Colength sequences for Kemer algebras. (Q2575687)

Let \(\chi_n=\sum m_\lambda(A)\chi^\lambda\), \(\lambda\vdash n\), be the \(n\)-th cocharacter of the polynomial identities of the algebra \(A\) over a field of characteristic 0, decomposed as a sum of irreducible \(S_n\)-characters. One may consider the asymptotics of the colength \(l_n(A)=\sum m_\lambda(A)\) as a first approximation to the description of the multiplicities \(m_\lambda(A)\). The paper under review is a sequel to the important paper of the author [\textit{A. Berele}, J. Algebra 283, No. 2, 700-710 (2005; Zbl 1074.16014)] where he described the exact asymptotics of the colengths of \(k\times k\) matrix algebras. Now the author considers a similar problem for another class of important algebras, \(M_{k,\ell}\), which plays a crucial role in the structure theory of T-ideals developed by Kemer. The main result is that there exist positive constants \(\alpha\) and \(\beta\) such that \(\alpha n^K\leq l_n(M_{k,\ell})\leq\beta n^K\), where \(K={k^2+\ell^2\choose 2}+{2k\ell+1\choose 2}\). Similar bounds hold if instead of the ordinary cocharacter sequence one uses the pure and the mixed trace cocharacter sequence or the cocharacter sequence for central polynomials. The author gives several conjectures. One of them is that each of the considered colength sequences for \(M_{k,\ell}\) is of the form \(\alpha n^K+{\mathcal O}(n^{K-1})\), i.e. behaves as a polynomial of degree \(K\), which is the exact analogue of the \(k\times k\) matrix case. As the paper for \(l_n(M_k)\), the present paper is very well written and uses several different techniques from the theory of PI-algebras, invariant theory (also of supergroups), combinatorics of hook symmetric functions and representation theory of classical groups, and even evaluation of multiple integrals.