Remarks on modules approximated by \(G\)-projective modules (Q855719)

\(R\) denotes a commutative, noetherian, henselian, local ring and mod\(R\) is the class of finitely generated \(R\)-modules; subcategories of mod\(R\) are assumed to be full subcategories, closed under isomorphisms. \(G\) stands for Gorenstein here. An \(R\)-module \(X\) is said to be \(G\)-projective, if a) the natural double dual map \(X\longrightarrow X^{**}\) is an isomorphism and b) \(\forall i>0\)\,\, Ext\(^i_R(X,R)=0\) and Ext\(^i_R(X^*,R)=0\). If \(\mathcal G\) denotes the subcategory of mod\(R\) of all \(G\)-projective \(R\)-modules, denote by \(\mathcal G^\perp=\{ Y : \forall X\in\mathcal G\, \forall i>0\,\, \text{Ext}^i_R(X,Y)=0\}\) and \(^\perp(\mathcal G^\perp)=\{ Z : \forall Y\in\mathcal G^\perp\, \forall i>0\,\, \text{Ext}^i_R(Z,Y)=0\}\) (the latter class is a superclass of \(\mathcal G\)). One of the author's results states that \(\mathcal G=^\perp(\mathcal G^\perp)\) iff every module in \(^\perp(\mathcal G^\perp)\) is torsionless. However he also asks a more specific question (as a conjecture), namely whether the former equality is always true; it is true for a generically Gorenstein Cohen-Macaulay local ring with canonical module. Given a subcategory \(\mathcal X\) of mod\(R\), an \(M\in\mathcal X\), an \(X\in\mathcal X\) and a homomomorphism \(\phi:X\longrightarrow M\); the map \(\phi\) (or \(X\) itself) is said to be a right \(\mathcal X\)-approximation of \(M\) if it is a universal arrow, namely, if for every homomorphism \(\phi^\prime:X^\prime\longrightarrow M\), \(X^\prime\in\mathcal X\), there is a homomorphism \(f:X^\prime\longrightarrow X\), with \(\phi^\prime=\phi f\). All modules of finite \(G\)-dimension admit right \(\mathcal G\)-approximation. It is a general purpose of this paper to look into cases when a given \(R\)-module has a right \(\mathcal G\)-approximation. To this end, the author introduces subcategories rap\(\mathcal G\) and lap\(\mathcal G\) as subcategories of mod\(R\) containing all modules that have right, and respectively left \(\mathcal G\)-approximation. Both of these subcategories of mod\(R\) contain \(\mathcal G\) and are closed under finite direct sums and direct summands. A subcategory \(\mathcal X\) of mod\(R\) closed under direct summands is called thick, if, for every exact sequence \(0\longrightarrow L\longrightarrow M\longrightarrow N\longrightarrow 0\) in mod\(R\), if any two of \(L, M, N\) belong to \(\mathcal X\) then so does the third. The author proves then that rap\(\mathcal G\) is the smallest subcategory of mod\(R\) that contains \(\mathcal G\) and \(\mathcal G^\perp\) and is closed with respect to direct summands and extensions; furthermore, rap\(\mathcal G\) is a thick subcategory of mod\(R\). A characterization of rap\(\mathcal G\) in terms of Tate cohomologies is also given. These are few of the author's results. The strength of the paper is that it introduces a number of notions that could (or should) be explored further.