scientific article; zbMATH DE number 4092471

From MaRDI portal
Publication:3819780





Complete convergence and convergence rates results for sums of random variables have been generalized to the case of randomly indexed partial sums. The following result is obtained. Let \(X_ n\) be i.i.d. random variables and \(\nu_ n\) be nonnegative, integer-valued random variables. Suppose that \(\alpha >1/2\), \(\alpha\) \(r\geq 1\), E \(| X_ 1|^ t<\infty\) and that E \(X_ 1=0\) if \(r\geq 1\). If, for some \(\epsilon >0\), \(\beta >0\), \[ (1)\quad \sum n^{\alpha r-2}P(\nu_ n<n^{\beta})<\infty \quad and\quad (2)\quad \sum n^{\alpha r-2}P(\max_{k\leq \nu_ n}| X_ k| \geq \epsilon \nu_ n^{\alpha})<\infty, \] then there exists a number N such that for \(\delta\geq N\epsilon\); \[ \sum n^{\alpha r-2}P(| S_ n| \geq \delta \nu_ n^{\alpha})<\infty. \] This result generalizes the theorem of \textit{A. Gut} in Acta Math. Hung. 42, 225-232 (1983; Zbl 0536.60036). Note that this result holds if E \(| X_ 1|^ t<\infty\) for \(t>1/\alpha\) and E \(X_ 1=0\) if \(\alpha\leq 1\), and conditions (1) and (2) are satisfied.











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 Q3819780)