Modeling the dynamics of a changing range genetic algorithm in noisy environments

Abstract

The paper discussed the dynamic model of a genetic algorithm (GA) with an adjustment of a search space size in a noisy environment. A linear pseudo-Boolean function with positive coefficients was used for modeling the GA with a search space size adjustment. The impact of an additive noise on the fitness value of the GA is assessed, and analytical expressions are derived that described how GA’s selection and size adjustment operators alter the macroscopic statistical properties of population. Calculations based on the theory are compared with experiments demonstrating the accurate predictions for the macroscopic statistical properties of population.