Uncertainties assessment in global sensitivity indices estimation from metamodels

Alexandre Janon (1, 2) , Maëlle Nodet (1) , Clémentine Prieur (1, 2)

1 MOISE - Modelling, Observations, Identification for Environmental Sciences
2 GdR MASCOT-NUM - Méthodes d'Analyse Stochastique des Codes et Traitements Numériques

Global sensitivity analysis is often impracticable for complex and resource intensive numerical models, as it requires a large number of runs. The metamodel approach replaces the original model by an approximated code that is much faster to run. This paper deals with the information loss in the estimation of sensitivity indices due to the metamodel approxima- tion. A method for providing a robust error assessment is presented, hence enabling significant time savings without sacrificing on precision and rigor. The methodology is illustrated on two different types of metamodels: one based on reduced basis, the other one on RKHS interpolation.

Mots clés: sensitivity analysis; reduced basis method; Sobol indices; bootstrap method; Monte Carlo method.; Monte Carlo method

Domaines: Calcul [stat.CO]; Statistiques [math.ST]; Théorie [stat.TH]; Analyse numérique [math.NA]

Alexandre Janon, Maëlle Nodet, Clémentine Prieur. Uncertainties assessment in global sensitivity indices estimation from metamodels. 2011. ⟨inria-00567977v2⟩