QMLearn is a pure-Python package. It needs Python ≥3.8, NumPy, SciPy, scikit-learn, ASE, h5py, and a quantum-chemistry engine (PySCF by default).
The development branch is dev on GitHub. Pin versions from requirements.txt if you need a reproducible stack.
Clone the repository and install in the current environment:
git clone https://github.com/Quantum-MultiScale/QMLearn.git
cd QMLearn
python -m pip install .
Or install directly from the remote without a local clone:
python -m pip install git+https://github.com/Quantum-MultiScale/QMLearn.git
To track the development branch explicitly: python -m pip install git+https://github.com/Quantum-MultiScale/QMLearn.git@dev.
python3 -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
python -m pip install -U pip
python -m pip install git+https://github.com/Quantum-MultiScale/QMLearn.git
python -m pip install matplotlib jupyter
python -c "import qmlearn; print(qmlearn.__version__)"
You should see a version string (currently 0.0.1). Next, generate a training database and fit a model: see the tutorials.
PySCF is the default engine for GTO integrals, 1-RDMs, and 2-RDMs. Psi4 is available as an alternative engine in qmlearn.drivers.psi4.