A method to establish the cooling scheme in simulated annealing like algorithms

Abstract

Since the publication of the seed paper of Simulated Annealing algorithm (SA) written by Kirkpatrich, several methods have been proposed to get the cooling scheme parameters. Although developed for SA, some of these methods can be extended to the algorithm known as Threshold Accepting (TA). SA and TA are quite similar and both are treated in this paper as a Simulated Annealing Like (SAL) algorithm. This paper presents a method to set the cooling scheme parameters in SAL algorithms; it establishes that both, the initial and the final temperatures are function of the maximum and minimum cost increment getting from the neighborhood structure. Experimentation with Traveling Salesman Problem and Hydraulic Network Design Problem shows that the cooling schemes getting through our method are more efficient than the previous ones. © Springer-Verlag Berlin Heidelberg 2004.