A characterization of the normal distribution using stationary max-stable processes

Abstract: Consider the max-stable process

e t a (t) = m a x_{i i n m a t h b b N} U_{i} m e^{l a n g l e X_{i}, t a n g l e - k a p p a (t)}

,

t i n m a t h b b R^{d}

, where

U_{i}, i i n m a t h b b N

are points of the Poisson process with intensity

u^{- 2} m d u

on

(0, i n f t y)

,

X_{i}

,

i i n m a t h b b N

, are independent copies of a random

d

-variate vector

X

(that are independent of the Poisson process), and

k a p p a : m a t h b b R^{d} o m a t h b b R

is a function. We show that the process

e t a

is stationary if and only if

X

has multivariate normal distribution and

k a p p a (t) - k a p p a (0)

is the cumulant generating function of

X

. In this case,

e t a

is a max-stable process introduced by R. L. Smith.

Recommendations

Cites work

Cited in

(3)

This page was built for publication: A characterization of the normal distribution using stationary max-stable processes