On existence of PI-exponents of unital algebras (Q779903)

Let \(A\) be an algebra over a field of characteristic 0. It is well known that the polynomial identities of \(A\) are determined by the multilinear ones. This justifies the interest in studying the multilinear identities satisfied by the algebras. One of the most important numerical invariants of a PI algebra (or rather of its T-ideal) is its codimension sequence \(c_n(A)\). It is defined as the dimension of the ``non-identities'' among the multilinear polynomials of degree \(n\). It was proved by \textit{A. Regev} [Isr. J. Math. 11, 131--152 (1972; Zbl 0249.16007)] that if \(A\) is associative then \(c_n(A)\le (d-1)^{2n}\) for every \(n\) where \(d\) stands for the degree of a polynomial identity satisfied by \(A\). The codimension sequence of an algebra is extremely difficult to compute; this fact lead to study its asymptotic behaviour. Amitsur conjectured that if \(A\) is an associative PI algebra then the limit \(\lim_{n\to\infty} \sqrt[n]{c_n(A)}\) exists and is an integer. The latter limit is the PI exponent \(\exp(A)\) of \(A\). \textit{A. Giambruno} and \textit{M. Zaicev} [Adv. Math. 140, No. 2, 145--155 (1998; Zbl 0920.16012); Adv. Math. 142, No. 2, 221--243 (1999; Zbl 0920.16013)] proved Amitsur's conjecture. Thus the associative case was settled. One can ask a similar question for nonassociative algebras as well. The answers are not that decisive as in the associative case: while it is known that for large and important classes of Lie and Jordan algebras it holds, there are counterexamples. There exist infinite dimensional Lie algebras whose PI exponent exists but is not an integer. Moreover there are examples where the PI exponent does not exist at all. The paper under review studies nonassociative unitary algebras. The authors construct a family of algebras \(T_\alpha\) indexed by the real numbers \(\alpha>2\) such that \(\liminf_{n\to\infty} \sqrt[n]{c_n(A)}=2\) for each \(\alpha\) but \(\alpha\le \limsup_{n\to\infty}\sqrt[n]{c_n(A)}\le \alpha+1\). It is clear that the PI exponent does not exist for either of these algebras. These are the first examples of unitary algebras without PI exponent.