When are definable classes tilting and cotilting classes? (Q2378606)

A class of modules is `definable' if it is closed under direct products, direct limits and pure submodules. By two correspondences \(_R\mathcal F@>\rho>>\mathcal T_R\) and \(_R\mathcal T@>\sigma>>\mathcal F_R\) a bijection is obtained between definable classes of \(R\)-Mod and definable classes of Mod-\(R\), such that resolving classes correspond to coresolving classes and classes of cofinite type correspond to classes of finite type. The asymmetry which appears between the notions of finite type and cofinite type for \(n\)-tilting and \(n\)-cotilting classes is mentioned. In particular, the tilting classes are of finite type, while cotilting classes are not always of cofinite type. To clarify this phenomenon, the following theorem for \(n\)-cotilting classes is proved, the analogue of which for \(n\)-tilting classes does not hold. For \(n\in\mathbb{N}\) and class \(\mathcal F\subseteq R\)-Mod the following are equivalent: 1) \({\mathcal F}\) is an \(n\)-cotilting class; 2) \(\mathcal F\) is resolving, definable and \(\mathcal F^\perp\subseteq\mathcal I_n\); 3) \(\mathcal F\) is resolving, definable and closed under \(n\)-submodules. For \(n\)-cotilting classes the following result is proved. Let \(n\in\mathbb{N}\) and \(\mathcal T\) be a definable class of \(R\)-Mod. The following are equivalent: 1) \(\mathcal T\) is an \(n\)-tilting class; 2) \(\sigma(\mathcal T)\) is an \(n\)-cotilting class of cofinite type; 3) \(\mathcal T\) is coresolving, special preenveloping and closed under \(n\)-images. As a conclusion is mentioned that if \(\mathcal F\) is an \(n\)-cotilting class of \(R\)-Mod, then \(\rho(\mathcal F)\) is a definable coresolving class closed under \(n\)-images and in this case \(\rho(\mathcal F)\) is an \(n\)-tilting class if and only if \(\mathcal F\) is an \(n\)-cotilting class of cofinite type.