PODI is a highly general evolutionary algorithm. It can be used as a normal genetic algorithm, or to carry out genetic programming or grammatical evolution, or to attack problems like TSP which require alternative representations. It achieves this generality using the same idea as GE: a variable-length integer array genome is evolved by a GA, and the genotype to phenotype mapping process consists of reading the genome one integer at a time, each one making one "decision" in the creation of the phenotype. The difference is that instead of the fixed grammar derivation process of GE, in PODI any non-deterministic program can be used. That program's possible outputs are the feasible solution space, ie the possible phenotypes. We can then see that non-deterministic program (NDP) as a mapping from integer-array genotype to phenotype.

This repository contains Python source code implementing the PODI idea, and several example NDPs. These examples allow PODI to emulate GE precisely, and to carry out two novel forms of GP (no claim is made as yet that they are efficient). There is also code for running most of the experiments in the EuroGP2013 paper (see below). Code for running the RSAP experiment in that paper is not provided since the heuristic NDP used there is not under the copyright of the author.

If you wish to cite this project, please cite this paper:

McDermott and Carroll, Program Optimisation with Dependency Injection, in Proceedings of EuroGP 2013, Vienna, Austria, Springer.

Abstract: For many real-world problems, there exist non-deterministic heuristics which generate valid but possibly sub-optimal solutions. The program optimisation with dependency injection method, introduced here, allows such a heuristic to be placed under evolutionary control, allowing search for the optimum. Essentially, the heuristic is "fooled" into using a genome, supplied by a genetic algorithm, in place of the output of its random number generator. The method is demonstrated with generative heuristics in the domains of 3D design and communications network design. It is also used in novel approaches to genetic programming.

Authors: James McDermott and Paula Carroll, Management Information Systems, Quinn School of Business, University College Dublin, Ireland.