A class of filled functions for finding global minimizers of a function of several variables (Q1077889)

This paper is concerned with the filled function methods for finding global minimizers of a function of several variables. A class of filled functions are defined. The advantages and disadvantages of every filled function in the class are analyzed. The best one in this class is pointed out. The idea of how to construct a better filled function is given and employed to construct the class of filled functions. A method is also explored as how to locate minimizers or saddle points of a filled function only through the use of the gradient of a function.