Abstract
Tabu search, simulated annealing and genetic algorithms are commonly used heuristic search methods. People attempt to improve these methods to reduce computation time. Therefore, multiple processors computation platform using parallel searching algorithms have been used and studied in recent decades.One of the parallel simulated annealing algorithms is the clustering algorithm. The main idea of this method is to make each processor generate its own adjacent solution from the same current solution and compute the acceptance probability for each adjacent solution. Then, it decides which adjacent solution should replace the current solution according to pre-specified rules.The purpose of using clustering parallel simulated annealing is to find an acceptable adjacent solution quickly. However, we may find too many acceptable adjacent solutions, especially when temperature is high. It is wasteful to compute discarded acceptable adjacent solutions except the solution that is chosen to replace the current solution.Furthermore, simulated annealing somewhat lacks of diversified searching. We attempt to tackle this problem by using multiple processors to perform several simultaneously independent searching sequences at the same time, and later make them to turn to clustering mode at the time when computation process meet a certain criteria.Because the method we discuss allocates computing resource dynamically, we call it Dynamic Parallel Simulated Annealing Algorithms (DPSA).