Comparison of steady state and elitist selection genetic algorithms

It is proposed that the comparison problem of two models of the genetic algorithm, the steady state genetic algorithm and the elitist selection genetic algorithm. The convergence speed, on-line and off-line performance of the two algorithms in different environments are compared. It is experimentally shown that the steady state genetic algorithm is a simple, effective genetic algorithm. The steady state genetic algorithm runs well in low-dimensional environment especially its good on-line performance. On the other band the elitist selection genetic algorithm runs well in high-dimensional environment, it has good capability in searching optimal value in a big space. The difference between the searching ability of two algorithms was explained by the theory of Implicit Parallelism. The difference between the two algorithms on-line performances was explained by the difference ways of reproduction the two models used.