Course: agent-based modelling with Modelly

Eight sessions that take a class from a first model to a validated, reproducible study of its own. Every session is built on a model of the zoo, so the students start from code that is known to be right and spend their time on modelling questions. Each session lists its goals, what to read before, and exercises; the solutions are folded under each exercise.

Session	Topic	Zoo model
1	Agents, a scenario and a step	Boltzmann wealth exchange
2	Data, seeds and reproducibility	Boltzmann, Bouchaud-Mezard
3	Space: grids, rasters and percolation	Forest fire
4	Networks and contact structure	Bass diffusion, age-structured SEIR
5	Thinking in arrays: columns and cohorts	Nagel-Schreckenberg, SEIR
6	Validation against theory	SEIR, traffic, Bass
7	Experiments: sweeps and replications	Deffuant, Hegselmann-Krause, Axelrod
8	From the zoo to your own project	any

Running the course

• Setup. pip install "modelly-abm[viz]" and a Jupyter environment. The commands in the sessions are the same in bash and PowerShell unless a session shows both.

• Format. Each session is a lecture on the topic, a live walk through the zoo model, then the exercises in pairs. The domain tutorials are ready-made walk-throughs.

• Assessment. A small project, started in session 8 from a zoo model with modelly zoo copy, and presented with its validation test, a seeded experiment and one figure.