Installation

PolyMon requires Python 3.8+ and several dependencies including PyTorch and PyTorch Geometric.

Prerequisites

PyTorch Installation

PolyMon requires torch>=2.2.2 and torch_geometric>=2.5.3. We recommend installing PyTorch first with CUDA support if available.

For CUDA 11.8:

conda install -y pytorch==2.3.0 torchvision==0.18.0 torchaudio==2.3.0 \
                 pytorch-cuda=11.8 -c pytorch -c nvidia
pip install torch_geometric
pip install torch_scatter torch_sparse -f https://data.pyg.org/whl/torch-2.3.0+cu118.html

For CPU only:

conda install -y pytorch==2.3.0 torchvision==0.18.0 torchaudio==2.3.0 -c pytorch
pip install torch_geometric
pip install torch_scatter torch_sparse -f https://data.pyg.org/whl/torch-2.3.0+cpu.html

For other CUDA versions, please refer to the PyTorch Geometric installation guide.

Install PolyMon

Via pip (recommended):

pip install polymon

From source:

git clone https://github.com/fate1997/polymon.git
cd polymon
pip install -e .

Verify Installation

After installation, verify that PolyMon is working:

polymon --help

You should see the help message with available commands: train, rec, and predict.

Optional Dependencies

For full functionality, the following packages are automatically installed:

  • RDKit: Chemical informatics and molecular featurization

  • Mordred: Additional molecular descriptors

  • XenonPy: Element-based descriptors

  • XGBoost/LightGBM/CatBoost: Gradient boosting frameworks

  • Optuna: Hyperparameter optimization

  • PyTorch Lightning: Training utilities

Troubleshooting

Issue: ImportError: No module named 'torch_geometric'

Solution: Make sure you installed PyTorch Geometric after PyTorch. Reinstall if needed:

pip install torch_geometric --force-reinstall

Issue: NumPy/AttributeError warnings when running polymon commands

Solution: RDKit is not yet compatible with NumPy 2.x. Downgrade NumPy:

pip install 'numpy<2'

Issue: CUDA errors

Solution: Verify your PyTorch installation includes CUDA support:

python -c "import torch; print(torch.cuda.is_available())"

If False, reinstall PyTorch with CUDA support using the commands above.