The derived category with respect to a generator (Q268197)

Let \(\mathcal{E}\) be a Grothendieck category equipped with a set of generators \(G_{i}\), and abbreviate \(G = \bigoplus_{i} G_{i}\). A map \(f : C \rightarrow D\) of chain complexes in \(\mathcal{E}\) is called a \(G\)-isomorphism if \(\hom_{\mathcal{E}} (G, f) = \prod_{i} \hom_{\mathcal{E}} (G_{i}, f)\) is a quasi-isomorphism (that is, a homology isomorphism) in the category of chain complexes of abelian groups. The \(G\)-derived category \(\mathcal{D}(G)\) is obtained from the category \(\mathrm{Ch}(\mathcal{E})\) of chain complexes in \(\mathcal{E}\) by inverting the \(G\)-isomorphisms. In case \(\mathcal{E}\) is the category of \(R\)-modules, for \(R\) a (unital) ring, and \(G = R\) this recovers the usual derived category such that \(G\)-equivalences are precisely the quasi-isomorphisms. In order to work with the \(\mathcal{E}\)-derived category it is convenient to describe it as the homotopy category of a model category structure on \(\mathrm{Ch}(\mathcal{E})\). Theorem A of the paper under review contains such a description generalising the usual projective model structure on chain complexes. In case each \(G_{i}\) is finitely presented, the \(G\)-derived category is compactly generated and there is a second model structure (with the same weak equivalences, and thus the same homotopy category) on \(\mathrm{Ch}(\mathcal{E})\) generalising the usual injective model structure on chain complexes. By connecting these model structures with the theory of cotorsion pairs, the author constructs several recollements involving \(G\)-derived categories and related constructions (Theorems B, C and D). Morphisms in the \(G\)-derived category are closely linked to \(G\text{-}\mathrm{Ext}\)-groups, that is, the groups of \(n\)-fold \(G\)-exact sequences. This notion is obtained from an exact structure (in the sense of Quillen) attached to the given data: a sequence \(A \rightarrow B \rightarrow C\) in \(\mathcal{E}\) is called \(G\)-exact if the sequence \[ 0 \rightarrow \hom_{\mathcal{E}}(G, A) \rightarrow \hom_{\mathcal{E}}(G, B) \rightarrow \hom_ {\mathcal{E}}(G, C) \rightarrow 0 \] is short exact in the category of abelian groups; exactness for longer sequences is defined in terms of splicing, and the Yoneda Ext groups thus resulting, viz., the \(G\text{-}\mathrm{Ext}\)-groups, can be computed in the \(G\)-derived category as \[ G\text{-}\mathrm{Ext}^{n} (A,B) = \hom_{\mathcal{D}(G)}(A, \Sigma^{n}B) \] as expected. The paper is carefully written, and includes a good list of situations to which the described machinery applies, containing among others the cases of pure, \(\lambda\)-pure and clean derived categories, and sheaves of modules over a ringed space.