Building kriging models using hierarchical codes with different levels of accuracy.

Loic Le Gratiet (1, 2)

1 LPMA - Laboratoire de Probabilités et Modèles Aléatoires
2 GdR MASCOT-NUM - Méthodes d'Analyse Stochastique des Codes et Traitements Numériques

This paper deals with the Gaussian process based approximation of a code which can be run at different levels of accuracy using co-kriging. This method allows us to improve a surrogate model of a complex computer code using fast approximations of it. In particular, we focus on the case of large number of code levels. A thermodynamic example is used to illustrate a 3-level co-kriging.

Mots clés: co-kriging; multi-level code; computer experiment; surrogate models; Gaussian process regression

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

Loic Le Gratiet. Building kriging models using hierarchical codes with different levels of accuracy.. 11th annual meeting of the European Network for Business and Industrial Statistics, Sep 2011, Coimbra, Portugal. ⟨hal-00654710⟩