Aims of the Project

One early aim in this project was to create a GP system using some of Perl's most convenient and well developed features (namely hash tables and regular expressions) rather than make a Perl clone of existing implementations. The initial goal was to evolve string manipulation code for protein secondary structure prediction, but through time the emphasis has shifted more towards making a robust GP system for general use.

This GP implementation has always tried to follow the evolution of biological organisms. It does not start from a baseline, such as Koza's work, but uses many borrowed and a few original ideas. The first release of this software contains code for running tournament based genetic algorithm on a population of tree-encoded programs. However, the object-oriented design allows exchange of components, for example a generational GA or Monte Carlo approach for the search algorithm, and other representations of individuals. Object method overloading lets you easily customise the algorithms to suit your needs.