Filled Function Method For Global Optimization

Optimization studies the characters of optimal choice on decision problems and develops numerical method to find the optimal solution.To find the effective methods for finding the global optimal solution of general multi-minimizers functions is one of the hot topics of the optimization problem.The filled function algorithm is one of the practical methods for the problem.This thesis investigated the new filled function algorithm.In the first chapter,we briefly introduced some classes of useful global optimization approaches and specially introduced the basic theory of the filled function method.We summarized the research progress of the algorithm,reviewed some classical filled functions and analyses their characteristics.Then,it provided theory basis for further research.In the second chapter,two new filled functions are proposed for finding global minimizers for unconstrained programming problems,which have only one parameter,it is easy to adjust in numerical calculation and it is better to find minimizers of the global optimization.The filled properties of the filled functions are proved,and two algorithms are developed based on the new filled functions.In the third chapter,numerical experiments of the algorithms on several common test problems were reported with satisfactory numerical result,which showed that the algorithms are efficient.