Global sensitivity analysis using derivative-based sparse Poincar\'e chaos expansions

TL;DR

This paper introduces derivative-based sparse Poincaré chaos expansions (PoinCE) for global sensitivity analysis, demonstrating their effectiveness in accurately estimating Sobol' indices and derivative-based sensitivity measures using spectral methods.

Contribution

The paper presents a novel derivative-based sparse Poincaré chaos expansion method that leverages the unique orthogonality properties of Poincaré basis functions for improved sensitivity analysis.

Findings

PoinCE provides accurate Sobol' index estimates, outperforming polynomial chaos expansions.

The derivative-based sensitivity measure (DGSM) can be efficiently computed from PoinCE coefficients.

Numerical examples validate the effectiveness of the proposed method in sensitivity analysis.

Abstract

Variance-based global sensitivity analysis, in particular Sobol' analysis, is widely used for determining the importance of input variables to a computational model. Sobol' indices can be computed cheaply based on spectral methods like polynomial chaos expansions (PCE). Another choice are the recently developed Poincar\'e chaos expansions (PoinCE), whose orthonormal tensor-product basis is generated from the eigenfunctions of one-dimensional Poincar\'e differential operators. In this paper, we show that the Poincar\'e basis is the unique orthonormal basis with the property that partial derivatives of the basis form again an orthogonal basis with respect to the same measure as the original basis. This special property makes PoinCE ideally suited for incorporating derivative information into the surrogate modelling process. Assuming that partial derivative evaluations of the computational model are available, we compute spectral expansions in terms of Poincar\'e basis functions or basis partial derivatives, respectively, by sparse regression. We show on two numerical examples that the derivative-based expansions provide accurate estimates for Sobol' indices, even outperforming PCE in terms of bias and variance. In addition, we derive an analytical expression based on the PoinCE coefficients for a second popular sensitivity index, the derivative-based sensitivity measure (DGSM), and explore its performance as upper bound to the corresponding total Sobol' indices.