Kaplansky test problem for \(R\)-modules (Q1180403)

This paper extends the following earlier result [Logique Anal., Nouv. Sér. 18, 241-308 (1975; Zbl 0359.02052)] of the author: Any ring \(R\) satisfies one of the following two conditions (1) All (left) \(R\)-modules are direct sums of countably generated modules. (2) For any cardinal \(\lambda > | R|\) there is an \(R\)-module of cardinality \(\lambda\) which is not the direct sum of \(R\)-modules of cardinality \(\leq \mu\) for some \(\mu < \lambda\). It is the aim to strengthen condition (2) in terms of endomorphism algebras. Assuming case (2), the author constructs in \(V=L\) certain modules \(M\) with prescribed endomorphisms modulo an ideal of ``small endomorphisms''. In a very special case (e.g. of abelian \(p\)-groups) these ``small endomorphisms'' can be viewed as small endomorphisms in the sense of R. S. Pierce. More generally the ideal of ``small endomorphisms'' consists of those maps which cannot be controlled and are determined by the cardinality of the images. The endomorphism results (Conclusion 4.2) allows us to replace (2) by certain bad decomposition behaviours of \(R\)-modules. (2) can be replaced by ``Kaplansky's test problems fail'', or by ``There are essentially indecomposable modules'', where ``essentially'' means: up to summands of small cardinality. The construction is based on clever embeddings into cartesian products of modules.