Doubly stochastic operators with zero entropy (Q1738221)

The paper is concerned with the study of doubly stochastic operators with zero entropy on $L^{p}$-spaces. Recall that if $\mu $ is a probability measure on a measurable space $(X,\Sigma )$, a linear operator $T$ on $L^{p}(X,\Sigma,\mu )$ ($1\le p<+\infty$) is called \textit{doubly stochastic} if $Tf\ge 0$ for every $f\in L^{p}(\mu )$ with $f\ge 0$, $T{1}={1}$, and $\int Tf\,d\mu =\int f\,d\mu $ for every $f\in L^{p}( \mu )$. Following \textit{T. Downarowicz} and \textit{B. Frej} [Ergodic Theory Dyn. Syst. 25, No. 2, 455--481 (2005; Zbl 1088.47006)], the authors define a notion of entropy of doubly stochastic operators, and generalize a classical theorem of Rokhlin by showing that doubly stochastic operators with zero entropy form a comeager subset of the set of doubly stochastic operators, for the SOT topology and for the norm topology on $L^{1}(\mu )$. Then they provide a version of Kushnirenko's theorem (that a probability-preserving transformation has discrete spectrum if and only if it is null, i.e., has zero sequence entropy along every sequence of positive integers): the discrete analogue of this statement does not hold true for doubly stochastic operators, but one can prove a related statement relying on the Jacobs-de Leeuw-Glicksberg decomposition of $L^{p}(\mu )$ associated with a doubly stochastic operator $T$. The authors also generalize to this context the Halmos-von Neumann theorem on representation of maps with discrete spectrum as group rotations.