Strong laws for \(L_p\)-norms of empirical and related processes (Q5933701)

Let \(X_1,X_2,\dots\) be i.i.d. random variables with the distribution function \(F\). The corresponding order statistics of the first \(n\) observations will be denoted by \(X_{1,n}\leq X_{2,n}\leq \dots\leq X_{n,n}\). As usual, the empirical process is denoted by \(\alpha_n(x)\), the quantile process by \(w_n(t)\). Furthermore the uniform empirical process \(e_n(t)\) and the uniform quantile process \(u_n(t)\) are considered, cf. \textit{G. R. Shorack} and \textit{J. A. Wellner} [``Empirical processes with applications to statistics'' (Wiley, 1986)] or \textit{G. R. Shorack} [``Probability for statisticians'' (2000; Zbl 0951.62005)]. In this paper the asymptotic properties of the weighted \(e_n(t),u_n(t),\alpha_n(x)\) and \(w_n(t)\) are investigated. Eight limsup-results and four liminf-results are obtained, e.g. for the \(L_p\)-norms (\(1\leq p <\infty\)) of the empirical and quantile processes. These results are proved by the help of combining theorems for sums of Banach space valued random variables with invariance principles. The references contain more than 40 bibliographical items.