Localization for infinite-dimensional hyperbolic measures (Q498193)

A Radon probability measure \(\mu\) on a locally convex space \(E\) is called \(\alpha\)-concave (\( - \infty \leq \alpha \leq 1\)) if, for any compact subsets \(A, B \subset E\) and \(0<t<1\), we have \[ \mu((1-t)A+tB)) \geq [(1-t) \mu(A)^{\alpha} +t \mu(B)^{\alpha}]^{\frac{1}{\alpha}}\,. \] In this paper, the authors extend some results about \(\alpha\)-concave measures from finite-dimensional to infinite-dimensional spaces. The main results are: {\parindent = 0.6 cm \begin{itemize}\item[I.] \((E, \mu)\) is a complete locally convex space with \(\mu\) a finite \(\alpha\)-concave measure, and \(u, v\) two lsc \(\mu\)-integrable functions with \(\int_{E} u \, d \mu >0\), \(\int_{E} v \,d \mu>0\). Then there are \(a, b \in E\) and a finite \(\alpha\)-concave measure \(\nu\) concentrated on \(\Delta= [a, b]\) such that \(\int_{E} u \,d \nu>0\), \(\int_{E} v \,d \nu>0\). \item[II.] Let \((E, \mu)\) be a complete locally convex space with a probability \(\alpha\)-concave measure \(\mu\) concentrated on a closed convex set \(F\) and with \(\mu^{*}\) the associated outer measure. Then, for any Borel set \(A \subset F\) and any \(\delta \in [0, 1]\), we have \(\mu(A) \geq [ \delta \mu^{*}(A_{\delta})^{\alpha} + (1- \delta)]^{\frac{1}{\alpha}}\) (here, \(A_{\delta} = \{ x \in A: \forall \Delta \subset F\,(x \in \Delta \Rightarrow m_{\Delta}(A) \geq 1- \delta) \} \), where \(m_{\Delta}\) is normalized Lebesgue measure on \(\Delta) \). \end{itemize}} Some additional comments are made.