Conception of large multi-agents systems using statistical analysis : application to the behavior of complex social systems

We deal with systems using massive multi-agent organizations and expressing complex problems like the representation of behavior of social organizations. We propose an analysis and an operating representation of multi-agent organization in a geometric way, using specific multi-agent organization in a morphologic agent space. We propose also an architecture expressing the behavior of the massive multi-agent organization. So we open the way to the implementation of self-adaptive systems. We present an application for prediction of the migrations in a very large city.