1
Define Individuals
Identify who the participants are and what state they possess - such as position, health status, speed, and resource levels.
Agent-Based Modeling
Observe How Collective Behavior Emerges from Individual Agents
Agent-based modeling doesn't start with aggregate equations. Instead, it defines "how each individual perceives, acts, and interacts". When these simple rules run simultaneously, emergent phenomena like congestion, diffusion, aggregation, and propagation emerge.
Individual Rules
Local Interaction
Spatial Movement
Emergent Behavior
Don't Think "Big System" First
The key to ABM isn't writing aggregate formulas first. It's defining each individual's rules clearly, then letting them interact together.
Who Are Agents?
People, vehicles, businesses, cells, animals - any independent decision-making unit.
What Are Rules?
Moving, contacting, infecting, collaborating, avoiding, buying - all are local rules.
How Does Environment Affect?
Spatial position, density, neighbor relationships, and constraints change outcomes.
What Do You Measure?
Not individual agents, but aggregate diffusion curves, clustering patterns, and collective behavior.
Understanding the Method
ABM's Core is "Bottom-Up"
Start by giving each individual a set of local rules, then let the system evolve on its own. The model's explanatory power comes from how these local rules combine into macro-level outcomes.
2
Define Local Rules
Each agent makes decisions based only on itself and its immediate neighbors, not the entire system.
3
Iterate Over Time
The system updates individual states at each time step. Contacts and feedback between agents accumulate over time.
4
Observe Emergent Results
Diffusion speed, peak size, clustering patterns, and final equilibrium all emerge naturally from individual behaviors.
Teaching Experiment on This Page
We use "infection spreading during movement" to demonstrate ABM: The rules are simple, but the results are complex - this is ABM's most important insight.
Interactive Lab
Adjust Local Rules, Observe How the Spread Curve Changes
Each dot is an agent. You're not adjusting an "aggregate infection rate" - you're adjusting population size, movement speed, contact radius, and infection probability.
Agent Spatial Evolution
Each dot is an independent agent. The system doesn't know the outcome in advance - it emerges step by step from local rules and spatial contacts.
Susceptible
Infected
Recovered
Population State Curve
This curve is typical ABM output: Macro results aren't directly input - they're aggregated from individual behaviors.How to Interpret Results
ABM is best for answering: "How does the whole change if local rules change?"Once the model starts running, interpretations based on current state will appear here.
When to Use
When Individual Differences, Neighbor Relationships, and Spatial Contacts Matter
If the key isn't averages but "who meets whom, who acts first, and how rules interact", ABM is often more natural than aggregate models.
Individuals Are Different
When participants have behavioral, informational, or spatial differences, an "average person" isn't sufficient.
Local Relationships Determine Outcomes
Spread, traffic, collaboration, market choices, and collective behavior often depend on local contact networks, not global averages.
You Care About Emergent Phenomena
Phenomena like congestion, clustering, synchronization, diffusion, and differentiation are well-suited for agent-based modeling.