On asymptotic behavior of almost surely extreme values of independent random variables

Abstract

The article studies the almost surely asymptotics of extreme values $\bar{\xi}_n = \max_{1\leq i \leq n} \xi_i$, where $ \xi , \xi_1 , \xi_2 , \ldots$ are discrete identically distributed random variables. One of the main results on this topic is related to the law of the iterated logarithm for the lim sup (LIL) and a law of the triple logarithm for the lim inf (LTL). But, taking into account the specifics of the discrete case, necessary and sufficient conditions are established for $\mathbf{P}(\bar{\xi}_n = a_n +l \quad\mbox{infinitely often} ) =1$, where $a_n$ is some given increasing sequence of integers and $l$ is a fixed integer. Note that in the case of discrete random variables whose distribution tails are close to the tails of the Poisson distribution or fall off even faster, Theorems 1-3 of this article are significantly more informative than LIL and LTL. For the geometric distribution (or discrete random variables whose tails fall off even more slowly), the results are given in Theorem 3 and Theorem B, which are an important complement to LIL and LTL.