On max-stable processes and the functional \(D\)-norm (Q385633): Difference between revisions

The work provides an ample and refined development of the extreme value theory for stochastic processes (functional EVT). A complex mathematical setup demanded by the study of the behavior of the extreme values related to stochastic processes was initiated by \textit{L. de Haan} [Ann. Probab. 12, 1194--1204 (1984; Zbl 0597.60050)]. The topic was detailed in a modern presentation by \textit{L. de Haan} and \textit{A. Ferreira} [Extreme value theory. An introduction. New York, NY: Springer (2006; Zbl 1101.62002)]. In this paper, the authors introduce an appropriate framework for extreme value theory in the space of continuous functions on compact intervals. A specific norm (called \(D\)-norm) on function spaces serves to understanding and characterizing the continuous max-stable processes on \([0,1]\). In this context, a kind of functional domain of attraction for stochastic processes is introduced and analyzed. Note that this new concept is more general than the usual one based on the convergence. The authors refer also to so-called sojourn time transformation and compare several types of convergence on functions spaces. Relevant examples (as the functional generalized Pareto distributions) contribute to an excellent consistency of the theory developed in this article.