In this paper we demonstrate the applicability of evolutionary algorithms (EAs) to the optimization of circuit designs. We examine the design of a full-adder cell, and show the capability of design of experiments (DOE) methods to improve the parameter-settings of EAs.

A software package for development, analysis and application of multiobjective evolutionary algorithms is described. The object-oriented design of this kit for evolutionary algorithms (KEA) offers a good suitable environment for various kinds of optimization tasks. It provides an interface to evaluate multi-objective fitness functions written in Java or C/C++ using a variety of multi-objective evolutionary algorithms (MOEA). In addition KEAcontains several state-of-the-art comparison methods for performance measure of algorithms. Furthermore KEAis able to display the progress of optimization in a dynamic display or just to display the results of optimization in a static visualization mode.