一类连续函数模拟退火算法及其收敛性分析  

A SIMULATED ANNEALING ALGORITHM FOR CONTINUOUS FUNCTIONS AND ITS CONVERGENCE PROPERTIES

作　　者：靳利霞[1] 唐焕文[1] 李斌[1] 计明军[1] 朱训芝[1] 

机构地区：[1]大连理工大学应用数学系,大连116024

出　　处：《计算数学》2005年第1期19-30,共12页Mathematica Numerica Sinica

基　　金：国家自然科学基金(90103033)资助.

摘　　要：高维连续函数的全局优化问题普遍存在于计算生物学、计算化学等领域.针对这类问题和现有连续函数模拟退火算法的某些不足,本文给出了一类改进的模拟退火算法.采用一种简单的方法证明了算法的全局收敛性.数值结果表明,对于高维连续函数,该算法能够快速有效地收敛到全局最优点,比较了两种新解产生方法的试验结果.The global optimization problems of continuous high-dimensional function frequently present in the fields of computational biology and computational chemistry. Regarding of the characters of this class of problems and the disadvantages of the existing simulated annealing algorithms, we propose an improved simulated annealing algorithm. Its convergence properties are proved by using a simple approach. Numerical tests on some typical optimization problems show that this is an effective algorithm. The testing results by two different new points generating approaches are compared.

关 键 词：连续函数 全局收敛性 模拟退火算法 高维 计算化学 全局优化问题 收敛性分析 全局最优 数值 快速 

分 类 号：O224[理学—运筹学与控制论]