Metaheuristic Agent Processes (MAPS)

Agent-based models have had a remarkable impact in many areas of science, engineering and business. To achieve their full potential, however, these models must be extended to meet challenges of optimization that have so far been sidestepped or left unattended. Because classical optimization procedures are incapable of handling the complex problems that give rise to this challenge, a need arises for agent-based models to draw support from the field of metaheuristics.

Accordingly, this situation motivates the creation of Metaheuristic Agent Processes (MAPs) that integrate agent-based models with metaheuristic procedures, and thereby offer a means for achieving further advances through the use of agent-based technology. In this paper, we demonstrate that fundamental metaheuristic strategies already encompass inherent agent-based components, providing a natural foundation for the form of integration necessary to produce MAPs. In addition, we identify a particular class of discrete optimization models that exhibits useful links to agent-based systems, and whose successful applications invite further exploration within the MAP context.