Solvent

Solvent is an open-source code for training highly accurate equivariant deep learning interatomic potentials for multiple electronic states.

Installation

Requires:

• Python >= 3.7
• CUDA >= 11.6

To install:

• Create virtual environment

python -m venv ./solvent_venv
source ./solvent_venv/bin/activate

• Install CUDA 11.6

wget https://developer.download.nvidia.com/compute/cuda/11.6.0/local_installers/cuda-repo-rhel7-11-6-local-11.6.0_510.39.01-1.x86_64.rpm
sudo rpm -i cuda-repo-rhel7-11-6-local-11.6.0_510.39.01-1.x86_64.rpm
sudo yum clean all
sudo yum -y install nvidia-driver-latest-dkms cuda
sudo yum -y install cuda-drivers

• Install torch with CUDA 11.6

pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116

• Install torch - nightly build for vmap - with CUDA (optional)

pip install --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cu116

• Install torch-geometric, torch-scatter, torch-cluster

pip install torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric -f https://data.pyg.org/whl/torch-1.12.0+cu116.html

• Install the rest of the dependencies:

pip install -r ./requirements.txt

To run demo:

cd ./demo
python example_training_from_preload.py