An agent is the colloquial term for pretty much any component in an ABM that has extent. Typically agents are motivated to do something, but sometimes other objects in a simulation will also be called "agents".

Agents have: - Internal data representations (memory or state) - Means for modifying their internal data representations (perceptions) - Means for modifying their environment (behaviors)

Why agent-based modelling?

traditional mathematical methods (ODE, PDE, statistical approaches): - can describe macroscopic properties of a system that is already known, but don't explain the origin of those properties. (e.g. rate constants) - cannot be easily extrapolated to situations where the assumptions behind the equations no longer hold. (e.g. Hookes law F = -kx) - don't handle discontinuous systems well - don't handle heterogenity in populations well

but: ABM complements and enhances rather than supplants, traditional approaches.

Means for agents to interact: - Direct spatial interaction - Indirect spatial interaction ownership resource depletion pheromone dispersal - Communication - Transactions (e.g. financial)