Skip to Main content Skip to Navigation
New interface

Etude des dispersions et incertitudes en optimisation et dans l'analyse des valeurs propres

Abstract : After an introduction concerning the techniques used to model uncertainties, the question of quantifying the latter is adressed. Focusing on two particular issues will make it possible to develop a new methodology on that purpose. First, we will tackle the issue of characterizing the dispersion affecting the solution ofan optimization problem whose objective or constraints are random. Afterwards, we will deal with computation of random matrices eigenelements. Original strategies based on transposing standard deterministic numerical schemes into stochastic framework are introduced. Besides, tools provided by reliability-based optimization turn out to be usefulto reach a balance between a lowest cost and a satisfying reliability. However, useof such approaches can lead to prohibitive computational effort. To overcome this difficulty, a new strategy is proposed to derive an approximated solution of initial problem. Depending on requirements on computational time or accuracy, it is shownthat several measurements of reliability can be used.
Complete list of metadata

Cited literature [186 references]  Display  Hide  Download
Contributor : ABES STAR :  Contact
Submitted on : Wednesday, October 10, 2012 - 2:55:17 PM
Last modification on : Tuesday, October 19, 2021 - 12:11:14 PM
Long-term archiving on: : Friday, January 11, 2013 - 3:41:19 AM


Version validated by the jury (STAR)


  • HAL Id : tel-00740583, version 1


Rémi Croquet. Etude des dispersions et incertitudes en optimisation et dans l'analyse des valeurs propres. Autre [cond-mat.other]. INSA de Rouen, 2012. Français. ⟨NNT : 2012ISAM0006⟩. ⟨tel-00740583⟩



Record views


Files downloads