Accurate machine learning force fields via experimental and simulation data fusion

TL;DR

This paper introduces a data fusion approach combining experimental measurements and DFT calculations to train more accurate machine learning force fields for materials, exemplified on titanium.

Contribution

It presents a novel data fusion strategy that improves ML potential accuracy by integrating experimental and simulation data, correcting DFT inaccuracies.

Findings

Fused data training yields higher accuracy models.

Corrects DFT functional inaccuracies at target properties.

Applicable to any material for improved ML potentials.

Abstract

Machine Learning (ML)-based force fields are attracting ever-increasing interest due to their capacity to span spatiotemporal scales of classical interatomic potentials at quantum-level accuracy. They can be trained based on high-fidelity simulations or experiments, the former being the common case. However, both approaches are impaired by scarce and erroneous data resulting in models that either do not agree with well-known experimental observations or are under-constrained and only reproduce some properties. Here we leverage both Density Functional Theory (DFT) calculations and experimentally measured mechanical properties and lattice parameters to train an ML potential of titanium. We demonstrate that the fused data learning strategy can concurrently satisfy all target objectives, thus resulting in a molecular model of higher accuracy compared to the models trained with a single data source. The inaccuracies of DFT functionals at target experimental properties were corrected, while the investigated off-target properties remained largely unperturbed. Our approach is applicable to any material and can serve as a general strategy to obtain highly accurate ML potentials.