Combinatorial lemmas and applications to dynamics (Q1011403)

Suppose \((X,T)\) is a (topological or measure-theoretic) dynamical system. If \({\mathcal A}\) is a strictly increasing subsequence of \(\{ 0, 1, 2, \dots \}\), then the (topological or measure theoretic) sequence entropy of \(T\) is defined analogously to the usual entropy but using only inverse iterates \(T^{-k}\) where \(k\in {\mathcal A}\). In the topological case, the maximal pattern entropy \(h_{t}^*(T)\) of \(T\) is defined to be the supremum over all finite open covers \({\mathcal U}\) of the compact space \(X\) of the exponential growth rate of the maximum over all finite sequences of length \(n\), \(0\leq t_1<t_2 <\cdots <t_n\), of the minimal cardinality of any subcover of \(X\) in \(\bigvee_1^nT^{-t_i}{\mathcal U}\). For a measure theoretic dynamical system \((X, T, \mu)\) there is an analogous maximal pattern entropy \(h_\mu^*(T)\). It is shown that each of these two invariants is the supremum of the corresponding sequence entropies. In the topological case, the \(\mu\) is a \(T\)-invariant measure then \(h_\mu^*(T) \leq h_t^*(T)\), and if \(h_t(T)<\infty\) then it is equal to \(\log(k)\) for some positive integer \(k\). If \(X\) is zero-dimensional then all of the topological sequence entropies of \(T\) are \(0\) if and only if the sequence of sizes of maximal patterns for each open cover \({\mathcal U}\) has at most polynomial growth.