scientific article; zbMATH DE number 3907438
[For the entire collection see Zbl 0562.00003.] Two types of convergence of random variables are considered in Rényi spaces (see the following review, Zbl 0568.60002): (*) \(\xi_ n\to \xi\) if \(P(| \xi_ n-\xi | >\epsilon | B)\to 0\) for each \(\epsilon >0\) and \(B\in {\mathcal B}\); (**) \(\xi_ n\to \xi\) if \(P(| \xi_ n-\xi | >\epsilon | B_ n)\to 0\) for each sequence \(B_ n\) in \({\mathcal B}\) such that \(P(B_ n| B)\to 1\) for some \(B\in {\mathcal B}\). Convergence (**) implies convergence (*), but not conversely. In Rényi spaces generated by a family \(\{\mu_{\alpha}\}\) of positive measures, convergence (*) is equivalent to the condition (*') \(\xi_ n\to \xi\) if \(\mu_{\alpha}((| \xi_ n-\xi | >\epsilon)\cap B)\to 0\) for each \(B\in {\mathcal B}\) such that \(\mu_{\alpha}(B)<\infty.\) It is proved in the paper that convergence (**) of random variables \(\xi_ n\) implies the respective convergence of the corresponding Rényi distribution functions \([F_{\xi_ n}]\). On the other hand, an example is given that this is not true for convergence (*), even in the case where the given Rényi space is generated by one measure.
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3684881)