Large indecomposable modules in torsion theories (Q1094469): Difference between revisions

The title of this paper is somewhat misleading. The author constructs arbitrarily large indecomposable R-modules over a commutative ring R without non-trivial idempotents within a class \({\mathfrak C}\). As mentioned at the end of the introduction, it is enough to assume that \({\mathfrak C}\) is closed under taking extensions, direct sums and images of homomorphisms in \({\mathfrak C}\). In order to get off the ground, the class \({\mathfrak C}\) must contain at least seven rigid R-modules \(A_ i\) \((i=0,...,6)\), i.e. \(Hom_ R(A_ i,A_ j)=R\cdot \delta_{ij}\) satisfying a non-projectivity-condition: R embeds into \(Ext^ 1_ R(A_ i,A_ 0)\) for all \(i=1,...,6\). This is used to derive a small fully rigid system of R-modules \(C_ X\quad (X\subseteq I)\) i.e. \(C_ X\subseteq C_ Y\) for \(X\subseteq Y\subseteq I\) and \(Hom(C_ X,C_ Y)=R\cdot \delta_{XY}\) (X,Y\(\subseteq I)\) with \(\delta_{XY}=1\) for \(X\subseteq Y\) and \(\delta_{XY}=0\) for \(X\not\subseteq Y.\) After the case \(| I| =6\) is established straight forwardly, the existence of the desired \({\mathfrak C}\)-modules follows from a general construction lemma of Corner's, which is based on \(| I| =6.\) The main application is a new proof of the existence of large indecomposable divisible torsion modules over certain valuation domains.