Evolutionary Strategic Computation

Evolution Strategy is a proven universal optimization tool. The mechanisms of biological evolution have been successfully transformed into optimization methods and backed with a sound theory. The theory proves: Evolution Strategies work more efficiently than many other optimization methods. Evolution Strategies show great strengths in practical experimental implementations compared to other optimization methods, especially regarding noisy fitness evaluations. Engineers and computer scientists increasingly use Evolution Strategies as an effective means of optimization.

Ingo Rechenberg formulated in 1962 the first version of the Evolution Strategy. Since then, a number of important development of the Evolution Strategies has been done. The most recent and important developments of ES initiated at the department are: efficient step size control algorithms (Covariance Matrix Adaptation, Ortho-ES), extension of the ES for noisy environments, multiobjective optimization and mechanisms for the optimization of multimodal goal functions. Some of the most remarkable applications of the last years are: e.g. optimization of ship propeller to reach greater efficiency, optimization of fuselage configurations to reduce drag, optimization of waterways to guarantee navigability and the optimization of tires to improve the aquaplaning properties.