Self-Organisation in Agent-Based Simulation

In this paper, we investigate mechanisms via which a group of autonomous agents can produce emergent phenomena in artificial systems. In this framework, emergence deals with the appearance of new properties within a system of interacting agents. More particularly, emergence is here studied from a self-organisation point of view, that is a dynamic appearance of “remarkable” structures of agents. This work proposes an intentional modelling of mechanisms needed to give birth to structures in a multiagent system and a model of such processes. These mechanisms bring appropriate computational powers of expression to the system; they can be summarised by a meta-function and a local mechanism. The article emphasises the necessity of an emergent level to represent emergent structures, intermediary between the multiagent system and agents, the system being then distributed through three levels of description. To validate selforganisation processes, an application of simulation in Geophysics, showing the emergence of earthquakes is also presented. The paper concludes with some comments and issues for future