On the rate of complete convergence for weighted sums of arrays of Banach space valued random elements with application to moving average processes (Q1871221)

A sequence of random variables \(U_n\) is said to converge completely to a constant \(c\) if \(\sum P\{|U_n-c|>\varepsilon\}<\infty \) for all \(\varepsilon \). This concept was introduced by \textit{P. L. Hsu} and \textit{H. Robbins} [Proc. Nat. Acad. Sci. USA 33, 25-31 (1947; Zbl 0030.20101)] in connection with the Borel-Cantelli lemma and the law of large numbers. Results of the Hsu and Robbins paper have been generalized and extended in many directions. The main result of the paper under review says that if \((V_{nk})\) is an array of rowwise independent Banach valued random elements, \((a_{nk})\) is an array of constants with some technical conditions and \(S_n=\sum_k a_{nk}V_{nk}\) converges to 0 in probability, then \(\|S_n\|\) converges completely to 0. The rate of the convergence is given. If the Banach space has the geometric property of being of an appropriate stable type, then the condition that \(S_n\) converges in probability can be dropped. An application to mowing average processes is presented. The main theorem of the paper is a generalization, with simpler proof, of a result of \textit{T.-C. Hu, D. Li, A. Rosalsky} and \textit{A. Volodin} [Theory Probab. Appl. 47, 533-547 (2002)].