Combining Resolution Search and Dynamic Programming for the 0-1 Multidimensional Knapsack Problem.
Résumé
We propose an exact method which combines resolution search, branch and bound, and dynamic programming algorithm for solving the 0-1 Multidimensional Knapsack Problem. Our method is a multi-level search strategy where the top level branches of the search tree are enumerated using Resolution Search, the middle level branches are enumerated using Branch & Bound and the lower level branches are solved with dynamic programming. The proposed algorithm is competitive with the existing heuristics in terms of convergence and is able to prove large-scale strong correlated instances of the OR-Library