Article ID: | iaor20081471 |
Country: | United States |
Volume: | 54 |
Issue: | 6 |
Start Page Number: | 1172 |
End Page Number: | 1184 |
Publication Date: | Nov 2006 |
Journal: | Operations Research |
Authors: | Righini Giovanni, Ceselli Alberto |
Keywords: | programming: integer |
The multilevel generalized assignment problem (MGAP) is a variation of the generalized assignment problem, in which agents can execute tasks at different efficiency levels with different costs. We present a branch-and-price algorithm that is the first exact algorithm for the MGAP. It is based on a decomposition into a master problem with set-partitioning constraints and a pricing sub problem that is a multiple-choice knapsack problem. We report on our computational experience with randomly generated instances with different numbers of agents, tasks, and levels; and with different correlations between cost and resource consumption for each agent–task–level assignment. Experimental results show that our algorithm is able to solve instances larger than those of the maximum size considered in the literature to proven optimality.