Installation¶

From PyPI¶

PyDSMC is published on PyPI:

pip install pydsmc

We recommend using a virtual environment and officially tested Python versions 3.10, 3.11, and 3.12.

Only the lightweight core dependencies (numpy, scipy, gymnasium, pandas, packaging) are installed by default. In particular, PyDSMC does not pull in a deep-learning framework, so you are free to bring your own (PyTorch, JAX, TensorFlow, …).

Optional extras¶

PyDSMC ships a few optional dependency groups:

Extra	Install	Contents
`examples`	`pip install "pydsmc[examples]"`	Environments/agents used by the bundled examples (`minigrid`, `ale-py`, `gymnasium-robotics`, `sb3-contrib`).
`dev`	`pip install "pydsmc[dev]"`	Test & tooling stack (`pytest`, `pre-commit`, `stable-baselines3`, `pgtg`, `ipython`).
`docs`	`pip install "pydsmc[docs]"`	Documentation stack (`mkdocs-material`, `mkdocstrings`, `mkdocs-marimo`).
`all`	`pip install "pydsmc[all]"`	Everything above.

Development setup (with mise)¶

This project uses mise to manage the Python toolchain and common tasks, backed by uv for fast installs. If you want to work on PyDSMC:

git clone https://github.com/neuro-mechanistic-modeling/PyDSMC.git
cd PyDSMC

# Installs the pinned Python + uv, creates the .venv, installs all extras, and sets up pre-commit.
mise run install

Handy tasks (mise tasks lists them all):

Task	Description
`mise run install`	Install all extras (`pydsmc[all]`) and pre-commit hooks.
`mise run test`	Install dev deps and run the test suite.
`mise run docs [--serve]`	Build [Serve] the documentation.
`mise run info`	Print project/environment information.

Prefer a manual setup? A classic virtual environment works too:

python3.11 -m venv .venv
source .venv/bin/activate
pip install -e ".[all]"
pre-commit install

See Contributing for the full contributor workflow.