Topological classification of spaces of probability measures for co-analytic sets (Q1897210): Difference between revisions

The authors' introduction: ``The problem of investigating the topology of a space \(P_\beta(Y)\) of probability measures with compact supports for a noncompact space \(X\) has been posed by Fedorchuk. Let us recall that for a Tikhonov space \(Y\) \[ P_\beta(Y)= \{\mu\in P(\beta Y)\mid \text{supp} (\mu)\subset Y\}\subset P(\beta Y), \] where \(\beta Y\) is the Stone-Čech extension of \(Y\). It is well-known that the topology of the space \(P_\beta(Y)\) does not depend on a compactification of \(Y\), that is, for any \(X\)-compactification of \(Y\) the space \[ P(Y)= \{\mu\in P(X)\mid \text{supp} (\mu)\subset Y\}\subset P(X) \] is naturally homeomorphic to the space \(P_\beta(Y)\). In the present paper we prove that for a co-analytic infinite space \(Y\) there exist exactly five types of the space \(P_\beta(Y)\). We obtain this result when investigating the topology of the pair \((P(X),P(Y))\), where \(Y\) is a dense co-analytic set in a metric compactum \(X\)''.