Finite aperiodic semigroups with commuting idempotents and generalizations (Q1976622)

Given \(k\) partial bijections on a set \(X\) with \(n\) elements, consider the semigroup \(U\) they generate. The authors construct an aperiodic subsemigroup \(S(U)\) of the symmetric inverse semigroup \(I_{2n}\) which is in fact the disjoint union of the aperiodic Brandt ideal of all bijections of rank at most~1 with \(k+1\) elements which generate a zero semigroup. The main theorem states that any finite inverse semigroup which is divided by~\(S(U)\) is also divided by~\(U\). This result affords many corollaries such as that the smallest pseudovariety generated by finite inverse semigroups which contains all finite aperiodic semigroups with commuting idempotents is the pseudovariety of all finite semigroups with commuting idempotents. As a complement to this result, the authors observe that a construction of \textit{J.-C.~Birget, S.~W.~Margolis} and \textit{J. Rhodes} [Bull. Aust. Math. Soc. 41, No.~2, 161-184 (1990; Zbl 0692.20046)] yields that, for every pseudovariety \(\mathbf H\) of groups, the pseudovariety consisting of all finite semigroups with commuting idempotents whose subgroups lie in \(\mathbf H\) is generated by its semigroups of partial bijections. Let \(\mathbf R\) denote the pseudovariety of all finite \(\mathcal R\)-trivial semigroups. For a pseudovariety \(\mathbf V\) let \(\mathbf{EV}\) stand for the pseudovariety of all finite semigroups whose idempotents generate a semigroup from~\(\mathbf V\). Answering a question posed by the reviewer, the authors go on to observe that, if \(\mathbf H\) is a proper non-trivial pseudovariety of groups and \(G\) is a finite group not in \(\mathbf H\), then the aperiodic semigroup \(S(G)\) belongs to \(\mathbf{ER}\) but not to the semidirect product \({\mathbf R}*{\mathbf H}\). More generally, for any pseudovariety \(\mathbf V\) containing the pseudovariety \(\mathbf{Sl}\) of semilattices and which is contained in the pseudovariety \(\mathbf{DA}\) of all finite aperiodic semigroups whose regular elements are idempotents, under the same hypotheses on \(\mathbf H\) and \(G\), \(S(G)\) lies in \(\mathbf{EV}\) but not in the Mal'cev product \({\mathbf V}m{\mathbf H}\). While this is an important paper which is generally well written, the last section contrasts with the rest of the paper in this respect. The proofs of Corollary 5.2 and Lemma~5.4 are obvious from the definitions in view of Lemma~5.1. In the proof of Theorem 5.3, one should read \(\mathbf H\) where \(\mathbf{V(H)}\) is found. The final remark states, without any references, that \textit{it is known} that \({\mathbf V}m{\mathbf G}=\mathbf{EV}\) holds for the pseudovariety \(\mathbf G\) of all finite groups whenever \(\mathbf{Sl}\subseteq{\mathbf V}\subseteq\mathbf{DA}\). While there are several published results which are instances of that equality, and which would deserve a reference, the reviewer would find it quite surprising if the equality holds in full generality.