Complete convergence and convergence rates for randomly indexed partial sums with an application to some first passage times (Q791984): Difference between revisions

There are given extensions of theorems on complete convergence for randomly indexed partial sums contained in the reviewer's note, Bull. Acad. Polon. Sci., Sér. Sci. Math. Astron. Phys. 20, 571-574 (1972; Zbl 0253.60057) and in a book by \textit{M. Csörgö} and \textit{P. Révész}, Strong approximations in probability and statistics. Akadémiai Kiadó, Budapest (1981), theorem 7.1.1, page 252. The main result is as follows. Theorem. Let \(\{X_ n,n\geq 1\}\) be a sequence of i.i.d. random variables, and let \(\{N_ n,n\geq 1\}\) be a sequence of nonnegative, integer valued random variables. a) Let \(\alpha\) \(r\geq 1\) and \(\alpha>1/2\). Suppose that \(E| X_ 1|^ r<\infty\) and that \(EX_ 1=0\) if \(r\geq 1\). If for \(\epsilon>0\) (1) \(\sum^{\infty}_{n=1}n^{\alpha r-2}P[| N_ n- Nn|>n\epsilon]<\infty\), where N is a positive random variable such that \(P[N\geq A]=1\) for some \(A>\epsilon>0\), then (2) \(\sum^{\infty}_{n=1}n^{\alpha r-2}P[| S_{N_ n}| \geq N_ n^{\alpha}\epsilon]<\infty.\) b) Let \(\alpha r=1\), \(\alpha>1/2\). Suppose that \(E| X_ 1|^ r \log^+| X_ 1|<\infty\) and that \(EX_ 1=0\) if \(1\leq r(<2)\). If (1) holds with N as above, then (2) holds. However, if (1) holds with \(P[A\leq N\leq B]=1\), where \(0<\epsilon<A\leq B<\infty\), then \(E| X_ 1|^ r<\infty\) and \(EX_ 1=0\) for \(r\geq 1\) imply that (2) holds. c) Let \(\alpha\) \(r\geq 1\) and \(\alpha>1/2\). Suppose that \(E| X_ 1|^ r<\infty\) and that \(EX_ 1=0\) if \(r\geq 1\). If (1) holds with \(P[N\leq B]=1\) for some \(B>0\), then \(\sum^{\infty}_{n=1}n^{\alpha r- 2}P[| S_{N_ n}|>n^{\alpha}\epsilon]<\infty.\) An application to some first passage times are given, too. Theorems 5.1 and 5.2 have to be modified slightly. The corrected results will be published by the author.