A computational framework for certified reduced basis methods: applications to multiphysics problems
Résumé
Over the last few years we have been developing an open-source computational framework for certified reduced basis methods based on the Feel++ framework. The latter provides (i) a very wide range of arbitrary order Galerkin methods (fem, sem, cG, dG) in 1D, 2D and 3D (ii) an interface to modern linear algebra framework PETSc/SLEPc and Trilinos and (iii) a language embedded in C++ for Galerkin methods with a syntax following closely the mathematical one. Our recent advances in the former allow to (i) treat a wide range of problems as exercised later, (ii) provide an (almost) automated framework including the use of reduced basis element methods, hp reduced basis approximations, empirical interpolation method, the successive contraints methods, a posteriori error estimation for linear and non-linear parametrized partial differential equations.
Fichier principal
eccomas12.pdf (341.82 Ko)
Télécharger le fichier
HR_modelling.png (195.39 Ko)
Télécharger le fichier
eccomas_rb_slides.pdf (2.39 Mo)
Télécharger le fichier
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
| Format | Figure, Image |
|---|
| Format | Autre |
|---|
Loading...