Bundle-Based Decomposition: Conditions for Convergence
Stephen M. Robinson · Annales de l Institut Henri Poincaré C Analyse Non Linéaire · 1989
Bundle-based decomposition is a recently proposed method for decentralized convex optimization. Computational tests indicate that it is very fast. In this paper we exhibit conditions for convergence of the method. In the process we study conditions for linearly-constrained approximate minimization of a convex function.