Codimensions of star-algebras and low exponential growth (Q2204397): Difference between revisions

The paper under review studies algebras with involution and their polynomial identities. Let \(F\) be a field of characteristic 0, and assume \(A\) is an \(F\)-algebra with involution \(*\). As in the case of ordinary identities, one of the most important numerical invariant of the ideals of \(*\)-identities of \(A\) is its codimension sequence \(c_n^*(A)\). The first author and \textit{A. Regev} [J. Pure Appl. Algebra 35, 133--149 (1985; Zbl 0563.16008)] proved that if \(A\) satisfies an ordinary PI then its \(*\)-codimension sequence is exponentially bounded, just as in the ordinary case. We recall here that computing the exact codimensions of an algebra (with or without involution) is an extremely difficult task, and it has been achieved in very few occasions. That is why one studies the asymptotic behaviour of the codimension sequence. \textit{E. Aljadeff} et al. [Proc. Am. Math. Soc. 145, No. 5, 1843--1857 (2017; Zbl 1406.16021)] proved that every variety of \(*\)-PI algebras is generated by the Grassmann envelope of a suitable finite dimensional superalgebra equipped with a superinvolution. By using this result, the first author et al. [J. Algebra 469, 302--322 (2017; Zbl 1353.16022)], like in the ordinary case, proved that if \(A\) is PI then its \(*\)-PI exponent exists and is a non-negative integer. It is also known that there exist constants \(C_1>0\), \(C_2\), \(t\) and \(s\) such that \(C_1 n^t \exp^*(A) \le c_n^*(A) \le C_2 n^s \exp^*(A)\). The paper under review captures the growth of the \(*\)-codimension sequence of \(A\) in more precise terms. The authors prove that if \(A\) satisfies an ordinary identity then its \(*\)-codimension sequence is eventually non-decreasing. Additionally they investigate and give a complete description of the varieties of \(*\)-algebras with slow \(*\)-codimension growth. They prove that there are only two varieties of \(*\)-algebras of almost polynomial growth, and also provide generating algebras of these two varieties. Furthermore, they characterize the varieties of \(*\)-algebras whose \(*\)-PI exponent equals 2. In fact, they describe also the varieties where the \(*\)-PI exponent is larger than 2, and as the description is in terms of excluding a list of 9 concrete algebras, the result concerning exponent 2 follows.