On fractional Gaussian random fields simulations

Alexandre Brouste (1) , Jacques Istas (2) , Sophie Lambert-Lacroix (2)

1 Département de Mathématiques [Le Mans]
2 SMS - Statistique et Modélisation Stochatisque

To simulate Gaussian fields poses serious numerical problems: storage and computing time. The midpoint displacement method is often used for simulating the fractional Brownian fields because it is fast. We propose an effective and fast method, valid not only for fractional Brownian fields, but for any Gaussian fields. First, our method is compared with midpoint for fractional Brownian fields. Second, the performance of our method is illustrated by simulating several Gaussian fields. The software FieldSim is an R package developed in R and C and that implements the procedures on which this paper focuses.

Mots clés: Gaussian fi elds; simulation; Hölder index

Domaines: Statistiques [math.ST]; Théorie [stat.TH]

Alexandre Brouste, Jacques Istas, Sophie Lambert-Lacroix. On fractional Gaussian random fields simulations. Journal of Statistical Software, 2007, 23 (1), pp.1-23. ⟨hal-00853131⟩