基于MPI的模拟退火算法并行性研究

Simulated Annealing Algorithm in Parallel Studies based on MPI Environment

针对标准模拟退火算法仅进行串行优化效率难以提高的问题,充分利用MPI并行环境的优势,提出了一种基于MPI环境的并行模拟退火算法,并将其应用于一些诸如TCP之类的NP问题的解决中。通过这种方法,可在加速、效率和解的质量上比以往传统标准模拟退火算法有显著的突破。又由于整个模拟退火算法是一个独立的模块,具有良好的可移植性,适用于任何多变量函数的最优化问题,可以更好地求解NP问题。

In order to solve the problem that standard simulated annealing using serial optimize method is difficult to improve its efficiency,the methods of making full use of MPI and forming the parallel simulated annealing realized on the MPI environment are presented.A number of NP problems such as TCP（Traveling Salesman Problem）are solved in this way,so that make a significant breakthrough in the aspects of the speeding up,the efficiency and the quality standards.The good results obtained from these indicate that the program is of high performance and is applicable to various regions.For the reason that the whole simulated annealing is an independent module and has a good portability,the NP problems can be better solved.