Convex geometry of max-stable distributions (Q1003326)

Representations for max-stable distributions are considered in terms of random sets, e.g., it is shown that if a random vector \(\xi\in \mathbb R^d\) is max stable with unit Fréchet marginals, then for any \(x=(x_1,\dots,x_d)\in\mathbb R^d\), \[ {\mathbf P}(\xi<x)=\exp(-ch({\mathbf E}\Delta_\eta,x^*)), \] where \(c\) is some constant, \(\Delta_\eta\) is a random compact set, \(K={\mathbf E}\Delta_\eta\) means its selection (Aumann) expectation, \(h(M,x)\) is the support function of a set \(M\), \(x^*=(1/x_1,\dots,1/x_d)\). Relations between spectral measures of max-stable laws and their dependency sets \(K\) are discussed. Geometrical interpretations for concepts of max-dependency and copula properties of max-stable random vectors are given.