A self adaptable evolutionary algorithm to tackle space planning problems
Résumé
In this paper we consider the space-planning problem, where a fixed set of objects have to be allocated on a single defined area. This is a Constraint Satisfaction Problem where the constraints are: the objects must be placed without overlapping. We have already designed a sequential evolutionary algorithm for space planning problem that has shown to be more performant than others approaches. Our aim is to be able to find more solutions than the sequential approach. The key idea is to design a self-adaptable distributed evolutionary algorithm. The first adaptation of our algorithm comes from the benefit of results already achieved by other nodes of the processors pool. In the paper we compare the first experimental results of our approach for solving space planning problem to the sequential ones. The implementation is running on an heterogeneous environment.