改进的模拟退火算法在L—J问题中的应用

AN IMPROVEMENT OF SIMULATED ANNEALING ALGORITHM AND ITS APPLICATION STUDY

针对高维连续函数的全局优化问题，笔者将两种确定性局部极小化过程分别引入到模拟退火算法当中，并将该算法应用到Lennard—Jones簇问题中．通过结果比较说明该算法可以提高计算的精度和成功率，并且Hooke—Jeaves方法在处理复杂的函数问题时比单纯形法有效．

In regarding of the characters of the global optimization problems of continuous multi - dimension function ,two determined local minimum methods are added to the simulated annealing respectively. As applications of this method, the Lennard - Jones Cluster model is computed. The results show that the new method can improve the precision and the percentage of success , and the Hooke -Jeaves method is more effective than Simplex Method in dealing with the complex function.