Estimating average convergence velocity of genetic algorithms
Guanqi Guo · Control theory & applications · 2003
The description of mutation and crossover operators independent of representation of solutions is formulized. The precisely quantitative Markov chain model of populations of a genetic algorithm is presented. Based on this model, the stochastic matrix of Markov chain of the best-so-far individual with the highest fitness in populations is derived. The average convergence velocity of a genetic algorithm is defined as the mathematical expectation of the mean absorbing time that the best-so-far individual transfers to the absorbing state. The theoretic method and computing process of estimating the average convergence velocity of a genetic algorithm are proposed.