Self-adaptive software is one of the key discoveries in the field of evolutionary computation, originally invented in the framework of so-called Evolution Strategies in Germany. Self-adaptivity enables the algorithm to dynamically adapt to the problem characteristics and even to cope with changing environmental conditions - as they occur in unforeseeable ways in many real-world business applications. In evolution strategies, self-adaptability is generated by means of an evolutionary search process that operates on the solutions generated by the method as well as on the evolution strategy's parameters, i.e., the algorithm itself. By focusing on a basic algorithmic variant of evolution strategies, the fundamental idea of self-adaptation is outlined in this paper. Applications of evolution strategies for NuTech's clients include the whole range of business tasks, including R&D, technical design, control, production, quality control, logistics, and management decision support. While such examples can of course not be disclosed, we illustrate the capabilities of evolution strategies by giving some simpler application examples to problems occurring in traffic control and engineering. (C) 2002 Elsevier Science Inc. All rights reserved.
