Publication detail

LOCALIZED APPROXIMATIONS BY POLYNOMIAL CHAOS EXPANSIONS FOR HIGHLY NON-LINEAR FUNCTIONS

NOVÁK, L. SHIELDS, M. SADÍLEK, V. VOŘECHOVSKÝ, M.

Original Title

LOCALIZED APPROXIMATIONS BY POLYNOMIAL CHAOS EXPANSIONS FOR HIGHLY NON-LINEAR FUNCTIONS

Type

conference paper

Language

English

Original Abstract

The paper presents a strategy for adaptive sequential decomposition of the design domain and for localized surrogate modeling via polynomial chaos expansion. The proposed method combines adaptivity of localized surrogate modeling and sequential sampling. The iteration process of the decomposition and sequential sampling balances between exploitation of the surrogate model and exploration of the design domain. The proposed methodology is especially beneficial for highly non-linear functions, which are extremely challenging for and approximation by polynomial chaos expansion. The obtained numerical results confirm its superiority over a single global surrogate model and a recently proposed technique based on similar concepts.

Keywords

Adaptive Sampling; Polynomial Chaos Expansion; Sequential Sampling

Authors

NOVÁK, L.; SHIELDS, M.; SADÍLEK, V.; VOŘECHOVSKÝ, M.

Released

12. 6. 2023

Publisher

National Technical University of Athens

Location

Athens, Greece

ISBN

9786185827021

Book

UNCECOMP 2023 Proceedings

Pages from

158

Pages to

164

Pages count

7

BibTex

@inproceedings{BUT187085,
  author="Lukáš {Novák} and Michael {Shields} and Václav {Sadílek} and Miroslav {Vořechovský}",
  title="LOCALIZED APPROXIMATIONS BY POLYNOMIAL CHAOS EXPANSIONS FOR HIGHLY NON-LINEAR FUNCTIONS",
  booktitle="UNCECOMP 2023 Proceedings",
  year="2023",
  pages="158--164",
  publisher="National Technical University of Athens",
  address="Athens, Greece",
  isbn="9786185827021"
}