Meta-RaPS approach for the 0-1 Multidimensional Knapsack Problem

A promising solution approach called Meta-RaPS is presented for the 0-1 Multidimensional Knapsack Problem (0-1 MKP). Meta-RaPS constructs feasible solutions at each iteration through the utilization of a priority rule used in a randomized fashion. Four different greedy priority rules are implemented within Meta-RaPS and compared. These rules differ in the way the corresponding pseudo-utility ratios for ranking variables are computed. In addition, two simple local search techniques within Meta-RaPS' improvement stage are implemented. The Meta-RaPS approach is tested on several established test sets, and the solution values are compared to both the optimal values and the results of other 0-1 MKP solution techniques. The Meta-RaPS approach outperforms many other solution methodologies in terms of differences from the optimal value and number of optimal solutions obtained. The advantage of the Meta-RaPS approach is that it is easy to understand and easy to implement, and it achieves good results.