SchNet - a deep learning architecture for quantum chemistry

Important: This package will not be further developed and supported. Please consider switching to our new pytorch-based package SchNetPack!

SchNet is a deep learning architecture that allows for spatially and chemically resolved insights into quantum-mechanical observables of atomistic systems.

Requirements:

• python 3.4
• ASE
• numpy
• tensorflow (>=1.0)

See the scripts folder for training and evaluation of SchNet model for predicting the total energy (U0) for the GDB-9 data set.

Install

python3 setup.py install

Examples

QM9

Download and convert QM9 data set:

python3 load_qm9.py <qm9destination>

Train QM9 energy (U0) prediction:

python3 train_energy_force.py <qm9destination>/qm9.db ./modeldir ./split50k.npz 
    --ntrain 50000 --nval 10000 --fit_energy --atomref <qm9destination>/atomref.npz

Potential energy surface

Predict force and energy for fullerene C20 configuration

python scripts/example_md_predictor.py ./models/c20/ ./models/c20/C20.xyz

Relax geometry:

python scripts/example_md_predictor.py ./models/c20/ ./models/c20/C20.xyz --relax

References

If you use SchNet in your research, please cite:

K.T. Schütt. P.-J. Kindermans, H. E. Sauceda, S. Chmiela, A. Tkatchenko, K.-R. Müller.
SchNet: A continuous-filter convolutional neural network for modeling quantum interactions.
Advances in Neural Information Processing Systems 30, pp. 992-1002 (2017)

K.T. Schütt. F. Arbabzadah. S. Chmiela, K.-R. Müller, A. Tkatchenko.
Quantum-chemical insights from deep tensor neural networks.
Nature Communications 8. 13890 (2017)
doi: 10.1038/ncomms13890