On the normalizing groupoids and the commensurability groupoids for inclusions of factors associated to ergodic equivalence relations-subrelations (Q860782): Difference between revisions

Consider a discrete measured equivalence relation \(\mathcal{R}\) on a standard probability space \((X,\Sigma,\mu),\) which is quasi-invariant with respect to \(\mu,\) and a normalized Borel 2-cocycle \(\omega\) from \(\mathcal{R}\) into \(S^{1}.\) A classical construction due to \textit{J. Feldman} and \textit{C. C. Moore} [Trans. Am. Math. Soc. 234, 325--359 (1977; Zbl 0369.22010)] allows us to associate to \(\mathcal{R}\) and \(\omega\) a von Neumann \(W^{\star}\left( \mathcal{R},\omega\right) .\) Theorem 5.16 shows that an ergodic equivalence subrelation \(\mathcal{S}\) is normal in \(\mathcal{R}\) if and only if the corresponding \(W^{\star}\left( \mathcal{R},\omega\right) \) is generated by the normalizing grupoid of \(W^{\star}\left( \mathcal{S},\omega\right) .\) As a consequence, for every ergodic subrelation \(\mathcal{S}\) of \(\mathcal{R}\) there always exists the largest intermediate equivalence subrelation which contains \(\mathcal{S}\) as a normal subrelation. Then the authors introduce a concept of commensurability groupoid as a generalization of normality and prove that the commensurability grupoid \(C\mathcal{G}(B),\) of \(B=W^{\star }\left( \mathcal{S},\omega\right) \) in \(A=W^{\star}\left( \mathcal{R} ,\omega\right) ,\) generates \(A\) if and only if the inclusion \(B\subseteq A\) is discrete in the sense of \textit{M. Izumi, R. Longo} and \textit{S. Popa} [J. Funct. Anal. 155, No. 1, 25--63 (1998; Zbl 0915.46051)]. In fact, Theorem 7.10 states that the inclusion \(B\subseteq C\mathcal{G}(B)^{\prime\prime}\) is discrete and \(C\mathcal{G} (B)^{\prime\prime}\) is the largest among the intermediate subfactors \(M\) of \(B\subseteq A\) such that \(B\subseteq M\) is discrete.