Lagrangian heuristic for a class of the generalized assignment problems

Abstract

A Lagrangian based heuristic is proposed for many-to-many assignment problems taking into account capacity limits for task and agents. A modified Lagrangian bound studied earlier by the authors is presented and a greedy heuristic is then applied to get a feasible Lagrangian-based solution. The latter is also used to speed up the subgradient scheme to solve the modified Lagrangian dual problem. A numerical study is presented to demonstrate the efficiency of the proposed approach. (C) 2010 Elsevier Ltd. All rights reserved.