A characterization of semigroup varieties of finite index on the ``forbidden divisors'' language (Q1910277)

A semigroup variety is called a variety of finite index if the nilpotency indexes of its nilpotent members are bounded by some positive integer. Such varieties have been characterized from several points of view in articles by \textit{M. Sapir} and \textit{E. Sukhanov} [Izv. Vyssh. Uchebn. Zaved., Mat. 1981, No. 4(227), 48-55 (1981; Zbl 0471.20042)] and \textit{A. Tishchenko} [ibid. 1990, No. 7(338), 79-83 (1990; Zbl 0728.20051)]. In the article under review, varieties of finite index are described in terms of so-called ``forbidden divisors''. To formulate the results of the article, we need notation for several semigroups. The first three of them are semigroups of \(2 \times 2\)-matrices (over any field): \(C=\{e_{11}+e_{22}, e_{12}, 0\}\), \(S_0=\{e_{11}, e_{12}+e_{21}, e_{12}, 0\}\), \(T=\{e_{11}, e_{22}, e_{12}, 0\}\). Here \(e_{ij}\) are matrix units and 0 is the zero matrix. Further, semigroups \(D_1\) and \(D_2\) are defined on the set \(\{a, f, p, q, r\}\), where \(p\), \(q\), \(r\) are left zeros, \(ax=p\), \(fx=x\) for all \(x\neq q\), and \(aq=r\) in both \(D_1\) and \(D_2\), and \(fq\) equals \(p\) in \(D_1\) and \(r\) in \(D_2\). Finally, semigroups \(D^*_1\) and \(D^*_2\) are antiisomorphic to \(D_1\) and \(D_2\) respectively. A Rees quotient of a subsemigroup of a semigroup \(X\) is called a Rees divisor of \(X\). The main result of the article is the following Theorem 1. A semigroup \(X\) belongs to a semigroup variety of finite index if and only if (i) \(X\) satisfies the identity \(x^n=x^{n+m}\) for some positive integers \(n\) and \(m\), (ii) none of the semigroups \(C\), \(S_0\) and \(T\) is a Rees divisor of \(X\), and (iii) none of the semigroups \(D_1\), \(D_2\), \(D^*_1\), and \(D^*_2\) divides \(X\). Theorem 1 immediately implies Corollary. A finite semigroup \(X\) belongs to a variety of finite index if and only if it satisfies the conditions (ii) and (iii) of Theorem 1. Results of the article by Sapir and Sukhanov mentioned above easily imply that the set \({\mathcal F}J\) of all finite semigroups belonging to some variety of finite index is a pseudovariety. The pseudovariety generated by a semigroup \(X\) (given by a pseudoidentity system \(\Sigma\)) is denoted by \(\text{pvar }X\) (respectively \(\text{pvar }\Sigma\)). Two concluding results of the article give characterizations of the pseudovariety \({\mathcal F}J\). Theorem 2. The following four pseudovarieties are all minimal non-\({\mathcal F}J\) pseudovarieties of semigroups: \(\text{pvar }C=\text{pvar} \{x^2=x^3,\;xy=yx\}\), \(\text{pvar } T=\text{pvar} \{x^2=x^3,\;xyx=x^2 y^2=y^2 x^2\}\), \(\text{pvar } D_1=\text{pvar }D_2=\text{pvar} \{x^2=x^3,\;xyx=xy^2\}\), and \(\text{pvar } D^*_1=\text{pvar } D^*_2=\text{pvar} \{x^2=x^3,\;xyx=y^2 x\}\). Theorem 3. \({\mathcal F} J = \text{pvar} \{(x^\omega yz^\omega)^{\omega + 1} = (x^\omega y)^{\omega + 1} z^\omega = x^\omega (yz^\omega)^{\omega + 1} = x^\omega yz^\omega\}\).