TEA - A C++ Library for the Design of Evolutionary Algorithms

Abstract

A library for the design of standard and non-standard EAs in C++ is described. The simple object-oriented design of the TEA library allows the fast configuration of new non-standard evolutionary algorithms. In TEA representation independent algorithms can be combined with non-standard genotypes. Complex genotypes can be build from existing simple genotypes. Furthermore, non-panmictic parallel population structures like neighbourhood and multipopulation EAs are supported. This paper introduces the main concepts of the TEA library. Examples illustrate how to build standard algorithms and how to design new kinds of algorithms and representations with TEA.