Overview of Agent-Based Models

Agent-based modeling (ABM) is an approach to understanding complex social, economic, and environmental systems. It is a simulation methodology that focuses on the interactions of individuals or households (agents) and the ways in which they interact to form systems. In this approach, the properties of complex systems arise from interactions between the agents, and this the modeled systems display emergent properties that cannot be induced simply by aggregating the characteristics of the individuals, nor deduced by indentifying the structures of the systems. Agent-based modeling is an important aid to research on coupled human-natural system dynamics and relationships between humans and the built environment; the investigation of changes at various spatial, temporal and social scales; and the development of improved spatially explicit land-change models that integrate people, place and environment.

Modeling Environment

We have chosen the Repast (Recursive Porous Agent Simulation Toolkit) modeling toolkit as the development environment for our agent-based models. Repast is one of the leading ABM development environments, and there is a large user base on the web that provides a substantial support network. Its use has been well established in the scientific literature.

Repast is a framework in which researchers can add components to create their own agent-based simulation models. This framework provides all of the basic functionality required to support and execute an ABM, including the scheduling mechanisms and a vast collection of modeling functions which are included in the Repast libraries. More information about Repast can be found at their website (http://repast.sourceforge.net).

Programming Languages

All of the code available on this site is written in Java and designed to operate within the Repast J modeling environment. This code, however, can also be used in models built within the Repast Py modeling environment or within the Agent Analyst extension for ArcGIS.