sGDML Symmetric Gradient Domain Machine Learning


Towards Exact Molecular Dynamics Simulations with Machine-Learned Force Fields

Chmiela, S., Sauceda, H. E., Müller, K.-R., Tkatchenko, A., Nature Communications, 9(1), 2018, 3887.

Article Supplement BibTex

Machine Learning of Accurate Energy-conserving Molecular Force Fields

Chmiela, S., Tkatchenko, A., Sauceda, H. E., Poltavsky, Igor, Schütt, K. T., Müller, K.-R., Science Advances, 3(5), 2017, e1603015.

Article Supplement BibTex


Replicate our numerical results with a Python implementation of sGDML.

GitHub Documentation

Validate a pre-trained model on one of the published datasets:

$ model aspirin
$ dataset aspirin
$ sgdml validate aspirin.npz aspirin_dft.npz 10000


All geometries in Å, energy labels in kcal mol-1 and force labels in kcal mol-1 Å-1.
Name Size Download
DFT [FHI-aims, light tier 1]
Benzene (from Chmiela et al., 2018) 49,863
Benzene (from Chmiela et al., 2017) 627,983
Uracil 133,770
Naphthalene 326,250
Aspirin 211,762
Salicylic acid 320,231
Malonaldehyde 993,237
Ethanol 555,092
Toluene 442,790
NEW Paracetamol 106,490
NEW Azobenzene 211,762
CCSD [Psi4, cc-pVDZ]
NEW Aspirin 1,500
CCSD(T) [Psi4, cc-pVDZ]
NEW Benzene 1,500
NEW Malonaldehyde 1,500
NEW Toluene 1,500
[Psi4, cc-pVTZ]
NEW Ethanol 2,000