Stationary self-similar extremal processes (Q1263870): Difference between revisions

Let \((\xi_ k)^{\infty}_{k=-\infty}\) be a stationary sequence of random variables, and, for \(A\subset R\), let \(M_ n(A):=\bigvee_{k/n\in A}\gamma_ n(\xi_ k)\) where \(\gamma_ n\) is an affine transformation of R (has the form \(a_ n\cdot +b_ n\), \(a_ n>0\), \(b_ n\in R)\). Then \(M_ n\) is a random sup measure, that is, \[ M_ n(\cup_{\alpha}G_{\alpha})=\bigvee_{\alpha}M_ n(G_{\alpha}) \] for arbitrary collections of open sets \(G_{\alpha}\). We show that the possible limiting random sup measures for such sequences \((M_ n)\) are those which are stationary \((M(\cdot +b)=_ dM\) for \(b\in R)\) and self-similar \((M(a\cdot)=_ d\delta^{\log \alpha}(M)\) for \(a>0\), where \(\delta\) is an affine transformation of R). By applying simple transformations, we need only study stationary M such that \(M(a\cdot)=_ daM\) for \(a>0.\) We show that these processes retain some but not all of the properties of the classical case. In particular, we display a nontrivial example such that \(t\mapsto M(0,t]\) is continuous wp 1. The classical planar point process representation of extremal processes is a special case of the present approach, but is not adequate for describing all possible limits.