Distributed computing of a stochastic algorithm for combinatorial optimization problems

Simulated annealing method, as a general stochastic algorithm, has proven to be particularly successful for combinatorial optimization problems. But it requires a long running time for some large scale problems. This paper introduces the synchronous and partially synchronous spatial process, instead of the Metropolis procedure, in the simulated annealing method, and shows the possibility of distributed computing. We use the module partition problem as an example to show the quality of the solutions obtained by our method and point out that a parallel computing is able to be executed to shorten the running time.