BIGDML: Towards Exact Machine Learning Force Fields for Materials
This work addresses the problem of practical applicability of MLFFs for materials scientists by reducing data requirements and improving accuracy, though it is incremental as it builds on existing gradient-domain methods.
The paper tackles the challenge of creating accurate and data-efficient machine-learning force fields (MLFFs) for materials by introducing the BIGDML approach, which achieves state-of-the-art energy accuracies with errors below 1 meV per atom using only 10-200 training geometries for various materials and adsorbates.
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 allow to rationalize the Arrhenius behavior of hydrogen diffusion coefficient in a Pd crystal for a wide range of temperatures.