Tail behavior of laws stable by random weighted mean (Q2431141)

Let \((W_n)_{n\geq 0}\) be the intrinsic martingale related to a supercritical branching random walk. Assuming that the martingale is uniformly integrable denote by \(W\) its \(L_1\)-limit. The present article provides two results concerning the tail behaviour of \(W\). Theorem 1.1 states that, under a side condition, the regular variation of \(P\{W_1>x\}\) is equivalent to the regular variation of \(P\{W>x\}\). This is a generalization of previously known results obtained by \textit{N. H. Bingham} and \textit{R. A. Doney} [Adv. Appl. Probab. 6, 711--731 (1974; Zbl 0297.60044)] and \textit{A. de Meye} [J. Appl. Probab. 19, 217--220 (1982; Zbl 0481.60077)] for Galton-Watson processes, by \textit{N. H. Bingham} and \textit{R. A. Doney} [Adv. Appl. Probab. 7, 66--82 (1975; Zbl 0308.60049)] for Crump-mode and Jirina processes and by \textit{A. Iksanov} and \textit{S. Polotskiy} [Theory Stoch. Process. 12, No.~28, Part~1--2, 38--54 (2006; Zbl 1142.60340)] for branching random walks. Following the approach developed in the first three cited papers the proof of Theorem 1.1 is based on using Abel-Tauberian theorems for Laplace transforms. Theorem 1.2 gives information on the asymptotic behaviour of \(EW1_{\{W>x\}}\) under the assumption that \(EW_11_{\{W_1>x\}}\) is slowly varying.