We study the temporal evolution of populations including thousands of mobile repulsive agents. In the first part, the agents follow the rules of a deterministic automaton. We show that, the population may evolve toward different types of high-symmetry steady distributions, depending on the agent-agent repulsion law, and on the presence (or absence) of borders around the playground. We observe the formation of hexagonal-condensed lattices that maximize the potential function of the agent distribution. In the second part, agents are faulty. They violate the repulsion law and follow the rule of a non-deterministic automaton. We observe fast and random oscillations between the ordered phase and a strongly disorganized state when the number of agents is small (say for population involving a few tens of agents). These oscillations are essentially averaged and disappear in the large populations, resulting in a partially-ordered homogeneous distribution
Published in:
Distributed Intelligent Systems: Collective Intelligence and Its Applications, 2006. DIS 2006. IEEE Workshop on
Date of Conference: 15-16 June 2006