The LIL for \(U\)-statistics in Hilbert spaces (Q939134)

This paper generalizes the results for real valued \(U\)-statistics of \textit{R. Adamczak} and \textit{R. Latała} [Ann. Probab. 36, 1023--1058 (2008; Zbl 1144.60015)] to obtain necessary and sufficient conditions for the bounded law of the iterated logarithm for \(U\)-statistics taking values in separable Hilbert spaces. The conditions are expressed in terms of the \(U\)-statistic kernel and the distribution of the underlying independent and identically distributed sequence. To achieve these results sharp moment and tail probability estimates are developed for canonical Hilbert space valued \(U\)-statistics of arbitrary order.