Geometry and dynamics of admissible metrics in measure spaces (Q1939364)

This paper continues an investigation by the first author [St. Petersbg. Math. J. 23, No. 1, 75--91 (2012); translation from Algebra Anal. 23, No. 1, 111--135 (2011; Zbl 1251.37009)] using properties of sequences of metrics in measure spaces to study ergodic theory. Given a measurable action of a countable group \(G\) on a standard Lebesgue space \((X,\mathcal{B})\) with a continuous measure, a measurable (semi-)metric \(\rho\) that is admissible (that is, has a measurable subset of full measure which is separable with respect to \(\rho\)), and a fixed sequence \((A_n)\) of finite subsets \(A_n\subset G\), write \(\rho^{n}(x,y)=| A_n|^{-1}\sum_{g\in A_n}\rho(gx,gy)\). The goal is to study the asymptotics of this sequence of (semi-)metrics along good sequences of subsets like Følner sequences, and to deduce measurable invariants. A norm is defined on the cone of admissible metrics and the notion of \(\epsilon\)-entropy of the measure space is combined with the averaged metrics to give a new discreteness criterion for the spectrum of an ergodic measure-preserving transformation: the spectrum is discrete if and only if the \(\epsilon\)-entropy of the averages of some (and hence of any) admissible metric is uniformly bounded. In contrast to \textit{A. G. Kushnirenko}'s result [Russ. Math. Surv. 22, No. 5, 53--61 (1967); translation from Usp. Mat. Nauk 22, No. 5 (137), 57--65 (1967; Zbl 0169.46101)] relating discrete spectrum to sequence entropies over all sequences, this characterization only requires an asymptotic calculation for a single metric.