On free spectra of finite completely regular semigroups and monoids. (Q1025072)

Given a variety \(\mathcal V\) of semigroups, for each positive integer \(n\) denote by \(f_n(\mathcal V)\) the cardinality of the free semigroup of rank \(n\) in the variety. The `free spectrum' of the variety is the sequence so produced. The free spectrum \(f_n(S)\) of a finite semigroup \(S\) is the free spectrum of the variety it generates, and therefore consists of positive integers. The semigroup has `log-exponential' free spectrum if eventually \(f_n(S)\geq 2^{2^{\alpha n}}\) for some positive real number \(\alpha\). Otherwise the spectrum is `sub-log-exponential'. \textit{S. Seif} [J. Pure Appl. Algebra 212, No. 5, 1162-1174 (2008; Zbl 1138.20049)] has proved that if a finite monoid \(S\) has sub-log-exponential free spectrum, then (a) the only regular elements in the subsemigroup generated by its idempotents are themselves idempotent and (b) its subgroups are nilpotent. He has conjectured that the converse is also true. If \(S\) is a regular semigroup, Seif's theorem implies that if its spectrum is sub-log-exponential, then it is locally orthodox (that is, \(eSe\) is orthodox for each idempotent \(e\) of \(S\)). The essence of the paper under review is that for finite completely regular semigroups, the converse is true (assuming, once again, that the subgroups are nilpotent). (In fact, a stronger result is proven: the free spectrum is `small', that is, \(\log f_n(S)\) is of polynomial order.) Thus Seif's conjecture is valid for completely regular monoids. The proof of this theorem is a beautiful application of Polák's deep theorem on the lattice of completely regular semigroup varieties (and of earlier solutions to the word problem for various band varieties).