On Banach support of an ergodic and quasi-invariant measure (Q808080)

Using the zero-one laws for the ergodic quasi-invariant measures, we discuss the Banach support of an ergodic and quasi-invariant measure. The results may be especially applied to the abstract Wiener space. As an application, we give a generalization of a result due to H. Sato. Let X be a locally convex space, \(\Omega =(X,{\mathcal B},\mu)\) a probability measure space. Let Z be a measurable linear subspace of full measure in X, P a norm on Z, and (Z,P) a Banach space. Then we call Z a Banach support of \(\mu\) if the injection (Z,P)\(\to X\) is continuous. We prove the following results: Theorem 1. Let X be a locally convex space, \(\Omega =(X,{\mathcal B},\mu)\) a convex-tight probability measure space, ergodic and quasi-invariant with respect to \({\mathfrak B}\), which is a subspace in X. Then \(\mu\) has a Banach support. - Theorem 2. Let X be a locally convex space, \(\Omega =(X,{\mathcal B},\mu)\) a convex-tight Radon probability measure space, ergodic and quasi-invariant with respect to \({\mathfrak B}\), where \({\mathfrak B}\) is a maximal quasi-invariant and quasi- continuous linear subspace of X, (Z,P) a Banach support of \(\mu\). We introduce a metric topology in \({\mathfrak B}\) by S-pseudo-distance \(\rho_ 1(h_ 1,h_ 2)=R_ 1(h_ 1-h_ 2),\) then the injection T: (\({\mathfrak B},\rho_ 1)\to (Z,P)\) is bounded. Hence, T is also continuous. - Theorem 3. Under the assumptions of Theorem 2 and the assumption that X is a separable Fréchet space, \(\mu\) has a separable Banach support Z, and the injection T: (\({\mathfrak B},\rho_ 1)\to Z\) is completely continuous. - As an application, our results may be used, especially, to the Gaussian measures.