single-objective optimization, adaptive differential evolution, reflection of simplex, CEC2013 functions

A modified competitive-adaptive differential evolution (DE) algorithm is proposed and tested experimentally on the test suite defined for the competition of optimization algorithms at CEC 2013. The algorithm combines twelve various DE strategies and a randomized reflection of simplex. The simplex is created from randomly selected points of population. The reflection generates a new trial point by a rotation-invariant strategy quite different from DE, which has appeared beneficial for the efficiency of the search in many test problems. The reasons of worse efficiency of the algorithm in some problems is also analyzed.