Towards more efficient and performant computations in quantum chemistry with machine learning

Pronobis, Wiktor

doi:10.14279/depositonce-9866

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the second option, the mathematical form of the BIGDML predictor can be reformulated by, for example, reducing the many-body complexity of the D PBC descriptor to keep interactions only up to a certain body order. This approach has been proven to give good results in molecules 102,103 . Despite current limitations of BIGDML, having access to a MLFF that can robustly represent global interactions in extended materials is a substantial achievement, as shown via extensive simulations in Section.…”

Section: Discussionmentioning

confidence: 98%

BIGDML—Towards accurate quantum machine learning force fields for materials

et al. 2022

View full text Add to dashboard Cite

Machine-learning force fields (MLFF) should be accurate, computationally and data efficient, and applicable to molecules, materials, and interfaces thereof. Currently, MLFFs often introduce tradeoffs that restrict their practical applicability to small subsets of chemical space or require exhaustive datasets for training. Here, we introduce the Bravais-Inspired Gradient-Domain Machine Learning (BIGDML) approach and demonstrate its ability to construct reliable force fields using a training set with just 10–200 geometries for materials including pristine and defect-containing 2D and 3D semiconductors and metals, as well as chemisorbed and physisorbed atomic and molecular adsorbates on surfaces. The BIGDML model employs the full relevant symmetry group for a given material, does not assume artificial atom types or localization of atomic interactions and exhibits high data efficiency and state-of-the-art energy accuracies (errors substantially below 1 meV per atom) for an extended set of materials. Extensive path-integral molecular dynamics carried out with BIGDML models demonstrate the counterintuitive localization of benzene–graphene dynamics induced by nuclear quantum effects and their strong contributions to the hydrogen diffusion coefficient in a Pd crystal for a wide range of temperatures.

show abstract

Section: Discussionmentioning

confidence: 98%