Abstract. This paper describes GAPP – a framework for the execution of distributed genetic algorithms (GAs) using the H2O metacomputing environment. GAs may be a viable solution technique to intractable problems; GAPP offers a distributed GA framework that can lead to rapid and efficient parallel execution of GAs from a variety of domains, with very little effort on behalf of the application scientist. It is premised upon the common phases embodied in GA lifecycles and contains modular implementations to handle each of them, whereas end applications simply provide domain-specific functions and parameters. GAPP is built for H2O, a component-oriented metacomputing system that enables cooperative resource sharing and flexible, reconfigurable concurrent computing on heterogeneous platforms. Experiences with the use of GAPP on H2O are described and preliminary results are very encouraging.
Tomasz Ampula, Dawid Kurzyniec, Vaidy S. Sunderam,