GPU Accelerated Metaheuristics for Integrated Production Lot Sizing and Scheduling Problems

Abstract

This paper presents an investigation of GPU-accelerated multi-population algorithms for two-stage multi-machine lot scheduling problems. While the literature suggests a variety of optimization techniques for this class of problems, here we investigate GPU vectorized Differential Evolutionary and Dispersive Flies Optimization algorithms combined with an exact Branch-and-Cut method. Computational tests with in-stances from the literature have shown that the GPU-accelerated heuristics can offer, in some cases, computational times that are not attainable with exact methods. Finally, in the conclusion potential areas for further study are discussed.