Measure concentration and the weak Pinsker property (Q1720271): Difference between revisions

An automorphism \(T\) on a probability space \((X,\mu)\) has the weak Pinsker property if it has a splitting into a direct product of a Bernoulli shift and an automorphism of entropy at most \(\varepsilon\), for any \(\varepsilon>0\). The main result in the present paper is the verification of a conjecture by \textit{J.-P. Thouvenot} [Isr. J. Math. 21, 208--214 (1975; Zbl 0329.28009)] that all ergodic automorphisms have the weak Pinsker property. In fact, the author proves a stronger result: a corresponding statement for free and ergodic measure-preserving actions of arbitrary countable and amenable groups. A key part of the proof of this result is the repeated application of a new result on concentration of measure. Given a finite alphabet \(A\), consider the corresponding product space \(A^n\) endowed with the Hamming metric. The author shows that a probability measure \(\mu\) on \(A^n\) may be represented as a mixture of measures in which most of the weight is on measures that have strong concentration properties, and the number of terms in the mixture may be controlled in terms of the Shannon entropy of the underlying measures. The author also discusses connections between this result and some important structural results in additive and extremal combinatorics. The paper concludes with a discussion of consequences of the main results and some related open problems.