A simulation application may be modeled as a set of interacting
	entities within an environment. Such applications can b e represented as a
	graph with a one-to-one mapping between vertices and entities and between
	edges and communications. As for classical applications, performances
	depend directly on a good load balancing of the entities between available
	computing devices and on the minimization of the impact of the
	communications b etween them. However, both objectives are contradictory
	and good performances may be achieved if and only if a good trade-o? is
	found. Our method for ?nding such a trade-o? leans on a bio-inspired
	method. We use comp etitive colonies of numerical ants, each one depositing
	colored pheromones, to ?nd organizations of highly communicating
	entities.