Introduction

The project started when I saw a person using lil-gp on his PC at the Artificial Intelligence Laboratory of the VUB. What striked me most was the lack of speed that these experiments exhibited. Since the problem was due to the exepensive evaluation of the individuals parallellisation seemed to me like the rigth answer.

Parallel lil-gp is nothing  more than what it says. It is a extension of lil-gp version 1.1 to can be used on a cluster as well as on a massively parallel computer. This relative platform independence of the parallellisation is due to the use of PVM. One of the objectives was to make the parallellisation transparant to the user. This in order to make it possible for people using the original version of lil-gp to make the transition without any problem. Next to that the expected increase in computation speed was meet as well and since this is what this is all about...

Advantages of Parallel Lil-gp

Parallel Lil-gp allows:

• Faster computations of a problem of the same size
• Experiments with bigger populations in the same time
• To use more expensive evaluations functions