Varieties of special Jordan algebras of almost polynomial growth (Q2419479)

One of the most important and studied numerical invariants of ideals of identities is their codimension sequence. The notion was introduced by \textit{A. Regev} [Isr. J. Math. 11, 131--152 (1972; Zbl 0249.16007)]. Regev proved that if \(R\) is a PI algebra then its codimension sequence \(c_n(R)\le (d-1)^{2n}\) where \(d\) is the degree of an identity satisfied by \(R\). He applied this important theorem to proving that the tensor product of two PI algebras is once again a PI algebra. The codimension sequence turned out to be quite elusive to compute explicitly, and its exact values are known for very few algebras. Clearly when one speaks about codimensions one considers the multilinear identities of the algebra in question, thus one considers algebras over fields of characteristic 0. The paper under review studies the codimension sequence of special Jordan algebras. The main result of the paper is that if \(J\) is a finite dimensional special Jordan algebra then its codimension sequence is bounded by a polynomial function if and only if \(UJ_2\) does not belong to the variety generated by \(J\). Here \(UJ_2\) is the Jordan algebra of the \(2\times 2\) upper triangular matrices. It follows from that result that \(UJ_2\) is the only finite dimensional special Jordan algebra that generates a variety of almost polynomial growth. (This means it generates a variety of exponential growth but every proper subvariety is of polynomial growth.) Another interesting corollary of the above theorem is that there does not exist a variety of special Jordan algebras generated by a finite dimensional algebra whose growth is faster than polynomial but slower that exponential.