What is agent-based modeling and why might it be useful?

Uri Wilensky, Ph.D.
Northwestern University

ABSTRACT:
Agent-based Modeling (ABM) is a methodology that enables modelers to create models that connect the micro and macro level. They represent the micro-level as agents with characteristic behaviors and as the agents execute their behaviors, macro-patterns emerge. Thus they enable a natural representation of "emergent phenomena",  which are notoriously difficult to comprehend. With ABM, we can model a wide variety of phenomena,  such as the emergence of galaxies from the interactions of stars, the emergence of traffic jams from the behaviors of drivers, economic patterns from the behavior of buyers and sellers, spreading of diseases from individual contacts on a network. We discuss how ABM methodology can be used to get insights into emergent properties of systems, and to create models of heterogeneous agents that can be readily adapted to changing conditions or hypotheses.