Parameter-less genetic algorithm: a worst-case time and space complexity analysis
Martin Pelikán, Fernando Graça Lobo · 2000
In this paper, the worst-case analysis of the time and space complexity of the parameter-less genetic algorithm versus the genetic algorithm with an optimal population size is provided and the results of the analysis are discussed. Since the assumptions in order for the analysis to be correct are very weak, the result is applicable to a wide range of problems. Various configurations of the parameter-less genetic algorithm are considered and the results of their time and space complexity are compared. 1 Introduction A parameter-less genetic algorithm (Harik & Lobo, 1999) is an alternative to a common trialand -error method of tweaking the values of the parameters of the genetic algorithm in order to find a set-up to accurately and reliably solve a given problem. The algorithm manages a number of independent runs of the genetic algorithm with different population sizes with the remaining parameters set to fixed values according to the theory of genetic algorithms' control maps introduce...