"Multiagent Dynamical Systems"

08/27/04: 10:30 AM
11th Floor Large Conference Room
Host: Kristina Lerman, schedule

Abstract: How is the collective functioning of intelligent agents different from, but still the consequence of, the agents' local, designed behavior? I'll review recent work on multiagent dynamical systems---a class of macroscopic differential equations that describe collective adaptation, but are derived from a discrete-time stochastic microscopic model. The behavior of each agent is a dynamic balance between adaptation that locally achieves the best action (exploitation) and memory loss that leads to randomized behavior (exploration). Although individual agents interact with their environment and other agents in a purely self-interested way, the macroscopic behavior that emerges can be interpreted as game dynamics. Application to several familiar, explicit game interactions shows that the adaptation dynamics exhibits a diversity of collective behaviors, including stable limit cycles, quasiperiodicity, intermittency, and deterministic chaos. The simplicity of the assumptions underlying the macroscopic equations suggests that these behaviors should be expected broadly in collective adaptation.

About Jim Crutchfield: James P. Crutchfield received his B.A. summa cum laude in Physics and Mathematics from the University of California, Santa Cruz, in 1979 and his Ph.D. in Physics there in 1983. He is currently a Research Professor at the Santa Fe Institute, where he runs the Dynamics of Learning Group in which post-doctoral researchers work and Ph.D. students pursue research on their dissertations. He is also Adjunct Professor of Physics in the Physics Department, University of New Mexico, Albuquerque. Before coming to SFI in 1997, he was a Research Physicist in the Physics Department at the University of California, Berkeley, since 1985. He has been a Visiting Research Professor at the Sloan Center for Theoretical Neurobiology, University of California, San Francisco; a Post-doctoral Fellow of the Miller Institute for Basic Research in Science at UCB; a UCB Physics Department IBM Post-Doctoral Fellow in Condensed Matter Physics; a Distinguished Visiting Research Professor of the Beckman Institute at the University of Illinois, Urbana-Champaign; and a Bernard Osher Fellow at the San Francisco Exploratorium. Over the last two decades Prof. Crutchfield has worked in the areas of nonlinear dynamics, solid-state physics, astrophysics, fluid mechanics, critical phenomena and phase transitions, chaos, and pattern formation. His current research interests center on computational mechanics, the physics of complexity, statistical inference for nonlinear processes, genetic algorithms, evolutionary theory, machine learning, and distributed intelligence. He has published over 90 papers in these areas; most are available from his website: www.santafe.edu/~chaos.

Last updated: Mon Jun 19 17:44:06 2006