Evolution and optimum seeking

Preface In 1963 two students at the Technical University of Berlin met and were soon to collaborate on experiments which used the wind tunnel of the Institute of Flow Engineering. During the search for the optimal shapes of bodies in a ow, which w as then a matter of laborious intuitive experimentation, the idea was conceived of proceeding strategically. However, attempts with the coordinate and simple gradient strategies were unsuccessful. Then one of the students, Ingo Rechenberg, now Professor of Bionics and Evolutionary Engineering, hit upon the idea of trying random changes in the parameters deening the shape, following the example of natural mutations. The evolution strategy was born. A third student, Peter Bienert, joined them and started the construction of an automatic experimenter, which w ould work according to the simple rules of mutation and selection. The second student, I myself, set about testing the eeciency of the new methods with the help of a Zuse Z23 computerr for there were plenty of objections to these random strategies. In spite of an occasional lack of nancial support, the Evolutionary Engineering Group which had been formed held rmly together. Ingo Rechenberg received his doctorate in 1970 for the seminal thesis: Optimierung technischer Systeme nach Prinzip-ien der biologischen Evolution. It contains the theory of the two m e m bered e v olution strategy and a rst proposal for a multimembered strategy, w h i c h in the nomenclature introduced here, is of the (+1) type. In the same year nancial support from the Deutsche Forschungsgemeinschaft (German Research Association) enabled the initiation of the work which comprises most of the present book. This work was concluded, at least temporarily, in 1974 with the thesis Evolutionsstrategie und numerische Optimierung and published by Birkhh auser, Basle, Switzerland, in 1977 under the title Numerische Optimierung von Computer-Modellen mittels der Evolutionsstrategie as well as by Wiley, Chichester, in 1981 as monograph Numerical optimization of computer models. Between 1976 and 1985 the author was not able to continue his work in the eld of Evolution Strategies (nowadays abbreviated: ESs). The general interest in this type of optimum seeking algorithms was not broad enough for there to be nancial support. On the other hand, the number of articles, journals, and books devoted to (mathematical) optimization has increased tremendously. Looking back upon the development from 1964 on, when the rst ES version was devoted to experimental optimization, …