Abstract
We provide an introduction to Gaussian process regression (GPR) machine-learning methods in computational materials science and chemistry. The focus of the present review is on the regression of atomistic properties: in particular, on the construction of interatomic potentials, or force fields, in the Gaussian Approximation Potential (GAP) framework; beyond this, we also discuss the fitting of arbitrary scalar, vectorial, and tensorial quantities. Methodological aspects of reference data generation, representation, and regression, as well as the question of how a data-driven model may be validated, are reviewed and critically discussed. A survey of applications to a variety of research questions in chemistry and materials science illustrates the rapid growth in the field. A vision is outlined for the development of the methodology in the years to come.
Highlights
At the heart of chemistry is the need to understand the nature, transformations, and macroscopic effects of atomistic structure
With the oftenquoted role of chemistry as the “central science”,1,2 its emphasis on atomistic understanding has a bearing on many neighboring disciplines: candidate drug molecules are made by synthetic chemists based on an atomic-level knowledge of reaction mechanisms; functional materials for technological applications are characterized on a range of length scales, which begins with increasingly accurate information about where exactly the atoms are located relative to one another in three-dimensional space
We review the application of Gaussian process regression (GPR) to computational chemistry, with an emphasis on the development of the methodology over the past decade
Summary
At the heart of chemistry is the need to understand the nature, transformations, and macroscopic effects of atomistic structure. Research progress in structural chemistry has largely been driven by advances in experimental characterization techniques, from landmark studies in X-ray and neutron crystallography to novel electron microscopy techniques which make it possible to visualize individual atoms directly Complementing these new developments, detailed and realistic structural insight is increasingly gained from computer simulations. Computations based on the quantum mechanics of electronic structure, currently most commonly within the framework of density-functional theory (DFT), are widely used to study structures of molecules and materials and to predict a range of atomic-scale properties.[9−11] Two approaches are of note here. It is hoped that the present work the entire thematic issue in which it appears will provide guidance and inspiration for research in this quickly evolving field and that it will help advance the transition of the methodology from relatively specialized to much more widely used
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
