Aller au contenu. | Aller à la navigation

Outils personnels

Navigation

UMR 5672

logo de l'ENS de Lyon
logo du CNRS
Vous êtes ici : Accueil / Séminaires / Machine Learning and Signal Processing / A fast semi-exact primal-dual method for steady viscoplastic flows

A fast semi-exact primal-dual method for steady viscoplastic flows

François Bouchut (DR CNRS, Université Gustave Eiffel)
Quand ? Le 01/07/2025,
de 13:00 à 14:00
Où ? M7 101
Participants François Bouchut
Ajouter un événement au calendrier vCal
iCal

FRANCOIS BOUCHUT

Title:  A fast semi-exact primal-dual method for steady viscoplastic flows

Abstract:  Constitutive laws for viscoplastic materials involve a multivalued nonlinearity. Duality methods for such equations are known to converge, but the convergence is slow, the error is in general of the order of 1/𝑘 with 𝑘 the number of iterations.
We consider here an iterative method of classical implicit primal-dual type with a particular form with exact conservation. An analysis via a Lyapunov functional yields admissible values of the parameters, whereas a linearized analysis for scalar problems indicates the potentially best values. An adaptive choice of the parameters enables to achieve an optimal convergence.
Numerical tests show an improved performance with respect to the classical Augmented Lagrangian method or the dual FISTA method, in particular when a large zero order term is present.
 
 

Website: https://perso.math.u-pem.fr/bouchut.francois/

In Room M7 101, 1st floor, Monod campus, ENSL.