Skip to Main content Skip to Navigation
Conference papers

méthodologie basée sur les mesures de dépendance HSIC pour l'analyse de sensibilité de second niveau

Abstract : We are interested in the sensitivity analysis of numerical simulators in the case where the probability distributions of the uncertain input variables are themselves (even partially) unknown. The objective is to quantify the impact of these uncertainties on the sensitivity analysis results. To achieve it, we propose a single Monte Carlo loop methodology based on Hilbert-Schmidt dependence measures and importance sampling techniques. This approach significantly limits the number of simulator evaluations. A numerical application is proposed to illustrate the whole methodology, while comparing its different options.
Complete list of metadatas

Cited literature [11 references]  Display  Hide  Download

https://hal-cea.archives-ouvertes.fr/cea-02339273
Contributor : Bibliothèque Cadarache <>
Submitted on : Saturday, December 14, 2019 - 6:23:14 PM
Last modification on : Tuesday, April 28, 2020 - 11:28:14 AM
Long-term archiving on: : Sunday, March 15, 2020 - 2:05:02 PM

File

201800000409.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : cea-02339273, version 1

Collections

Citation

A. Meynaoui, A. Marrel, B. Laurent-Bonneau. méthodologie basée sur les mesures de dépendance HSIC pour l'analyse de sensibilité de second niveau. 50èmes Journées de Statistique (JdS2018), May 2018, Palaiseau, France. ⟨cea-02339273⟩

Share

Metrics

Record views

49

Files downloads

149