In this work, a theoretical analysis of proximal point methods is done for solving nonconvex optimization problems and more specifically DC programming problems. The aim is to study convergence properties of an approximate proximal point algorithm for minimizing DC functions and to suggest an algorithmic scheme for computing the proximal point. The method used in this work is the bundle method well-known in convex optimization. Our analysis also allows us to consider the exact scheme.
Une méthode de point proximal pour la minimisation d'une fonction DC
Simonet, C. (Author). 2006
Student thesis: Master types › Master in Mathematics