Probabilistic adaptations of point generation schemes in some global optimization algorithms

Abstract

Modifications in the trial point generation scheme are suggested to the differential evolution (de) and the controlled random search (crs) algorithm to improve their performances. Each of the modified de and the crs algorithm uses two different point generation schemes. A probabilistic adaptation of point generation schemes in each of the algorithms is proposed. Localizations in acceptance rule and in trial point generation schemes are also suggested in the resulting de and the crs algorithm, respectively. Numerical experiments indicate that the resulting algorithms are considerably better than their original counterparts. Therefore, they offer a reasonable alternative to many currently available stochastic algorithms, especially for problems requiring ‘direct search type’ methods.

Keywords:

• Global optimization
• Population set based method
• Modified differential evolution
• Modified controlled random search
• Probabilistic adaptation
• Continuous variable