An efficient simple cooling schedule for simulated annealing

Abstract

The capability of Global solution of an optimization problem is the forte of Simulated Annealing (SA). Theoretically only the infinite-time algorithm can guarantee the global solution. The finite-time characteristics of the algorithm depend largely on the ensemble of certain control parameters. Since the main parameter is dubbed temperature, the dynamics of how it is manipulated is widely known as cooling schedule. A variety of methods, from simple geometric to highly complex, have been proposed in the literature. While global solution capability has been the overall goal for all implementation, few schedules combined effective solution with simplicity of the cooling schedule. A novel schedule is proposed which combines efficiency with simplicity into an easily implementable algorithm. Several fundamental cooling schemes are compared with the proposed one based on 2 test problems. Our schedule faired competitively with most while being the simplest. © Springer-Verlag Berlin Heidelberg 2004.