Adaptive Simulated Annealing for maximum temperature

It is difficult to determine the appropriate temperature parameters which control the acceptance probability in Simulated Annealing, which is a typical meta-heuristic method in the optimization methods. In this paper, we propose a new simulated annealing method that determines the maximum temperature adaptively. The proposed method is base on an important temperature where optimum solutions were sought effectively. The proposed method determines the maximum temperature by finding an upper limit of the important temperature in a heating process from the lowest temperature. Using this method, the total annealing steps can be decreased to half without making the accuracy of solution worse. We apply this method to some of the Traveling Salesman Problems and confirmed its effectiveness.