Performance Comparison of Parameter Variation Operators in Self-Adaptive Differential Evolution Algorithms

Differential Evolution (DE) algorithm is an important Evolutionary Algorithm (EA) for global optimization over continuous spaces, which can also work with discrete variables. The success of DE in solving a specific problem is closely related to appropriately choosing its control parameters, in this context, self-adaptation allows the algorithm to reconfigure itself, automatically adapting to the problem being solved. In self-adaptation the control parameters are encoded into the genotype of the individuals and undergo the actions of variation operators. In the literature, there are several different operators proposed to vary the encoded parameters, however, there is a lack of information about their influence on the algorithms performance. To cover part of this lack of knowledge, in this paper a comparison of variation operators, commonly used to adapt parameters in self-adaptive versions of DE, is presented. The experiments on well know benchmark functions indicates that operators which maintain the control parameters diversity work better than the others.