Limit theorems for nonnegative independent random variables with truncation (Q2341928)

The paper concerns the asymptotic behaviour of sums of random variables with distributions \(P\{0\leq X<\infty \}=1\), \(P\{X>x\}\asymp x^{-\alpha}\) illustrated by the generalized St. Petersburg and the generalized Feller games. \(\{X_i\}_{i=1}^n\) is a sequence of i.i.d.\ random variables, \(\{c_n\}_{n=1}^{\infty}\subset {\mathbb R}^+\) a sequence of positive real numbers, \(\{X_i^{(c_n)}\}_{i=1}^n\) a sequence of truncated random variables constructed from \(\{X_i\}_{i=1}^n\) for \(n\in \mathbb {N}\). Putting \(S_n^{(c_n)}:=\sum_{i=1}^nX_i^{(c_n)}\), the asymptotic behaviour of \(S_n^{(c_n)}\) is investigated when \(X_i^{(c_n)}\) are specified by the above distribution. Let \(\{\ell_n\}\subset \mathbb N\) be a sequence of positive integers (\(\ell_n\) denotes the number of games, \(\lim \ell_n=\infty \)). The paper generalizes the results of \textit{L. Györfi} and \textit{P. Kevei} [Period. Math. Hung. 62, No. 1, 13--37 (2011; Zbl 1240.60011)] when the game is realized \(\ell_n\) times instead of \(n\) and studies the behaviour of the above sum with respect not only to \(\{c_n\}\) but also \(\alpha >0\) and \(\{\ell_n\}\).