Asymptotics for statistical functionals of long-memory sequences (Q765881)

This paper establishes two general results that can be used to obtain the asymptotic behaviour for statistical functionals based on long memory sequences. The idea of quasi-Hadamard differentiability was introduced by \textit{E. Beutner} and \textit{H. Zähle} [J. Multivariate Anal. 101, No. 10, 2452--2463 (2010; Zbl 1213.62055)] to enable a modified functional delta method to be applied to a class of functionals. In general, to apply this technique a weak convergence result is required for the underlying empirical process with respect to a weighted-sup norm. The first result of this paper establishes such a result for the empirical process of a linear long memory sequence. This is then applied to obtain properties of certain L\,- and V-statistics. The second result enables the asymptotics for degenerate V-statistics to be developed when the kernel is only of locally bounded variation.