INVESTIGATION OF A STOP CRITERION IN EVOLUTIONARY ALGORITHMS

Abstract

The limit theory of Evolutionary Algorithms for optimization problems with single objective has been developed well recently. The result will become quite difference for the domain of the evolutionary optimization of mulitple objective functions. Since these problem classes can be considered as a special case for determine a set of minimal elements (maximal) in the partially order sets, the limit theory of the evolutionary algorithm can satisfy for this kind of problems, it allow to transfer all the results and properties to the above special case. Another problem concerned to the efficiency of the algorithms, that is the execution time of the algorithm, after developing of the dynamic model for the algorithm, many study had been tried to test the stop criterion with the best value which obtained in the previous generation for comparing and developing the methods for finding the responsible criterion. This paper deal with a well known method, in which the Expected Waiting Time (EWT) will be calculated clearly base upon the Markov model of Nix and Vose which developed for the Simple Genetic Algorithm (SGA).