This paper may be considered as a practical reference for those who wish to add (now sufficiently matured) Agent Based modeling to their analysis toolkit and may or may not have some System Dynamics or Discrete Event modeling background. We focus on systems that contain large numbers of active objects (people, business units, animals, vehicles, or even things like projects, stocks, products, etc. that have timing, event ordering or other kind of individual behavior associated with them). We compare the three major paradigms in simulation modeling: System Dynamics, Discrete Event and Agent Based. Modeling with respect to how they approach such systems. We show in detail how an Agent Based model can be built from an existing System Dynamics or a Discrete Event model and then show how easily it can be further enhanced to capture much more complicated behavior, dependencies and interactions thus providing for deeper insight in the system being modeled. Commonly understood examples are used throughout the paper; all models are specified in the visual language supported by AnyLogicTM tool. We view and present Agent Based modeling not as a substitution to older modeling paradigms but as a useful add-on that can be efficiently combined with System Dynamics and Discrete Event modeling. Several multi-paradigm model architectures are suggested.
Analytical (Static) and Simulation (Dynamic) Modeling