Agent Based Modeling

A couple of paragraphs from call for papers of Journal of Simulation:

Agent-based models (ABMs), also sometimes called agent distillations or cellular automata models, have found tremendous success in an especially wide spectrum of applications. Such models have also been used in financial modeling, personnel modeling, in military applications such as examining the impacts of tactical airpower or logistical supply, in social science modeling and even in a variety of manufacturing applications. Multiagent systems (MAS) are a special kind of ABM in which a group of systems or entities interact with each other and their environment. An agent in a MAS can represent a human, group of humans, a machine or a software system with the capability to perceive their environment, react to changes in the environment and affect the environment with their own actions.

In an ABM, entities are constructed to have specified goals or actions. Agents can control their own destiny which means they may change their internal state depending on their knowledge of the artificial environment in which they reside and function. This cognitive capability of the agents is necessary so agents can achieve the specified goal or criterion within the context of the application. The agents are usually modeled in an object-oriented manner but suitably extended to include representation of their knowledge and roles within an environment. Thus, an agent can have knowledge of itself, knowledge received from other agents through communication and sensory channels, knowledge based on perceptions of its environment, and even memory of previous states the agent found itself. Among the challenges to ABM are designing these agents to encapsulate this information in such a way that the knowledge is controlled and exploited by the software agent, ensuring the ABM realistically captures the actual system of interest, and analyzing the emergent behavior that often arises in the use of these models.