A DFT-driven multifidelity framework for constructing efficient energy models for atomic-scale simulations

The reliability of atomistic simulations depends on the quality of the underlying energy models providing the source of physical information, for instance for the calculation of migration barriers in atomistic Kinetic Monte Carlo simulations. Accurate (high-fidelity) methods are often available, but since they are usually computationally expensive, they must be replaced by less accurate (low-fidelity) models that introduce some degrees of approximation. Machine-learning techniques such as artificial neural networks can be employed to work around this limitation and extract the needed parameters from large databases of high-fidelity data. However, the latter are often computationally expensive to produce. This work introduces an alternative method based on the multifidelity approach. Correlations between high-fidelity and low-fidelity predictions are exploited to make an educated guess of the high-fidelity value based only on quick low-fidelity estimations, to be used for instance as an efficient and reliable source of physical data for atomistic simulations. With respect to neural networks, this approach requires less training data because of the lower amount of fitting parameters involved. The method is tested on the prediction of ab initio formation and migration energies of vacancy diffusion in iron-copper alloys, and compared with the neural networks trained on the same database.