Retractions and Gorenstein homological properties. (Q742932): Difference between revisions

The paper under review studies Gorenstein projectives over an Artin algebra by using localizing simples. Recall that a simple module of projective dimension at most one over an Artin algebra \(A\) is said to be localizable if all of its self-extentions are split. Let \(S\) be such a module. The perpendicular category \(S^\perp\) consists of all modules \(X\) such that \(\Hom(S,X)=0\) and \(\mathrm{Ext}^1(S,X)=0\). By a result of Geigle and Lenzing, \(S^\perp\) is a reflective subcategory of the category of finitely generated modules, i.e., the inclusion functor \(i\) has a left adjoint \(i_\lambda\) (on page 715 there is a claim that the inclusion has a right adjoint, but that seems to be an isolated misprint). The unit of adjunction \(\eta\colon\mathrm{Id}\to ii_\lambda\) gives rise to an endofunctor \(L:=ii_\lambda\) on the ambient module category. As the inclusion functor is fully faithful, the counit \(i_\lambda i\to\mathrm{Id}_{S^\perp}\) of adjunction is an isomorphism. Consequently, one gets an isomorphism \(L(\eta)\colon L\to L^2\). In other words, the pair \((L,\eta)\) is a localization functor. These data allow to define an algebra structure on \(L(A)\) such that \(\eta_A\colon A\to L(A)\) is an algebra homomorphism. Moreover, the induced \(A\)-module structure on \(L(A)\) coincides with the original one. This homomorphism is called the left retraction associated with the localizable simple \(S\). The authors apply left retractions to the study of Gorenstein properties of Nakayama algebras. In particular, they classify Nakayama algebras with at most three simples in the following three classes: Gorenstein algebras, non-Gorenstein algebras all of whose Gorenstein projectives are projective, and non-Gorenstein algebras admitting non-projective Gorenstein projectives.