An interview with Hans-Paul Schwefel: with an introduction by Günter Rudolph

The sounds of Sputnik from its orbit around the earth in 1957 were the origin of the young Hans-Paul Schwefel's desire to become a space traveller. In thoughtful preparation he soon began his studies in Aero- and Space Technology at the Technical University Berlin (TUB). But while he was junior assistant at the Hermann-Föttinger Institute for Hydrodynamics at TUB another event changed his plans: In the 1960s difficult multimodal and noisy optimization problems from engineering sciences awaited their solution at the TUB. At that time mathematical models or numerical simulations of these particular problems were not available. As a consequence, the optimization had to be done experimentally with the real object at hardware level. The three students Peter Bienert, Ingo Rechenberg and Hans-Paul Schwefel (HPS) envisioned an automated cybernetic system that alters the object parameters mechanically or electrically, runs the experiment, measures the outcome of the experiment and uses this information in the context of some optimization strategy for the decision how to alter the object parameters of the real object for the experiment in the subsequent iteration. Obvious candidates for the optimization methods in the framework of this early 'hardware-in-the-loop' approach were all kinds of gradient-like descent methods for minimization tasks. But these methods failed. Inspired by lectures on biological evolution they tried a randomized method that may be regarded as the simplest algorithm driven by mutation and selection---a method nowadays known as the (1+1)-Evolution Strategy. This approach was successful and this event may be seen as the trigger that turned the career of HPS towards the emerging scientific field of evolutionary computation (EC).