On a question of matlis

From MaRDI portal





The subject of this paper concerns necessary and sufficient conditions for a regular ring \(R\) to be a self-injective ring. The motivation for this study comes from a paper by \textit{E. Matlis} [Ill. J. Math. 27, 353-391 (1983; Zbl 0519.13004)] where the following question was posed: Is the ring \(K/J\) self-injective if \(K=\prod_{i\in I}K_i\), \(J=\sum_{i\in I}\oplus K_i\), where \(K_i\) is a field and \(I\) an infinite set? The author easily settles this question in the negative. Since the ring \(K/J\) is a regular ring, a natural question to ask is when a regular ring will be self-injective. From two earlier results of the same author, it is derived that in self-injective rings a maximal right ideal is projective if and only if it is a direct summand. In \(K/J\) not a single maximal ideal is a direct summand, leading one to suspect that the converse may be true. However, the author shows by means of an example that there are regular rings in which projective maximal ideals are direct summands, but which are still not self-injective. Characterizations of regular noncommutative rings which are self-injective have appeared in the literature, but all of them involved the category of \(R\)-modules and are therefore difficult to implement. The main theorem gives an intrinsic characterization of a self-injective reduced (not necessarily commutative) ring. This result is then used to give a second solution to the question of Matlis mentioned earlier.











This page was built for publication: On a question of matlis

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4351690)