Intermediate sums and stochastic compactness of maxima (Q1890869)

Let \(X_ 1, X_ 2, \dots\) be a sequence of i.i.d. nondegenerate real r.v.'s and \(X_{1,n} \leq \cdots \leq X_{n,n}\) denote the order statistics based on \(X_ 1, \dots, X_ n\). The main result of the paper establishes a precise connection between the stochastic compactness of the sequence of maxima \(\{X_{n,n}\}\) (this means that it can be centered and normalized so that each subsequence has a further subsequence converging in distribution to a nondegenerate limit) with a similar property of the intermediate sums \[ I_ n (a,b;k_ n) = \sum_{i = \lceil ak_ n \rceil + 1}^{\lceil b k_ n \rceil} X_{n + 1 - i,n}, \] where \(0 < a < b\) and the \(k_ n\)'s are integers tending to infinity such that \(k_ n/n \to 0\) \((\lceil x \rceil\) is the smallest integer not smaller than \(x)\). The convergence in distribution of these sums has been studied by the authors [Ann. Probab. 22, No. 1, 145-159 (1994; Zbl 0793.60020)]. A result of \textit{L. de Haan} and \textit{S. I. Resnick} is used [ibid. 12, 588-608 (1984; Zbl 0538.60025)].