On duality theory and \(AB5^*\) modules (Q1369596): Difference between revisions

A subcategory \({\mathcal K}\subseteq R\)-Mod is called finitely closed if it is closed with respect to submodules, factor modules and finite direct sums. Dualities between the finitely closed subcategories of module categories are studied with the help of \(AB5*\)-modules and the functor \(D\colon R\text{-Mod}\to\text{Mod-}S\), where \(S\subseteq\text{End}(_RU)\) and \(_RU=\bigoplus_{i\in I} E_i\) is a minimal cogenerator of \(R\)-Mod: \(D(M)=\{f\colon M\to U\mid\text{Im} f\subseteq\bigoplus_{i\in B_f} E_i\) for some finite \(B_f\subseteq I\}\). The module \(_RM\) is an \(AB5^*\)-module if and only if the lattices of submodules \(L(_RM)\) and \(L(D(M)_S)\) are anti-isomorphic. For a finitely closed subcategory \({\mathcal K}\subseteq R\)-Mod the following conditions are equivalent: (a) \({\mathcal K}\) is an \(AB5^*\)-category; (b) \(D({\mathcal K})\) is a finitely closed subcategory in Mod-\(S\) and the functor \(D\) induces a duality between \({\mathcal K}\) and \(D({\mathcal K})\); (c) \({\mathcal K}\) is dual to some finitely closed subcategory \({\mathcal L}\subseteq\text{Mod-}T\) for some ring \(T\). In particular, \(M\) is locally an \(AB5^*\)-module if and only if there exists a Morita duality (in the sense of Anh-Wiegandt) between \(\sigma[M]\) and \(\sigma[N]\subseteq\text{Mod-}T\) for some ring \(T\) and \(N\in\text{Mod-}T\). Every finite direct sum \(M^n\) of \(M\) is an \(AB5^*\)-module if and only if \(M\) is a linearly compact object in the category \(\sigma[M]\).