On Convergence of Genetic Algorithm
Angyan Tu · Journal of Hubei University of Education · 2008
Based on the methods of Markov chain,the probability distribution of populations over the solution space and the probability of converging to the optimal solution were analyzed.It proves that classic GA is not converge to the optimal solution until the GA retains the best individuals of every generation.After that,the reasons of global convergence and premature convergence are discussed.Finally that GA operations should be guided by the principles,which is the key of improving the search performance of GA,is proposed.