A tail empirical process approach to some nonstandard laws of the iterated logarithm (Q2640223)

Let \(X_ 1,X_ 2,..\). be independent and identically distributed random variables with non-degenerate distribution F. If the variance of \(X_ 1\) is finite (say \(\sigma^ 2)\), the law of the iterated logarithm states that with probability one \(\limsup_{n\to \infty}(\sum^{n}_{i=1}X_ i/\sqrt{2 \log \log n})=\sigma.\) Recently several non-standard LILs have been shown [e.g. the authors and \textit{E. Haeusler}, Bull. Sci. Math., II. Sér. 114, No.1, 61-95 (1990; Zbl 0698.62023) or the authors in the paper reviewed above]. Just as \(\sigma\) in the usual LIL may be computed as the extreme value of a functional over the Strassen set \((K=\{f\) absolutely continuous on [0,1] with \(f(0)=0\) and \(\int^{\infty}_{0}(\dot f(u))^ 2\leq 1\})\), so may the corresponding constants be found in the case of nonstandard limiting compact sets K, too. By solving the corresponding extremal problem on K, the authors show how these constants may be generated. Several examples of classical statistics with non-standard LIL behaviour are given.