Parallel simulated annealing for the job shop scheduling problem

Abstract

This paper describes two parallel simulated annealing algorithms for the job shop scheduling problem with the sum of job completion times criterion. Some properties of the problem associated with the block theory have been presented and discussed. These properties allow us to introduce the effective neighborhood based on the adjacent swap type moves. In this paper, an original method for parallel calculation of optimization criterion value for set of solutions, recommended for the use in metaheuristics with single- and multiple- search trajectories is proposed. Additionally, the vector calculation method, that uses multiple mathematical instructions MMX supported by suitable data organization, is presented. Properties of parallel calculations are empirically verified on the PC with Intel Core 2 Duo processor on Taillard's benchmarks. © 2009 Springer Berlin Heidelberg.