ParaDisEO-Based Design of Parallel and Distributed Evolutionary Algorithms

S. Cahon (1) , N. Melab (1) , El-Ghazali Talbi (1) , Marc Schoenauer (2)

1 LIFL - Laboratoire d'Informatique Fondamentale de Lille
2 FRACTALES - Fractals, complex models and artificial evolution

ParaDisEO is a framework dedicated to the design of parallel and distributed metaheuristics including local search methods and evolutionary algorithms. This paper focuses on the latter aspect. We present the three parallel and distributed models implemented in ParaDisEO and show how these can be exploited in a user-friendly, flexible and transparent way. These models can be deployed on distributed memory machines as well as on shared memory multi-processors, taking advantage of the shared memory in the latter case. In addition, we illustrate the instantiation of the models through two applications demonstrating the efficiency and robustness of the framework.

Domaines: Intelligence artificielle [cs.AI]

S. Cahon, N. Melab, El-Ghazali Talbi, Marc Schoenauer. ParaDisEO-Based Design of Parallel and Distributed Evolutionary Algorithms. Evolution Artificielle, Oct 2003, Marseille, France. pp.216-228, ⟨10.1007/b96080⟩. ⟨inria-00201075⟩