1.4.2: Tools for other paradigms#
Agent-based modelling is not the only simulation paradigm, and the two main alternatives, system dynamics (SD) and discrete-event simulation (DES), have their own established tools. They are listed here for comparison [1]. But there is a payoff beyond the roster itself: because each tool is built around one paradigm, the way a tool represents the world is a signature that tells you which paradigm it serves. Learning to read that signature lets you size up unfamiliar software quickly, so this page pairs the catalogue of tools with the skill of recognising a tool’s paradigm from how it works.
1.4.2.1: The tools#
The tabs below are a short reference catalogue of the main SD and DES tools, each with the interface signature that gives away its paradigm and the kind of work it is typically used for. The comparison is not only about software names. It shows what each paradigm treats as the basic unit of a model: stocks and flows in system dynamics, events and resource queues in discrete-event simulation, and individual entities in agent-based modelling. That underlying representation determines which questions are natural to express and which require awkward workarounds [1].
When comparing tools, look beyond whether an interface appears easy to use. Consider how the tool represents state and time, what data it can exchange, how experiments are automated, and whether its outputs connect to the analysis the project requires. Those practical capabilities matter only after the modelling paradigm fits the problem.
Vensim is a long-standing SD package, widely used in policy analysis, public health, and resource management, and free for educational use. Stella / iThink is a more visual, teaching-oriented SD environment. Both use stock-and-flow diagrams as the main interface: the modeller draws the system’s structure, and the tool generates the underlying differential equations [1].
SimPy is a free, open-source DES library for Python; because it shares the Python ecosystem with Mesa, the two can be combined. AnyLogic is a commercial, multi-paradigm platform supporting ABM, SD, and DES in one model, widely used in industry for logistics, healthcare, and transport, with academic licences available [1].
1.4.2.2: Reading a tool’s paradigm#
The tools above split cleanly along one line: software tends to be built around a single simulation paradigm, and knowing which paradigm a tool serves tells you a great deal about what it is good for and where it will resist you. System-dynamics packages think in stocks and flows, discrete-event tools think in queues and events, and agent-based tools think in individuals and their interactions, so reaching for the wrong one means forcing a problem into a shape it does not fit. Being able to place a named tool in its paradigm is a quick way to orient yourself when you meet unfamiliar software.
The activity below puts that skill to work across all three paradigms, so it also draws in the ABM tools introduced in 1.4.1: Agent-based modelling platforms: NetLogo, Mesa and GAMA all model individuals and their interactions. Match each tool to the paradigm it is designed for.
The activity above named the tools for you. In practice you often meet software cold, and the skill that transfers is reading a tool’s paradigm from how it represents the world, with no name to lean on. The activity below isolates exactly that: each row is an interface signature rather than a product.
Learning activity: read the signature#
Read this way, an unfamiliar tool announces its paradigm through its interface long before you commit to it: stocks and flows point to system dynamics, queues and events to discrete-event simulation, and interacting individuals to agent-based modelling.
With the ABM platforms of Section 1.4.1 and the other-paradigm tools here on the table, 1.4.3: How to choose a tool turns to the practical question they raise: how to choose the right tool for a given project.
Key takeaways#
Key takeaways
System dynamics tools (Vensim, Stella) work through stock-and-flow diagrams; discrete-event tools (SimPy, AnyLogic) work through queues and events.
Each tool announces its paradigm through the way it represents the world, so recognising that signature lets you size up unfamiliar software before you have written a line in it.
SimPy shares Python with Mesa; AnyLogic can combine all three paradigms in a single model.