Nil restricted Lie algebras of oscillating intermediate growth (Q2236076)

Different versions of Burnside Problem ask what one can say about finitely generated periodic groups under additional assumptions. For associative algebras, Kurosh type problems ask similar questions about properties of finitely generated nil (or more generally, algebraic) algebras. In case of finitely generated Lie algebras, the periodicity is replaced by the condition that the adjoint mapping is nil. In particular, for Lie \(p\)-algebras one assumes that the \(p\)-mapping is nil. One of recent important directions in these areas is to study the growth of finitely generated periodic groups and nil algebras. The present paper is devoted to the construction of finitely generated nil restricted Lie algebras with extreme growth properties. In fact the research is motivated by a construction of groups of oscillating growth by \textit{M. Kassabov} and \textit{I. Pak} [Ann. Math. (2) 177, No. 3, 1113--1145 (2013; Zbl 1283.20027)] and a description of possible growth functions of finitely generated associative algebras by \textit{J. Bell} and \textit{E. Zelmanov} [``On the growth of algebras, semigroups, and hereditary languages'', Preprint, \url{arXiv:1907.01777}]. The author addresses both, the question of possible growth functions in case of Lie algebras, and the Kurosh problem, since the proposed examples of restricted Lie algebras have a nil \(p\)-mapping, which is an analogue of nillity for associative algebras or periodicity for groups, i.e., for any field of positive characteristic, construct a family of 3-generated restricted Lie algebras of intermediate oscillating growth. The author calls them Phoenix algebras because, for infinitely many periods of time, the algebra is ``almost dying'' by having a quasi-linear growth, i.e., the lower Gelfand-Kirillov dimension is one, more precisely, the growth is of type \(n(ln\cdots ln(n))^k\), \(q\)-times, where \(q \in\mathbb{N}, k > 0\) are constants. On the other hand, for infinitely many \(n\) the growth function has a rather fast intermediate behavior of type \(\exp(n/(\ln n)^{\lambda})\), \(\lambda\) being a constant determined by characteristic, for such periods the algebra is ``resuscitating''. Moreover, the growth function is bounded and oscillating between these two types of behavior. These restricted Lie algebras have a nil \(p\)-mapping, thus addressing the Kurosh problem as well.