Notes on reduced Rickart rings. I. Representation and equational axiomatizations (Q1646660)

This paper has two major aims. First, to generalize results on reduced Rickart rings from the literature. Second, to give explicit equational axiomatizations of the class of such rings. Recall that a ring \(R\) is right Rickart when every principal right ideal is projective as a right \(R\)-module. When \(R\) is reduced (i.e., there are no nonzero nilpotent elements), the Rickart condition becomes a left-right symmetric notion and it has been studied in a number of papers. The authors of this paper give a thorough review of the theory of such rings in the first four sections of their paper, especially concentrating on the work of \textit{I. Sussman} from [Math. Ann. 136, 326--338 (1958; Zbl 0083.02902)]. In Section 5, the authors show that reduced Rickart rings are (up to isomorphism) exactly the subdirect products of domains that contain certain ``associate idempotents''. This generalizes Theorem 3.3 from Sussman's paper, by dropping the commutativity hypothesis, as well as filling a small gap in Sussman's argument. This generalization also follows from two other well-known results. The first is that reduced rings are exactly the subdirect products of domains, proved by \textit{V. A. Andrunakievich} and \textit{Yu. M. Ryabukhin} in [Sov. Math., Dokl. 9, 565--568 (1968; Zbl 0174.32803); translation from Dokl. Akad. Nauk SSSR 180, 9--11 (1968)]. The second is that minimal prime ideals in reduced rings consist of zero-divisors, which allows one to guarantee that such subdirect products of domains can be chosen with the additional condition that any zero-divisor has a zero component. Sussman's original argument is now sufficient to complete the proof. Using methods from \textit{J. A. Fraser} and \textit{W. K. Nicholson}'s paper [Math. Japon. 34, No. 5, 715--725 (1989; Zbl 0688.16024)], the authors of this paper are able to do more, by giving an alternative characterization of when zero-divisors have a zero component in such decompositions. Finally, in Section 6 the authors give multiple examples of explicit collections of identities which demonstrate that the class of reduced Rickart rings forms a variety (in the sense of universal algebra). For example, they show in Theorem 6.2 that a unital ring \(R\) is a reduced Rickart ring if and only if it admits an unary operation \('\) satisfying the identities \(xx'=0=x'x\), \(x'+x''=1\), and \((xy)'=x'+y'-x'y'\). There were a few minor errors in the statements of some results. These have been fixed in a subsequent corrigendum of the authors, but we mention them here for clarity. In Proposition 2.4(f), the righthand side should be \((a''b)'\), and in the proof the reference to (e) should be to (f). In Section 6, the reader should be aware that the authors are now implicitly suspending their assumption that all rings are nonzero. The statement of Theorem 6.5 is incorrect as given, but can be rectified by adding the axiom \(0^{\circ}=0\). The hole in the proof comes when the authors assert \(x''=x^{\circ}\), which is equivalent to showing \((1-x^{\circ})^{\circ}=1-x^{\circ}\). With the axiom \(0^{\circ}=0\), in conjunction with the other axioms from Theorem 6.5, the authors provide the following computation, which fills the hole: First, \[ 1-x^{\circ} = (1-x^{\circ})(1-x^{\circ})^{\circ}=(1-x^{\circ})^{\circ}-x^{\circ}(1-x^{\circ})^{\circ}, \] which is equal to \[ (1-x^{\circ})^{\circ}-(x(1-x^{\circ}))^{\circ} = (1-x^{\circ})^{\circ}-(x-xx^{\circ})^{\circ}, \] which finally simplifies to \[ (1-x^{\circ})^{\circ}-0^{\circ}=(1-x^{\circ})^{\circ}. \] It appears that this computation has the added benefit that the axiom \(x'+x''=1\) in Theorem 6.2 can be weakened to \(0'=1\). It also shows that the first of Cornish's axioms listed in the displayed equation (6.2) is redundant (but the existence of \(1\) should be posited).