Building and assurance of agent-based models: An example and challenge to the field

Abstract The assurance, that is, the verification and validation, of agent-based models is difficult, because of the heterogeneity of agents, and the possibility of the emergence of new patterns of macro behavior as a result of the interactions of these agents at the micro-level. This paper uses an agent-based model of the complex interactions among consumers, retailers, and manufacturers to explore issues of model assurance. These explorations involve two challenges for the agent-based model's field. The first challenge is to address the critical issue of software verification. The second challenge is to overcome the many methodological obstacles that exist in empirically validating these models. This paper will outline some of them. The authors propose a method based on the Genetic Algorithm to address both these challenges, but the experiments required, and a lack of good data on many kinds of agents, generally call for a minimalist approach to building and assuring agent-based models.