Load balanced parallel simulated annealing on a cluster of SMP nodes

Abstract

The paper focuses on a parallel implementation of a simulated annealing algorithm. In order to take advantage of the properties of modern clustered SMP architectures a hybrid method using a combination of OpenMP nested in MPI is advocated. The development of the reference implementation is proposed. Furthermore, a few load balancing strategies are introduced: time scheduling at the annealing process level, clustering at the basic annealing step level and suspending - inside of the basic annealing step. The application of the algorithm to VRPTW - a generally accepted benchmark problem - is used to illustrate their positive influence on execution time and the quality of results. © Springer-Verlag Berlin Heidelberg 2006.