Improving the Neighborhood Selection Strategy in Simulated Annealing Using the Optimal Stopping Problem

Abstract

The search method utilized by metaheuristic algorithms is in fact a local search in the solu- tion space. Investigations of these methods show that their efficiency in solving different types of problems significantly depends on the applied strategy in searching the solution space. Tradi- tional neighborhood selection methods have disregarded the essential characteristics embodied in the chosen neighborhood. In this research, we enhance the efficiency of solving combinatorial optimization problems using simulated annealing method by taking the advantages embodied in neighborhood structures. This new algorithm improves simulated annealing in two different aspects: First, by employing multiple neighborhood structures, it performs a more powerful search and second, using optimal stopping problem, it finds the best time to change the tem- perature which is a critical issue in simulated annealing. The novel algorithm is illustrated on the traveling salesman problem.