Resumen en español

The present project seeks to develop a metaheuristic for solving a variant of the Multiple Knapsack Problem, based on a recently developed metaheuristic for solving continuous domain optimization problems. To achieve this, an algorithm that has the logic behind the initial model was developed and new functions were added to make it work properly with the binary encoding of the decision variables of the problem. In order to make comparison, a couple of variants of the main model were designed and an experiment was built in which several runs were performed, with predefined and tested with other algorithms instances, in a random order for each variant, and finally get some descriptive data, GAP’s and average execution times. Results indicate a competent performance of the metaheuristics compared to exact algorithms like CPLEX, while it’s less competent compared to other metaheuristics developed previously for this type of optimization problems, in terms of both target value and execution time.