Smart energy models for atomistic simulations using a DFT-driven multifidelity approach

Abstract

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 are usually employed to work around this limitation and extract the needed parameters from large databases of high-fidelity data, but the latter are often computationally expensive to produce. This work introduces an alternative method based on the multifidelity approach, where correlations between high-fidelity and low-fidelity outputs are exploited to make an educated guess of the high-fidelity outcome based only on quick low-fidelity estimations, hence without the need of running full expensive high-fidelity calculations. With respect to neural networks, this approach is expected to require 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.

arXiv commentProceedings of the COSIRES 2018 conference, submitted for peer-review in Nuclear Instruments and Methods in Physics Research B: Beam Interactions with Materials and Atoms

References in corpus (8)

• On representing chemical environments
• A Spectral Analysis Method for Automated Generation of Quantum-Accurate Interatomic Potentials
• Extending the Accuracy of the SNAP Interatomic Potential Form
• KineCluE: a Kinetic Cluster Expansion code to compute transport coefficients beyond the dilute limit
• Mechanism of Re precipitation in irradiated W-Re alloys from kinetic Monte Carlo simulations
• Automatic numerical evaluation of vacancy-mediated transport for arbitrary crystals: Onsager coefficients in the dilute limit using a Green function approach
• A novel set of rotationally and translationally invariant features for images based on the non-commutative bispectrum
• Gaussian Approximation Potential: an interatomic potential derived from first principles Quantum Mechanics

Cited by in corpus (4)

• A Deep Dive into Machine Learning Density Functional Theory for Materials Science and Chemistry
• Extensive deep neural networks for transferring small scale learning to large scale systems
• Predicting activation energies for vacancy-mediated diffusion in alloys using a transition-state cluster expansion
• Inverting the Kohn-Sham equations with physics-informed machine learning