ABSTRACT
Minimizing the total drag of three-dimensional slender bodies in a turbulent flow was, and still is, a general goal of research in institutes of hydrodynamics. Three students (Peter Bienert, Ingo Rechenberg, and Hans-Paul Schwefel) met each other at such an institute, the Hermann Fo¨ttinger Institute of the Technical University of Berlin, in 1964. Since they were fascinated not only by aerodynamics, but also by cybernetics, they hit upon the idea to solve the analytically (and at that time also numerically) intractable form design problem with the help of some kind of robot. The robot should perform the necessary experiments by iteratively manipulating a flexible model positioned at the outlet of a wind tunnel. An experimentum crucis was set up with a two-dimensional foldable plate. The iterative search strategy-first performed by hand, a robot was developed later on by Peter Bienert-was expected to end up with a flat plate: the form with minimal drag. But it did not, since a one-variable-at-a-time as well as a discrete gradient-type strategy always got stuck in a local minimum: an S-shaped folding of the plate. Switching to small random changes that were only accepted in the case of improvements-an idea of Ingo Rechenbergbrought the breakthrough, which was reported at the joint annual meeting of WGLR and DGRR in Berlin, 1964 (Rechenberg 1965). The interpretation of binomially distributed changes as mutations and of the decision to step back or not as selection (on 12 June 1964) was the seed for all further developments leading to evolution strategies (ESs) as they are known today. So much about the birth of the ES.
