On self-adaptive features in real-parameter evolutionary algorithms

Due to the exibility in adapting to diierent tness landscapes, self-adaptive evolutionary algorithms (SA-EAs) have been gaining popularity in the recent past. In this paper, we postulate the properties that SA-EA operators should have for successful applications in real-valued search spaces. Speciically, population mean and variance of a number of SA-EA operators, such as various real-parameter crossover operators and self-adaptive evolution strategies, are calculated for this purpose. Simulation results are shown to verify the theoretical calculations. The postulations and population variance calculations explain why self-adaptive GAs and ESs have shown similar performance in the past and also suggest appropriate strategy parameter values which must be chosen while applying and comparing diierent SA-EAs.