Criteria for a direct sum of modules to be a multiplication module over noncommutative rings (Q2035822)

A multiplication module is defined as an \(R\)-module \(M\) of which every submodule equals \(IM\) for some two-sided ideal \(I\) of the ring \(R\) (see [\textit{A. Barnard}, Multiplication modules, J. Algebra 71, 174--178 (1981; Zbl 0468.13011)]). These modules have been studied intensively over the years, but most results were on modules over commutative rings. In the paper under review, rings are allowed to be noncommutative. The main results are a series of necessary and sufficient conditions for a direct sum of nonzero \(R\)-modules to be a multiplication module. In the run-up to the main theorems, several characterizations of multiplication modules are given. It is shown, inter alia, that the algebras \(\mathbb{I}_{n}\) form a subclass of the class of multiplication rings and that, in a multiplication module that is a direct sum of modules, the annihilators of the summands are incomparable. If an \(R\)-module \(M\) has two direct decompositions \(\oplus M_{i \in I}\) and \(\oplus N_{j \in J}\), then the decompositions are said to have the refinement condition if \(M = \oplus_{i \in I, j \in J} (M_{i} \cap N_{j})\) and if any two of the direct decompositions of \(M\) satisfy this property, then \(M\) is said to satisfy the refinement condition. Every multiplication module is proven to satisfy the refinement condition. Multiplication modules are also shown to be a unique direct sum of indecomposable modules and to satisfy the direct sum cancellation property.