Self-similarity of ergodic measure preserving transformations (Q1202216)

The authors consider the so-called induced transformation \(T_ A\) of a measurable set \(A\) for a dynamical system \((X,T)\), where a probability measure \(P\) on \(X\) is given. \(T_ A(x)\) is the \(n_ A(x)\)-times composition of \(T\) applied to \(x\), where \(n_ A(x)\) denotes the first return time of \(x\) to the set \(A\). If \(T\) and \(T_ A\) are isomorphic, i.e., there exists an invertible measure-preserving map \(\phi\) from \((X,P)\) onto \((A,P_ A)\) (note that \(P_ A(B)=P(B\cap A)/P(A))\) such that \(T_ A\phi x=\phi T(x)\) for almost all \(x\) in \(X\), then the set \[ R(T)=\bigl\{ P(A)\mid T_ A\text{ isomorphic to }T\bigr\} \] has some interesting properties: (1) \(R(T)\) is a semigroup of (0,1] with respect to ordinary multiplication. (2) For ergodic measure-preserving transformations \(T\) with positive finite entropy \(R(T)=\{1\}\). (3) Bernoulli transformations \(T\) with infinite entropy satisfy \(R(T)=(0,1]\). (4) For any \(0<\lambda<1\) there exists an ergodic measure-preserving transformation \(T\) with zero entropy such that \(\lambda\in R(T)\). The first title word of this paper does not match to the content of the paper.