Indecomposability of treed equivalence relations (Q1124002)

The author gives a rigorous definition of a ``treed'' equivalence relation, which roughly speaking is an equivalence relation together with a measurably varying tree structure on each equivalence class. This idea, first suggested by Alain Connes in the late 1970's was studied in a special case by the author in Ergodic Theory Dyn. Syst. (to appear). In that example the treed equivalence relation comes from considering the orbits of the action of a finitely generated free group F acting freely on a measure space. The tree structure is then obtained on each orbit by saying that two points x and y are adjacent if there is a generator \(g\in F\) such that \(gx=y\) or \(gy=x.\) The main theorem of the author, an analogue of a theorem of \textit{R. J. Zimmer} [Ann. Math., II. Ser. 118, 9-19 (1983; Zbl 0545.22010)], is the following: Let R be a countable, non-amenable, measure preserving, properly ergodic equivalence relation on a finite measure space M. Assume that (M,R) admits a treeing. Then there do not exist equivalence relations \(R_ 1\) and \(R_ 2\) on measure spaces \(M_ 1\) and \(M_ 2\) such that R is stably isomorphic to \(R_ 1\times R_ 2\).