n 进制编码遗传算法的收敛速度

利用一个特殊的minorization条件给出了n进制编码的经典遗传算法收敛速度的一个上界,分析了种群的规模、编码串的长度以及变异概率等变量对算法收敛速度的影响,它推广了已有的结论,并对算法的参数设计有参考价值.

Convergent Rate of Genetic Algorithms with Arbitrary Encoding

At present,any mode for studying the convergence of genetic algorithm is limit. Thus,it is important and significant to study the convergence of genetic algorithms from another viewpoint,which help build right stopping criteria and appropriate measure one for comprehensively and unbiasedly judging any algorithm.However,there are few results about convergence rate.Moreover,nearly all these results limit to binary encoding genetic algorithms.In this paper,an upper bound on the rates of convergence to distributions of canonical genetic algorithms with arbitrary encoding is presented by using a special minorization condition,and effects of the population size,the encoding strength and the mutation probability on the convergence rate are given.It generalizes the existing results,and is of referenced value for designing algorithms.