Graph simulation software: how to choose

Search for graph simulation software and you will find tools that have almost nothing in common. Some generate graphs. Some analyse them. Some simulate computer networks. Some simulate a system whose structure is a graph. This guide sorts them into four groups so you can tell which one you actually need. If the term itself is new to you, start with "What is graph simulation?".

Start with the question you are asking

The right category depends on what the graph is for.

You want to...	Look for	Examples
Generate synthetic graphs to test an algorithm	Graph generators	Libraries such as NetworkX, which can produce random, scale-free and small-world graphs
Study the structure of an existing network	Graph analysis and visualisation	Gephi, NetworkX, igraph
Test a computer network design	Network emulators	GNS3
Watch a system change over time, where parts affect each other	Dynamic-system simulation on a graph	Konjugate

None of these is better in general. They answer different questions, and the confusion in search results comes from all of them using the word "graph".

Graph generators

These produce graphs as output: nodes and edges with a chosen statistical shape. Use them when the graph is your test data. They do not model anything evolving on the graph, although you can build that yourself on top.

Graph analysis and visualisation

These take a graph you already have and measure it: degree distributions, communities, shortest paths. They answer "what is the structure?" rather than "what will happen next?".

Network emulators

Tools in this group simulate computer networks: routers, links and protocols. The word network is literal here. They are built for one domain.

Dynamic-system simulation on a graph

Here the graph is the model. Each node holds state such as a temperature, a stock level or a balance. Each edge is a relationship describing how one node's state affects another's. The software steps the whole structure through time and shows you the trajectory.

Look for these properties when you evaluate a tool in this group:

• Visible structure. You can see which components depend on which, rather than reading one large block of equations.
• Composability. You can build subsystems separately and connect them.
• Flexible relationships. You can write relationships as equations or as code, and change one without touching the rest.
• Speed. Large graphs run in reasonable time, ideally on a native engine.
• Inspectable results. You can explore how each state evolved, not only the final values.
• Openness. You can read the source, export the model and run it outside the app.

Konjugate is open-source software in this last group. You build the graph visually, write relationships as equations or in C++ or Python, and run it on a native C++ engine. It can also export a simulation as standalone code and recover a graph from measured data. It is early-stage software, so check the docs for what is supported today.

A quick decision rule

• If your output is a graph, you want a generator.
• If your input is a graph and your output is a measurement, you want an analysis tool.
• If your subject is a computer network, you want an emulator.
• If your input is a graph of components and your output is behaviour over time, you want dynamic-system simulation software.

To try the last option, open Konjugate in your browser or follow "Getting started with Konjugate".