Extremes of subexponential Lévy-driven random fields in the Gumbel domain of attraction (Q2121642)

The authors study the extremal behaviour of a spatial Lévy-driven moving average given by \(X_v = \int_{\mathbb{R}^d} f(v-u) \Lambda(du), v \in \mathbb{R}^d,\) where \(\Lambda\) is an infinitely divisible, independently scattered random measure on \(\mathbb{R}^d,\) that is, a Lévy basis which is assumed to be stationary and isotropic, and \(f\) is an appropriate non-negative kernel function. Assuming the Lévy measure \(\rho\) of the basis \(\Lambda\) to have a subexponential right tail, which is furthermore in the maximum domain of attraction of the Gumbel distribution, and integrability of \(f\) and a couple of other assumptions, it is shown that \(P(\sup_{v \in B} X_v> x) \sim \rho((x, \infty)) C_B\) as \(x \rightarrow \infty,\) where \(\sim\) denotes asymptotic equivalence and \(C_B\) is a constant given in terms of the bounded and full dimensional index set \(B\) and the index sets of maximal and minimal values of \(f.\) It is also shown that, for a very general class of expanding index sets, the running supremum of the field, under a suitable scaling, converges to the Gumbel distribution.