On the codimension sequence of \(G\)-simple algebras. (Q279734): Difference between revisions

Let \(G\) be a finite group, \(F\) an algebraically closed field of characteristic 0, and \(A\) a \(G\)-graded algebra over \(F\). The paper under review studies the asymptotic behaviour of the sequence of the graded codimensions of \(A\). One of the most important numerical invariants of a PI algebra \(A\) is its codimension sequence \(c_n(A)\). It is defined as follows. Let \(T(A)\) be the T-ideal of \(A\) in the free associative algebra \(F\langle X\rangle\) of infinite rank, \(X=\{x_1,x_2,\ldots\}\), and let \(P_n\) be the vector space in \(F\langle X\rangle\) spanned by all multilinear monomials of degree \(n\) in \(x_1,\ldots,x_n\). Then \(c_n(A)=\dim P_n/(P_n\cap T(A))\). A fundamental theorem due to Regev gives that if \(A\) satisfies a polynomial identity then \(c_n(A)\leq d^n\) for some constant \(d\). In other words \(c_n(A)\) grows at most exponentially. The exact values of \(c_n(A)\) are known in very few instances thus the asymptotic behaviour of the codimensions of an algebra is of significant interest. Another fundamental result, due to Giambruno and Zaicev, yields that the limit \(\root n\of{c_n(A)}\) always exists and is an integer. This integer is the PI exponent \(\exp(A)\) of \(A\). Also Berele and Regev proved that the codimension sequence of a PI algebra \(A\) with 1 behaves asymptotically like \(\alpha n^bd^n\). (The sequences \(a_n\) and \(b_n\) have the same asymptotic behaviour if \(\lim_{n\to\infty}a_n/b_n\) exists and equals 1.) Here \(\alpha>0\), \(b\) is an integer or half integer, and \(d=\exp(A)\). The graded PI exponent of a graded algebra is defined in analogy with the ungraded case. A deep result of Aljadeff, Giambruno and La Mattina guarantees that the graded PI exponent exists and as in the ungraded case is always an integer. They conjectured that the asymptotic behaviour of the graded codimensions is similar to that of the ungraded case, a conjecture that was confirmed for affine unitary algebras by the second named author of the present paper. We recall that when dealing with gradings on matrix algebras (and their subalgebras) two types of gradings are of importance. The first is the elementary grading: in it all matrix units are homogeneous. The fine gradings are the ones where each homogeneous component is of dimension \(\leq 1\). The paper under review considers mainly elementary gradings. The first main result of the paper, Theorem A, describes the exact asymptotic behaviour of the graded codimensions of the matrix algebras \(M_m(F)\) of order \(m\) equipped with an elementary \(G\)-grading. The authors prove that the graded codimensions behave asymptotically as \(\alpha n^bd^n\) where \(d=m^2=\dim M_m(F)\), and \(b=(1-\dim A_e)/2\) where \(A_e\) is the neutral component in the grading. The constant \(\alpha\) is computed explicitly as well for the case of matrix algebras. Furthermore the authors prove a similar result (only without giving the exact value of \(\alpha\)) for finite dimensional \(G\)-graded algebras that are simple as graded ones (that is algebras having no non-trivial homogeneous ideals).