A graph theoretic approach to graded identities for matrices. (Q1939326): Difference between revisions

Let \(M_k(\mathbb C)\) be the algebra of \(k\times k\) matrices with complex entries and let \(G\) be an arbitrary group of order \(k\). The embedding of \(G\) in the symmetric group \(S_k\) via the regular representation and of \(S_k\) in \(M_k(\mathbb C)\) in the usual way induces a natural \(G\)-grading on \(M_k(\mathbb C)\) which the authors call a crossed-product grading. This is the elementary grading defined by any permutation \((g_1,g_2,\dots,g_k)\) of the elements of \(G\). In the paper under review the authors study the \(G\)-graded polynomial identities (with coefficients in \(\mathbb Q\)) for \(M_k(\mathbb C)\) equipped with the crossed-product \(G\)-grading. They associate a directed labeled graph to each multilinear monomial in the free associative \(G\)-graded algebra and describe the multilinear \(G\)-graded polynomial identities for \(M_k(\mathbb C)\) in the language of these graphs. This approach gives new proofs of a result of \textit{Yu. Bahturin} and \textit{V. Drensky} [Linear Algebra Appl. 357, No. 1-3, 15-34 (2002; Zbl 1019.16011)] on the generators of the T-ideal of the graded identities of matrices with elementary grading and of the Amitsur-Levitsky theorem. The most important new result is the asymptotic formula for the sequence of \(G\)-graded codimensions of \(M_k(\mathbb C)\): \[ c_n^G(M_k(\mathbb C))\sim {k^{k/2+1}\over(4\pi)^{(k-1)/2}}n^{-(k-1)/2}k^{2n},\quad n\to\infty. \] In particular, the authors show that the sequence \(c_n^G(M_k(\mathbb C))\), \(n=0,1,2,\dots\), does not depend on the group \(G\). For \(k=2\) (and the cyclic group \(G\) of order 2) the explicit formula for the codimension sequence is \[ c_n^G(M_2(\mathbb C))={2n+1\choose n}-2^n+1\sim{1\over\sqrt{\pi}}n^{-1/2}2^{2n+1}. \] The proof uses essentially asymptotic results from \textit{L. B. Richmond} and \textit{J. Shallit} [Electron. J. Comb. 16, No. 1, Research Paper R72 (2009; Zbl 1191.68479)]. Reviewer's remark: In the proofs the authors use that the field of complex numbers contains an arbitrary number of algebraically independent over \(\mathbb Q\) elements. Although not mentioned in the paper, the results hold for the matrix algebra \(M_k(F)\) over any field \(F\) of characteristic 0. It is sufficient to replace \(\mathbb Q\) with \(F\) and \(M_k(\mathbb C)\) with the generic \(G\)-graded \(k\times k\) matrix algebra over \(F\).