Installation Guide¶
Requirements¶
MMSBM requires:
Python >= 3.7
NumPy
tqdm
Basic Installation¶
The easiest way to install the base mmsbm package is via pip:
pip install mmsbm
This will install the package with the default numpy backend.
Installing Optional Backends¶
For enhanced performance, you can install optional dependencies for accelerated backends.
Numba Backend (CPU JIT Compilation)¶
To enable the numba backend, install the Numba dependency:
pip install mmsbm[numba]
CuPy Backend (GPU Acceleration)¶
To enable the cupy backend for NVIDIA GPUs, you first need a compatible CUDA toolkit installed on your system. Then, you can install the required dependency:
pip install mmsbm[cupy]
Note
The CuPy backend requires Python 3.12 or lower. As of December 2025, CuPy does not provide pre-built wheels for Python 3.13+.
You can also install all extras at once:
pip install mmsbm[numba,cupy]
Development Installation¶
For development, you can install MMSBM from source:
git clone https://github.com/eudald-seeslab/mmsbm.git
cd mmsbm
pip install -e ".[dev]"
This will install additional dependencies needed for development:
pytest for testing
sphinx for documentation
black for code formatting
Verifying Installation¶
To verify the installation, you can run:
python -c "import mmsbm; print(mmsbm.__version__)"
Running Tests¶
To run the test suite:
python -m pytest tests/
Building Documentation¶
To build the documentation locally:
cd docs
make html
The documentation will be available in docs/_build/html/.
Troubleshooting¶
Common Issues¶
ImportError: No module named ‘mmsbm’
Make sure you’ve installed the package correctly:
pip install --upgrade mmsbm
Version conflicts
Try creating a new virtual environment:
python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install mmsbm
Getting Help¶
If you encounter any issues:
Check the troubleshooting guide
Search existing GitHub issues
Open a new GitHub issue with:
Your Python version
MMSBM version
Complete error traceback
Minimal example reproducing the issue