Distinguishing Environmental and Agent Dynamics: A Case Study in Abstraction and Alternate Modeling Technologies

Two factors can confound the interpretation of an MAS application or model. First, the MAS’s dynamics interact in complex ways with those of its environment, and agent engineers need to distinguish the two. Second, “mean field” approximations of the behavior of the system may be useful for qualitative examination of the dynamics, but can differ surprisingly from the behavior that emerges from the interactions of discrete agents. This paper examines these effects in the context of a project applying synthetic pheromones for agent-based control to a military air operations scenario.