scholarly journals Machine learning the Hubbard U parameter in DFT+U using Bayesian optimization

2020 ◽  
Vol 6 (1) ◽  
Author(s):  
Maituo Yu ◽  
Shuyang Yang ◽  
Chunzhi Wu ◽  
Noa Marom
Keyword(s):  
Machine Learning ◽  
Transition Metal Oxides ◽  
Density Functional ◽  
Functional Results ◽  
Bayesian Optimization ◽  
Hybrid Functional ◽  
Band Structures ◽  
Local Exchange ◽  
Functional Theory ◽  
Exchange Correlation

AbstractWithin density functional theory (DFT), adding a Hubbard U correction can mitigate some of the deficiencies of local and semi-local exchange-correlation functionals, while maintaining computational efficiency. However, the accuracy of DFT+U largely depends on the chosen Hubbard U values. We propose an approach to determining the optimal U parameters for a given material by machine learning. The Bayesian optimization (BO) algorithm is used with an objective function formulated to reproduce the band structures produced by more accurate hybrid functionals. This approach is demonstrated for transition metal oxides, europium chalcogenides, and narrow-gap semiconductors. The band structures obtained using the BO U values are in agreement with hybrid functional results. Additionally, comparison to the linear response (LR) approach to determining U demonstrates that the BO method is superior.

scholarly journals Machine learning the derivative discontinuity of density-functional theory

2021 ◽  
Author(s):  
Johannes Gedeon ◽  
Jonathan Schmidt ◽  
Matthew Hodgson ◽  
Jack Wetherel ◽  
Carlos Benavides-Riveros ◽  
...  
Keyword(s):  
Machine Learning ◽  
Density Functional Theory ◽  
Density Functional ◽  
Correlation Energy ◽  
Integer Number ◽  
Local Exchange ◽  
Functional Theory ◽  
Exchange Correlation ◽  
Derivative Discontinuity ◽  
Number Of Particles

Abstract Machine learning is a powerful tool to design accurate, highly non-local, exchange-correlation functionals for density functional theory. So far, most of those machine learned functionals are trained for systems with an integer number of particles. As such, they are unable to reproduce some crucial and fundamental aspects, such as the explicit dependency of the functionals on the particle number or the infamous derivative discontinuity at integer particle numbers. Here we propose a solution to these problems by training a neural network as the universal functional of density-functional theory that (i) depends explicitly on the number of particles with a piece-wise linearity between the integer numbers and (ii) reproduces the derivative discontinuity of the exchange-correlation energy. This is achieved by using an ensemble formalism, a training set containing fractional densities, and an explicitly discontinuous formulation.

scholarly journals Exact Generalized Kohn-Sham Theory for Hybrid Functionals

2020 ◽  
Author(s):  
Rachel Garrick ◽  
Amir Natan ◽  
Tim Gould ◽  
Leeor Kronik
Keyword(s):  
Density Functional Theory ◽  
Density Functional ◽  
Formal Theory ◽  
Density Functional Theory Calculations ◽  
State Density ◽  
Hybrid Functional ◽  
Functional Theory ◽  
Exchange Correlation ◽  
Hybrid Functionals ◽  
The Many

Hybrid functionals have proven to be of immense practical value in density functional theory calculations. While they are often thought to be a heuristic construct, it has been established that this is in fact not the case. Here, we present a rigorous and formally exact analysis of generalized Kohn-Sham (GKS) density functional theory of hybrid functionals, in which exact remainder exchange-correlation potentials combine with a fraction of Fock exchange to produce the correct ground state density. First, we extend formal GKS theory by proving a generalized adiabatic connection theorem. We then use this extension to derive two different definitions for a rigorous distinction between multiplicative exchange and correlation components - one new and one previously postulated. We examine their density-scaling behavior and discuss their similarities and differences. We then present a new algorithm for obtaining exact GKS potentials by inversion of accurate reference electron densities and employ this algorithm to obtain exact potentials for simple atoms and ions. We establish that an equivalent description of the many-electron problem is indeed obtained with any arbitrary global fraction of Fock exchange and we rationalize the Fock-fraction dependence of the computed remainder exchange-correlation potentials in terms of the new formal theory. Finally, we use the exact theoretical framework and numerical results to shed light on the exchange-correlation potential used in approximate hybrid functional calculations and to assess the consequences of different choices of fractional exchange.<br><br>

scholarly journals Exact Generalized Kohn-Sham Theory for Hybrid Functionals

2020 ◽  
Author(s):  
Rachel Garrick ◽  
Amir Natan ◽  
Tim Gould ◽  
Leeor Kronik
Keyword(s):  
Density Functional Theory ◽  
Density Functional ◽  
Formal Theory ◽  
Density Functional Theory Calculations ◽  
State Density ◽  
Hybrid Functional ◽  
Functional Theory ◽  
Exchange Correlation ◽  
Hybrid Functionals ◽  
The Many

Hybrid functionals have proven to be of immense practical value in density functional theory calculations. While they are often thought to be a heuristic construct, it has been established that this is in fact not the case. Here, we present a rigorous and formally exact analysis of generalized Kohn-Sham (GKS) density functional theory of hybrid functionals, in which exact remainder exchange-correlation potentials combine with a fraction of Fock exchange to produce the correct ground state density. First, we extend formal GKS theory by proving a generalized adiabatic connection theorem. We then use this extension to derive two different definitions for a rigorous distinction between multiplicative exchange and correlation components - one new and one previously postulated. We examine their density-scaling behavior and discuss their similarities and differences. We then present a new algorithm for obtaining exact GKS potentials by inversion of accurate reference electron densities and employ this algorithm to obtain exact potentials for simple atoms and ions. We establish that an equivalent description of the many-electron problem is indeed obtained with any arbitrary global fraction of Fock exchange and we rationalize the Fock-fraction dependence of the computed remainder exchange-correlation potentials in terms of the new formal theory. Finally, we use the exact theoretical framework and numerical results to shed light on the exchange-correlation potential used in approximate hybrid functional calculations and to assess the consequences of different choices of fractional exchange.<br><br>

scholarly journals Weight dependence of local exchange–correlation functionals in ensemble density-functional theory: double excitations in two-electron systems

2020 ◽  
Vol 224 ◽  
pp. 402-423 ◽  
Author(s):  
Clotilde Marut ◽  
Bruno Senjean ◽  
Emmanuel Fromager ◽  
Pierre-François Loos
Keyword(s):  
Density Functional Theory ◽  
Density Functional ◽  
Electron Systems ◽  
Local Exchange ◽  
Functional Theory ◽  
Exchange Correlation

We discuss the construction of first-rung weight-dependent exchange–correlation density-functional approximations for He and H2 specifically designed for the computation of double excitations within Gross–Oliveira–Kohn-DFT.

Sign in / Sign up

Export Citation Format

Share Document