A Unified Framework of Regularization Methods for Degenerate Non-Linear Optimization Models - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2018

A Unified Framework of Regularization Methods for Degenerate Non-Linear Optimization Models

Résumé

In this paper, we study three classes of difficult non-linear optimization, problems with complementarity (MPCC), vanishing (MPVC) and cardinality constraints (OMCC). They all have in common degenerate constraints, which fail to satisfy classical constraint qualifications in a generic way. The feasible sets of these problems are non-convex, possibly non-connected, with an empty relative interior. This causes several difficulties in practice. In a recent work (Dussault et al. 2017), we propose a unified framework of methods that consider a regularization-penalization-active set method to solve the MPCC, which possesses the best-known convergence properties. In this paper, we extend this unified framework to MPVC and OMCC and consider some applications on optimal control problems.
Fichier principal
Vignette du fichier
article_degenerateNLP_HAL.pdf (411.85 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01734832 , version 1 (15-03-2018)

Identifiants

  • HAL Id : hal-01734832 , version 1

Citer

Tangi Migot, Jean-Pierre Dussault, Mounir Haddou. A Unified Framework of Regularization Methods for Degenerate Non-Linear Optimization Models. 2018. ⟨hal-01734832⟩
342 Consultations
178 Téléchargements

Partager

Gmail Facebook X LinkedIn More