Uniform sets for infinite measure-preserving systems (Q393423)

Let \(T\) be a measure-preserving ergodic automorphism of a non-atomic Lebesgue space \((Y, \mathcal C, \nu)\). When \(\nu\) is a probability measure, \textit{R. I. Jewett} [J. Math. Mech. 19, 717--729 (1970; Zbl 0192.40601)] for \(T\) weakly mixing and \textit{W. Krieger} [in: Proc. 6th Berkeley Sympos. math. Statist. Probab., Univ. Calif. 1970, 2, 327--346 (1972; Zbl 0262.28013)] for the general case proved that \((Y, \mathcal C, \nu, T)\) is measure-theoretically isomorphic to a strictly ergodic dynamical system. The result was extended by \textit{B. Weiss} who proved [Bull. Am. Math. Soc., New Ser. 13, 143--146 (1985; Zbl 0615.28012)] that every measure-preserving ergodic action of a commutative group has a strictly ergodic model. In this paper, the author extends Jewett-Krieger theorem to infinite-measure dynamical systems: he shows that, when \((Y, \mathcal C, \nu)\) is a Lebesgue space with an infinite measure \(\nu\), the system is isomorphic to a minimal homeomorphism on a locally compact metric space which admits a unique (up to scaling) invariant Radon measure. The notion of uniform set, which plays a central role in the proof of B. Weiss, is formalized for the infinite-measure case as follows: If \(C\) and \(K\) are sets of positive finite measure, \(C\) is said to be \textit{uniform} relative to \(K\) if, for every \(\varepsilon > 0\), there exists \(m \in \;N\) such that \(\sum_{k=0}^{N-1}T^k 1_K (y) \geq m\) implies \(\displaystyle \left|{\sum_{k=0}^{N-1}T^k 1_C(y) \over \sum_{k=0}^{N-1}T^k 1_K (y)}- {\nu(\mathcal C) \over \nu(K)}\right| < \varepsilon\) for a.e. \(y \in K\). All partitions considered in the paper are finite and have a unique atom of infinite measure. A finite partition \(\alpha\) of \(Y\) is said to be uniform if all the atoms of finite measure in \(\bigcup_{n \in \;N} \, \bigvee_{k=-n}^{n-1} T^{-k} \alpha\) are uniform relatively to the complement of the unique set of infinite measure in \(\alpha\). Uniform partitions give rise to almost minimal factors admitting a unique invariant Radon measure up to scaling. The main theorem is based on the existence of a refining sequence of uniform partitions in the system \((Y, \mathcal C, \nu, T)\) which generate the \(\sigma\)-algebra \(\mathcal C\). Finally the author obtains a Bratelli-Vershik representation of Hajian-Kakutani's transformation with an invariant Radon measure which is unique up to scaling.