be injective. As a consequence, we prove the following, which also generalizes similar results of [BBOS] for tilting objects. Corollary 0.5. For a lw-cotorsion-torsion triple (C,T ,F), we have C/(C ∩ T ) ≃ F
Mathematics