28
112

Sensitivity analysis for multidimensional and functional outputs

Abstract

Let X:=(X1,,Xp)X:=(X_1, \ldots, X_p) be random objects (the inputs), defined on some probability space (Ω,F,P)(\Omega,{\mathcal{F}}, \mathbb P) and valued in some measurable space E=E1××EpE=E_1\times\ldots \times E_p. Further, let Y:=Y=f(X1,,Xp)Y:=Y = f(X_1, \ldots, X_p) be the output. Here, ff is a measurable function from EE to some Hilbert space H\mathbb{H} (H\mathbb{H} could be either of finite or infinite dimension). In this work, we give a natural generalization of the Sobol indices (that are classically defined when YRY\in\mathbb R ), when the output belongs to H\mathbb{H}. These indices have very nice properties. First, they are invariant. under isometry and scaling. Further they can be, as in dimension 11, easily estimated by using the so-called Pick and Freeze method. We investigate the asymptotic behaviour of such estimation scheme.

View on arXiv
Comments on this paper