Grothendieck groups of subcategories (Q1194199)

The author considers pairs \(({\mathbf I},{\mathbf P})\) of full subcategories of the category mod-\(A\) of all finitely generated right \(A\)-modules, \(A\) an Artinian algebra. Both categories are assumed to be closed under extensions and direct summands. The pair is also supposed to be Ext- orthogonal, that is, \(\text{Ext}^ i_ A(P,I) = 0\) for all \(i \geq 1\), \(I\in {\mathbf I}\), \(P\in {\mathbf P}\), and left complete: for any \(X\) in mod- \(A\), there is a short exact sequence \(0 \to X\to I\to P\to 0\), \(I \in {\mathbf I}\), \(P\in {\mathbf P}\). The main result is that there is then a short exact sequence \[ 0 \to G({\mathbf I} \cap {\mathbf P}) \to G({\mathbf I}) \oplus G({\mathbf P}) \to G(\text{mod- }A) \to 0 \] of Grothendieck groups of categories. This sequence is in fact split, the splitting of the injection being given explicitly in terms of a defect module \(d(X)\) attached to an arbitrary module \(X\). First, the defect \(d({\mathbf E})\) is defined for a short exact sequence \({\mathbf E}: 0 \to X \to Y \to Z \to 0\), and then for a module \(X\) by summing the defects of the sequences associated to any composition series of \(X\). It is nearly additive on exact sequences: \(d(X) + d(Z) + d({\mathbf E}) = d(Y)\). An explicit example arising from tilting modules is discussed.