Generalized augmented Lagrangian methods for equality constrained optimization problems

X.X. Huang, Xiaoqi Yang

Research output: Journal article publicationJournal articleAcademic researchpeer-review

Abstract

In this paper we consider generalized augmented Lagrangian methods including a classical augmented Lagrangian method and some " lower order" augmented Lagrangian methods as special cases for a mathematical program with only equality constraints. Since generalized augmented Lagrangians are in general not differentiable or even not locally Lipschitz, we carry out convergence analysis of first-order and second-order stationary points of generalized augmented Lagrangian methods by applying the Borwein-Preiss approximate smooth variational principle.
Original languageEnglish
Pages (from-to)81-99
Number of pages19
JournalPacific Journal of Optimization
Volume1
Issue number1
Publication statusPublished - 2005

Keywords

  • Equality constrained optimization
  • Augmented Lagrangian
  • Constraint qualification
  • Optimality condition

ASJC Scopus subject areas

  • Applied Mathematics
  • Computational Mathematics
  • Control and Optimization

Cite this