This one-day seminar provides a comprehensive introduction to agent-based simulation, which is currently the "hottest" topic in simulation modeling. In an agent-based simulation, autonomous agents (people, vehicles, organizations, etc.), which have attributes and potentially complex behaviors, interact with each other and their environment over time toward the accomplishment of their goals. This allows an agent's behavior to depend on the current state of its environment, rather than being "scripted" as in many traditional models. The interactions of the "low-level" agents often result in complex emergent behavior for the system as a whole.
Agent-based simulation has been successfully applied to a diverse set of problems, and improved software packages have facilitated the model-development process. However, learning agent-based simulation on one's own is difficult, at best, due to the real lack of technically precise literature on the subject. The development of this seminar has benefited from funding by the U.S. Army.
Recent attendees include Aerospace Corporation, Allstate, AT&T, BAE Systems, Boeing, CACI, Canadian Air Force, Center for Naval Analysis, Computer Sciences Corporation, General Motors, Institute for Defense Analyses, Johns Hopkins University Applied Physics Lab, Joint Warfare Analysis Center, Lockheed Martin, Memorial Sloan-Kettering Cancer Center, NASA, Raytheon, Rockwell, RTI International, Sikorsky, University of Central Florida, University of Maryland, University of Virginia, U.S. Army, U.S. Coast Guard Academy, U.S. Navy, and Zurich Airport.
What You Will Learn:
1. Agents and Agent-Based Simulation
Agents as autonomous entities with attributes and complex behaviors
Reactive versus adaptive (learn from previous experiences) agents
Bottom-up modeling and the resulting emergent system-level behavior
When to use agent-based simulation (ABS)
Relationship of ABS to discrete-event simulation (DES)
2. The Structure of an Agent and Time-Advance Mechanisms
Agents with memory, rules, and a behavior engine
Time stepping versus next-event time advance
3. Historical Perspective
Cellular automata
Schelling's segregation model
Growing artificial societies
Complex adaptive systems
4. Software for Agent-Based Simulation
Commercial DES software with capabilities for ABS
Public-domain toolkits for ABS and their modeling flexibility, ease of use, features, and quality of documentation
Important defense-related ABS
5. Development of Several Agent-Based Simulation Models
Example to illustrate how to implement basic reactive rules for agents
Example to illustrate learning and adaptation for agents
6. Validation of Agent-Based Simulation Models
Developing an assumptions document (AD)
Performing a structured walk-though of the AD
Comparing model output data with the comparable output data from a similar existing system
Confirming agent behaviors and emergent structures
7. Design and Analysis of Simulation Experiments
Specifying the simulation run length and the number of independent replications
Determining model sensitivities using parameter sweeps
8. Successful Applications of Agent-Based Simulation
Defense (e.g., irregular warfare)
Homeland security (evacuation of crowds, border control, etc.)
Traffic modeling
Epidemiology and bio-warfare
Supply chains
Consumer behavior
Sociology
Economics
Anthropology
And many more ...
Public Simulation Seminars Home Page
