Generalized Density-Functional Tight-Binding Repulsive Potentials from Unsupervised Machine Learning

Authors:

Abstract
We combine the approximate Density-Functional Tight-Binding (DFTB) method with unsupervised machine learning . This allows to improve transferability and accuracy, make use of large quantum chemical data sets for the parametrization, and to efficiently automatize the parametrization process of DFTB. For this purpose, generalized pair-potentials are introduced, where the chemical environmental is included during the learning process leading to more specific effective two-body potentials. We train on energies and forces of equilibrium and non-equilibrium structures of 2100 molecules, and test on ∼ 130.000 organic molecules containing O, N, C, H and F atoms. Atomization energies of the reference method can be reproduced within an error of ∼ 2.6 kcal/mol, indicating drastic improvement over standard DFTB.
Funding Information
  • Schweizerischer Nationalfonds zur F?rderung der Wissenschaftlichen Forschung (PP00P2_138932)
  • Deutsche Forschungsgemeinschaft (SFB 1249)

This publication has 42 references indexed in Scilit: