skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Discovering charge density functionals and structure-property relationships with PROPhet: A general framework for coupling machine learning and first-principles methods

Authors:
; ;
Publication Date:
Research Org.:
Energy Frontier Research Centers (EFRC) (United States). Center for Next Generation of Materials by Design: Incorporating Metastability (CNGMD)
Sponsoring Org.:
USDOE Office of Science (SC), Basic Energy Sciences (BES) (SC-22)
OSTI Identifier:
1388993
DOE Contract Number:
AC36-99GO10337
Resource Type:
Journal Article
Resource Relation:
Journal Name: Scientific Reports; Journal Volume: 7; Journal Issue: 1; Related Information: CNGMD partners with National Renewable Energy Laboratory (lead); Colorado School of Mines; Harvard University; Lawrence Berkeley National Laboratory; Massachusetts Institute of Technology; Oregon State University; SLAC National Accelerator Laboratory
Country of Publication:
United States
Language:
English
Subject:
solar (photovoltaic), solar (fuels), solid state lighting, phonons, thermoelectric, hydrogen and fuel cells, defects, charge transport, optics, materials and chemistry by design, synthesis (novel materials)

Citation Formats

Kolb, Brian, Lentz, Levi C., and Kolpak, Alexie M. Discovering charge density functionals and structure-property relationships with PROPhet: A general framework for coupling machine learning and first-principles methods. United States: N. p., 2017. Web. doi:10.1038/s41598-017-01251-z.
Kolb, Brian, Lentz, Levi C., & Kolpak, Alexie M. Discovering charge density functionals and structure-property relationships with PROPhet: A general framework for coupling machine learning and first-principles methods. United States. doi:10.1038/s41598-017-01251-z.
Kolb, Brian, Lentz, Levi C., and Kolpak, Alexie M. Wed . "Discovering charge density functionals and structure-property relationships with PROPhet: A general framework for coupling machine learning and first-principles methods". United States. doi:10.1038/s41598-017-01251-z.
@article{osti_1388993,
title = {Discovering charge density functionals and structure-property relationships with PROPhet: A general framework for coupling machine learning and first-principles methods},
author = {Kolb, Brian and Lentz, Levi C. and Kolpak, Alexie M.},
abstractNote = {},
doi = {10.1038/s41598-017-01251-z},
journal = {Scientific Reports},
number = 1,
volume = 7,
place = {United States},
year = {Wed Apr 26 00:00:00 EDT 2017},
month = {Wed Apr 26 00:00:00 EDT 2017}
}
  • Modern ab initio methods have rapidly increased our understanding of solid state materials properties, chemical reactions, and the quantum interactions between atoms. However, poor scaling often renders direct ab initio calculations intractable for large or complex systems. There are two obvious avenues through which to remedy this problem: (i) develop new, less expensive methods to calculate system properties, or (ii) make existing methods faster. This paper describes an open source framework designed to pursue both of these avenues. PROPhet (short for PROPerty Prophet) utilizes machine learning techniques to find complex, non-linear mappings between sets of material or system properties. Themore » result is a single code capable of learning analytical potentials, non-linear density functionals, and other structure-property or property-property relationships. These capabilities enable highly accurate mesoscopic simulations, facilitate computation of expensive properties, and enable the development of predictive models for systematic materials design and optimization. Here, this work explores the coupling of machine learning to ab initio methods through means both familiar (e.g., the creation of various potentials and energy functionals) and less familiar (e.g., the creation of density functionals for arbitrary properties), serving both to demonstrate PROPhet’s ability to create exciting post-processing analysis tools and to open the door to improving ab initio methods themselves with these powerful machine learning techniques.« less
  • A very active area of materials research is to devise methods that use machine learning to automatically extract predictive models from existing materials data. While prior examples have demonstrated successful models for some applications, many more applications exist where machine learning can make a strong impact. To enable faster development of machine-learning-based models for such applications, we have created a framework capable of being applied to a broad range of materials data. Our method works by using a chemically diverse list of attributes, which we demonstrate are suitable for describing a wide variety of properties, and a novel method formore » partitioning the data set into groups of similar materials to boost the predictive accuracy. In this manuscript, we demonstrate how this new method can be used to predict diverse properties of crystalline and amorphous materials, such as band gap energy and glass-forming ability.« less
  • Here, we apply unsupervised machine learning techniques, mainly principal component analysis (PCA), to compare and contrast the phase behavior and phase transitions in several classical spin models - the square and triangular-lattice Ising models, the Blume-Capel model, a highly degenerate biquadratic-exchange spin-one Ising (BSI) model, and the 2D XY model, and examine critically what machine learning is teaching us. We find that quantified principal components from PCA not only allow exploration of different phases and symmetry-breaking, but can distinguish phase transition types and locate critical points. We show that the corresponding weight vectors have a clear physical interpretation, which ismore » particularly interesting in the frustrated models such as the triangular antiferromagnet, where they can point to incipient orders. Unlike the other well-studied models, the properties of the BSI model are less well known. Using both PCA and conventional Monte Carlo analysis, we demonstrate that the BSI model shows an absence of phase transition and macroscopic ground-state degeneracy. The failure to capture the 'charge' correlations (vorticity) in the BSI model (XY model) from raw spin configurations points to some of the limitations of PCA. Finally, we employ a nonlinear unsupervised machine learning procedure, the 'antoencoder method', and demonstrate that it too can be trained to capture phase transitions and critical points.« less
    Cited by 10
  • As a representative lanthanide endohedral metallofullerene, Gd@C{sub 82} has attracted a widespread attention among theorists and experimentalists ever since its first synthesis. Through comprehensive comparisons and discussions, as well as references to the latest high precision experiments, we evaluated the performance of different computational methods. Our results showed that the appropriate choice of the exchange-correlation functionals is the decisive factor to accurately predict both geometric and electronic structures for Gd@C{sub 82}. The electronic structure of the ground state and energy gap between the septet ground state and the nonet low-lying state obtained from pure density functional methods, such as PBEmore » and PW91, are in good agreement with current experiment. Unlike pure functionals, the popularly used hybrid functionals in previous studies, such as B3LYP, could infer the qualitative correct ground state only when small basis set for C atoms is employed. Furthermore, we also highlighted that other geometric structures of Gd@C{sub 82} with the Gd staying at different positions are either not stable or with higher energies. This work should provide some useful references for various theoretical methodologies in further density functional studies on Gd@C{sub 82} and its derivatives in the future.« less