Exchange-correlation Energy Density Research Articles

A numerical Poisson solver with improved radial solutions for a self-consistent locally scaled self-interaction correction method

The universal applicability of density functional approximations is limited by self-interaction error made by these functionals. Recently, a novel one-electron self-interaction-correction (SIC) method that uses an iso-orbital indicator to apply the SIC at each point in space by scaling the exchange-correlation and Coulomb energy densities was proposed. The locally scaled SIC (LSIC) method is exact for the one-electron densities, and unlike the well-known Perdew–Zunger SIC (PZSIC) method recovers the uniform electron gas limit of the uncorrected density functional approximation, and reduces to PZSIC method as a special case when isoorbital indicator is set to the unity. Here, we present a numerical scheme that we have adopted to evaluate the Coulomb potential of the electron density scaled by the iso-orbital indicator required for the self-consistent LSIC calculations. After analyzing the behavior of the finite difference method (FDM) and the green function solution to the radial part of the Poisson equation, we adopt a hybrid approach that uses the FDM for the Coulomb potential due to the monopole and the GF for all higher-order terms. The performance of the resultant hybrid method is assessed using a variety of systems. The results show improved accuracy than earlier numerical schemes. We also find that, even with a generic set of radial grid parameters, accurate energy differences can be obtained using a numerical Coulomb solver in standard density functional studies.

Application of two-component neural network for exchange-correlation functional interpolation

Density functional theory (DFT) is one of the primary approaches to solving the many-body Schrodinger equation. The essential part of the DFT theory is the exchange-correlation (XC) functional, which can not be obtained in analytical form. Accordingly, the accuracy improvement of the DFT is mainly based on the development of XC functional approximations. Commonly, they are built upon analytic solutions in low- and high-density limits and result from quantum Monte Carlo or post-Hartree-Fock numerical calculations. However, there is no universal functional form to incorporate these data into XC functional. Instead, various parameterizations use heuristic rules to build a specific XC functional. The neural network (NN) approach to interpolate the data from higher precision theories can give a unified path to parametrize an XC functional. Moreover, data from many existing quantum chemical databases could provide the XC functional with improved accuracy. We develop NN XC functional, which gives exchange potential and energy density without direct derivatives of exchange-correlation energy density. Proposed NN architecture consists of two parts NN-E and NN-V, which could be trained in separate ways, adding new flexibility to XC functional. We also show that the developed NN XC functional converges in the self-consistent cycle and gives reasonable energies when applied to atoms, molecules, and crystals.

Improving Density Functional Prediction of Molecular Thermochemical Properties with a Machine-Learning-Corrected Generalized Gradient Approximation.

The past decade has seen an increasing interest in designing sophisticated density functional approximations (DFAs) by integrating the power of machine learning (ML) techniques. However, application of the ML-based DFAs is often confined to simple model systems. In this work, we construct an ML correction to the widely used Perdew-Burke-Ernzerhof (PBE) functional by establishing a semilocal mapping from the electron density and reduced gradient to the exchange-correlation energy density. The resulting ML-corrected PBE is immediately applicable to any real molecule and yields significantly improved heats of formation while preserving the accuracy for other thermochemical and kinetic properties. This work highlights the prospect of combining the power of data-driven ML methods with physics-inspired derivations for reaching the heaven of chemical accuracy.

Exchange-Correlation Energy Densities and Response Potentials: Connection between Two Definitions and Analytical Model for the Strong-Coupling Limit of a Stretched Bond.

We analyze in depth two widely used definitions (from the theory of conditional probability amplitudes and from the adiabatic connection formalism) of the exchange-correlation energy density and of the response potential of Kohn–Sham density functional theory. We introduce a local form of the coupling-constant-dependent Hohenberg–Kohn functional, showing that the difference between the two definitions is due to a corresponding local first-order term in the coupling constant, which disappears globally (when integrated over all space), but not locally. We also design an analytic representation for the response potential in the strong-coupling limit of density functional theory for a model single stretched bond.

Design and analysis of machine learning exchange-correlation functionals via rotationally invariant convolutional descriptors

In this work we explore the potential of a new data-driven approach to the design of exchange-correlation (XC) functionals. The approach, inspired by convolutional filters in computer vision and surrogate functions from optimization, utilizes convolutions of the electron density to form a feature space to represent local electronic environments and neural networks to map the features to the exchange-correlation energy density. These features are orbital free, and provide a systematic route to including information at various length scales. This work shows that convolutional descriptors are theoretically capable of an exact representation of the electron density, and proposes Maxwell-Cartesian spherical harmonic kernels as a class of rotationally invariant descriptors for the construction of machine-learned functionals. The approach is demonstrated using data from the B3LYP functional on a number of small-molecules containing C, H, O, and N along with a neural network regression model. The machine-learned functionals are compared to standard physical approximations and the accuracy is assessed for the absolute energy of each molecular system as well as formation energies. The results indicate that it is possible to reproduce B3LYP formation energies to within chemical accuracy using orbital-free descriptors with a spatial extent of 0.2 A. The findings provide empirical insight into the spatial range of electron exchange, and suggest that the combination of convolutional descriptors and machine-learning regression models is a promising new framework for XC functional design, although challenges remain in obtaining training data and generating models consistent with pseudopotentials.

Embedded Cluster Density Approximation for Exchange-Correlation Energy: A Natural Extension of the Local Density Approximation.

We developed a local correlation method in the framework of Kohn-Sham density functional theory (KS-DFT). The method is termed ''embedded cluster density approximation'' (ECDA) and is a logical extension of the local density approximation. In ECDA, an embedded cluster is defined for each atom based on the finite-temperature density functional embedding theory. The clusters' exchange-correlation (XC) energy densities are calculated using high-level XC functionals. The system's XC energy is then constructed by patching these locally computed high-level XC energy densities over the system in an atom-by-atom manner. We derive the relationship between the embedding potential and the system's KS potential. We show how to efficiently compute the system's XC potential, which is the functional derivative of the patched XC energy with respect to the system's electron density. The accuracy of ECDA is examined by patching the exact exchange (EXX) and the random phase approximation (RPA) correlation energy densities in a one-dimensional hydrogen chain, as well as by patching EXX energy densities in several molecules. The agreement between ECDA and KS-DFT on total energies, dipole moments, XC potentials, and KS eigenvalues is good in general as the clusters are made larger. Based on these encouraging results, we expect ECDA to be a simple yet effective method to scale up high-level KS-DFT simulations in large systems.

Communication: Two types of flat-planes conditions in density functional theory.

Using results from atomic spectroscopy, we show that there are two types of flat-planes conditions. The first type of flat-planes condition occurs when the energy as a function of the number of electrons of each spin, Nα and Nβ, has a derivative discontinuity on a line segment where the number of electrons, Nα + Nβ, is an integer. The second type of flat-planes condition occurs when the energy has a derivative discontinuity on a line segment where the spin polarization, Nα - Nβ, is an integer, but does not have a discontinuity associated with an integer number of electrons. Type 2 flat planes are rare-we observed just 15 type 2 flat-planes conditions out of the 4884 cases we tested-but their mere existence has implications for the design of exchange-correlation energy density functionals. To facilitate the development of functionals that have the correct behavior with respect to both fractional number of electrons and fractional spin polarization, we present a dataset for the chromium atom and its ions that can be used to test new functionals.

Exchange-Correlation Functionals via Local Interpolation along the Adiabatic Connection.

The construction of density-functional approximations is explored by modeling the adiabatic connection locally, using energy densities defined in terms of the electrostatic potential of the exchange–correlation hole. These local models are more amenable to the construction of size-consistent approximations than their global counterparts. In this work we use accurate input local ingredients to assess the accuracy of a range of local interpolation models against accurate exchange–correlation energy densities. The importance of the strictly correlated electrons (SCE) functional describing the strong coupling limit is emphasized, enabling the corresponding interpolated functionals to treat strong correlation effects. In addition to exploring the performance of such models numerically for the helium and beryllium isoelectronic series and the dissociation of the hydrogen molecule, an approximate analytic model is presented for the initial slope of the local adiabatic connection. Comparisons are made with approaches based on global models, and prospects for future approximations based on the local adiabatic connection are discussed.

Insights into the phase diagram of bismuth ferrite from quasiharmonic free-energy calculations

We have used first-principles methods to investigate the phase diagram of multiferroic bismuth ferrite (BiFeO${}_{3}$ or BFO), revealing the energetic and vibrational features that control the occurrence of various relevant structures. More precisely, we have studied the relative stability of four low-energy BFO polymorphs by computing their free energies within the quasiharmonic approximation, introducing a practical scheme that allows us to account for the main effects of spin disorder. As expected, we find that the ferroelectric ground state of the material (with $R3c$ space group) transforms into an orthorhombic paraelectric phase ($Pnma$) upon heating. We show that this transition is not significantly affected by magnetic disorder, and that the occurrence of the $Pnma$ structure relies on its being vibrationally (although not elastically) softer than the $R3c$ phase. We also investigate a representative member of the family of nanotwinned polymorphs recently predicted for BFO [S. Prosandeev et al., Adv. Funct. Mater. 23, 234 (2013)] and discuss their possible stabilization at the boundaries separating the $R3c$ and $Pnma$ regions in the corresponding pressure-temperature phase diagram. Finally, we elucidate the intriguing case of the so-called supertetragonal phases of BFO: Our results explain why such structures have never been observed in the bulk material, despite their being stable polymorphs of very low energy. Quantitative comparison with experiment is provided whenever possible, and the relative importance of various physical effects (zero-point motion, spin fluctuations, thermal expansion) and technical features (employed exchange-correlation energy density functional) is discussed. Our work attests the validity and usefulness of the quasiharmonic scheme to investigate the phase diagram of this complex oxide, and prospective applications are discussed.

Construction of a general semilocal exchange-correlation hole model: Application to nonempirical meta-GGA functionals

Using a reverse-engineering method, we construct a meta-generalized gradient approximation (meta-GGA) angle-averaged exchange-correlation (XC) hole model which has a general applicability. It satisfies known exact hole constraints and can exactly recover the exchange-correlation energy density of any reasonable meta-GGA exchange-correlation energy functional satisfying a minimal set of exact properties. The hole model is applied to several nonempirical meta-GGA functionals: the Tao-Perdew-Staroverov-Scuseria (TPSS), the revised TPSS (revTPSS), and the recently Balanced LOCalization (BLOC) meta-GGA [L. A. Constantin, E. Fabiano, and F. Della Sala, J. Chem. Theory Comput. 9, 2256 (2013)]. The empirical M06-L meta-GGA functional is also considered. Real-space analyses of atoms and ions as well as wave-vector analyses of jellium surface energies show that the meta-GGA hole models, in particular the BLOC one, are very realistic and can reproduce many features of benchmark XC holes. In addition, the BLOC hole model can be used to estimate with good accuracy the Coulomb hole radius of small atoms and ions. Thus, the proposed meta-GGA hole models provide a valuable tool to validate in detail existing meta-GGA functionals, and can be further used in the development of density functional theory methods beyond the semilocal level of theory.

Current-density functional theory study of the H2 molecule evolving under a strong ultrashort magnetic field

Hydrogen molecule in a strong ultrashort magnetic field is investigated through a current-density functional theory (CDFT) and quantum fluid dynamics (QFD) based approach employing current-density dependent vector exchange-correlation potential and energy density functional derived with a vorticity variable. The numerical computations through the CDFT based approach are performed for the H2 molecule, starting initially from its field-free ground state, in a parallel internuclear axis and magnetic field-axis configuration with the internuclear separation R ranging from 0.1 a.u. to 14.0 a.u., and the strength of the time-dependent (TD) magnetic field varying between 0−1011 G over a few femtoseconds. The numerical results are compared with that obtained using an approach based on the current-density independent approximation under similar computational constraints but employing only scalar exchange-correlation potential dependent on the electronic charge-density alone. The current-density based approach yields exchange- and correlation energy as well as electronic charge-density of the H2 molecule drastically different from that obtained using current-independent approach, in particular, at TD magnetic field-strengths >109 G during a typical time-period of the field when the magnetic-field had attained maximum applied field-strength and is switched to a decreasing ramp function. This nonadiabatic behavior of the TD electronic charge-density is traced to the TD vorticity-dependent vector exchange-correlation potential of the CDFT based approach. The interesting electron dynamics of the H2 molecule in strong TD magnetic field is further elucidated by treating electronic charge-density as an ‘electron-fluid’. The present work also reveals interesting real-time dynamics on the attosecond time-scale in the electronic charge-density distribution of the hydrogen molecule.

He 2++ molecular ion in a strong time-dependent magnetic field: A current-density functional study

The He 2++ molecular ion exposed to a strong ultrashort time-dependent (TD) magnetic field of the order of 10(9) G is investigated through a quantum fluid dynamics (QFD) and current-density functional theory (CDFT) based approach using vector exchange-correlation (XC) potential and energy density functional that depend not only on the electronic charge-density but also on the current density. The TD-QFD-CDFT computations are performed in a parallel internuclear-axis and magnetic field-axis configuration at the field-free equilibrium internuclear separation R = 1.3 au with the field-strength varying between 0 and 10(11) G. The TD behavior of the exchange- and correlation energy of the He 2++ is analyzed and compared with that obtained using a [B-TD-QFD-density functional theory (DFT)] approach based on the conventional TD-DFT under similar computational constraints but using only scalar XC potential and energy density functional dependent on the electronic charge-density alone. The CDFT based approach yields TD exchange- and correlation energy and TD electronic charge-density significantly different from that obtained using the conventional TD-DFT based approach, particularly, at typical magnetic field strengths and during a typical time period of the TD field. This peculiar behavior of the CDFT-based approach is traced to the TD current-density dependent vector XC potential, which can induce nonadiabatic effects causing retardation of the oscillating electronic charge density. Such dissipative electron dynamics of the He 2++ molecular ion is elucidated by treating electronic charge density as an electron-"fluid" in the terminology of QFD.

Calculation of the Binding Energies of Different Types of Hydrogen Bonds Using GGA Density Functional and Its Long-Range, Empirical Dispersion Correction Methods

We investigated eleven exchange-correlation energy density functionals including generalized gradient approximation (GGA) (PBE, PW91), meta-GGA (M06-L), hyper-GGA (M06-2X, B3LYP, X3LYP), LC-DFT methods (CAM-B3LYP, LC-ωPBE, ωB97X), and density functional theory with dispersion corrections (DFT-D) methods (ωB97X-D, B97-D) for their performance in describing systems with conventional and non-conventional hydrogen bonds. After comparing the results using the benchmark CCSD(T)/aug-cc-pVQZ approach we found that the M06-2X and ωB97X-D functionals provided the most accurate and reliable results for the fifteen systems studied in this work with strong, moderate, and weak hydrogen bonds. It is important to employ an appropriate basis set to predict the binding energy of hydrogen bonds for all DFT methods and we found that the basis set of 6-311++G(2d, 2p) or aug-cc-pVDZ is adequate. The effect of the basis set superposition error (BSSE) is relatively small for the DFT methods tested. All the methods except for ωB97X and ωB97X-D were found to produce equally accurate or even more accurate results without BSSE correction.

Rung 3.5 density functionals

This work proposes a new term on the "Jacob's ladder" of approximate exchange-correlation functionals in Kohn-Sham density functional theory. In these Rung 3.5 functionals, the exchange-correlation energy density at a point depends linearly (rather than quadratically) on the nonlocal Kohn-Sham one-particle density matrix in a finite neighborhood around the point. These functionals are intermediate in complexity between the semilocal approximations of Rungs 1-3, and fully nonlocal Rung 4 approximations such as global hybrids. Rung 3.5 functionals built on the model for exchange in [B. G. Janesko, J. Chem. Phys. 131, 234111 (2009)] predict molecular thermochemistry and kinetics with accuracy intermediate between their "parent" semilocal functional and the corresponding global hybrid. The best Rung 3.5 functional presented here gives mean absolute errors of 5.7 kcal/mol for G3/99 thermochemistry, 5.2 kcal/mol for HTBH38/04 hydrogen-transfer reaction barriers, and 5.7 kcal/mol for NHTBH38/04 nonhydrogen-transfer reaction barriers, while incorporating only two empirical parameters.

Communications: Making density functional theory and the quantum theory of atoms in molecules converse: A local approach

A first (local) bridge between Kohn-Sham density functional theory and the quantum theory of atoms in molecules of Bader is built by means of a second order reduced density gradient expansion of the exchange-correlation energy density at a given bond critical point. This approach leads to the definition of new "mixed" descriptors that are particularly useful for the classification of the chemical interactions for which the traditional atoms in molecules characterization reveals insufficient, as for instance the distinction between hydrogen and agostic bonds.

Density functional theory for strongly-interacting electrons: perspectives for physics and chemistry

Improving the accuracy and thus broadening the applicability of electronic density functional theory (DFT) is crucial to many research areas, from material science, to theoretical chemistry, biophysics and biochemistry. In the last three years, the mathematical structure of the strong-interaction limit of density functional theory has been uncovered, and exact information on this limit has started to become available. The aim of this paper is to give a perspective on how this new piece of exact information can be used to treat situations that are problematic for standard Kohn-Sham DFT. One way to use the strong-interaction limit, more relevant for solid-state physical devices, is to define a new framework to do practical, non-conventional, DFT calculations in which a strong-interacting reference system is used instead of the traditional non-interacting one of Kohn and Sham. Another way to proceed, more related to chemical applications, is to include the exact treatment of the strong-interaction limit into approximate exchange-correlation energy density functionals in order to describe difficult situations such as the breaking of the chemical bond.

Exchange-correlation energy functional based on the Airy-gas reference system

In recent work, generalized gradient approximations (GGAs) have been constructed from the energy density of the Airy gas for exchange but not for correlation. We report the random-phase approximation (RPA) conventional correlation energy density of the Airy gas, the simplest edge electron gas, in which the auxiliary noninteracting electrons experience a linear potential. By fitting the Airy-gas RPA exchange-correlation energy density and making an accurate short-range correction to RPA, we propose a simple beyond RPA GGA density functional (``$\text{ARPA}+$'') for the exchange-correlation energy. Our functional, tested for jellium surfaces, atoms, molecules, and solids, improves mildly over the local spin-density approximation for atomization energies and lattice constants without much worsening the already good surface exchange-correlation energies.

Reconstruction of Density Functionals from Kohn-Sham Potentials by Integration along Density Scaling Paths.

We demonstrate by specific examples that if a Kohn-Sham exchange-correlation potential is given explicitly in terms of the electron density and its derivatives, then one can easily reconstruct the parent density functional by evaluating analytically (or numerically with one-dimensional quadratures) the van Leeuwen-Baerends line integral (Phys. Rev. A 1995, 51, 170-178) along a path of (coordinate)-scaled densities. The choice of a density scaling path amounts to defining the gauge of the resultant exchange-correlation energy density. The well-known Levy-Perdew virial relation for exchange potentials can be viewed as an analytical line integral along the electron-number-conserving uniform density scaling path. Energies obtained from model exchange-correlation potentials should be interpreted with caution because the reconstructed density functional is unique (up to a gauge transformation) only if the model Kohn-Sham potential is a functional derivative.

Numerical integration of exchange-correlation energies and potentials using transformed sparse grids

A new numerical integration procedure for exchange-correlation energies and potentials is proposed and "proof of principle" results are presented. The numerical integration grids are built from sparse-tensor product grids (constructed according to Smolyak's prescription [Dokl. Akad. Nauk. 4, 240 (1963)] ) on the unit cube. The grid on the unit cube is then transformed to a grid over real space with respect to a weight function, which we choose to be the promolecular density. This produces a "whole molecule" grid, in contrast to conventional integration methods in density-functional theory, which use atom-in-molecule grids. The integration scheme was implemented in a modified version of the DEMON2K density-functional theory program, where it is used to evaluate integrals of the exchange-correlation energy density and the exchange-correlation potential. Ground-state energies and molecular geometries are accurately computed. The biggest advantages of the grid are its flexibility (it is easy to change the number and distribution of grid points) and its whole molecule nature. The latter feature is potentially helpful for basis-set-free computational algorithms.

Interaction energies in non-covalently bound intermolecular complexes derived using the subsystem formulation of density functional theory

Interaction energies for a representative sample of 39 intermolecular complexes are calculated using two computational approaches based on the subsystem formulation of density functional theory introduced by Cortona (Phys. Rev. B 44:8454, 1991), adopted for studies of intermolecular complexes (Wesolowski and Weber in Chem. Phys. Lett. 248:71, 1996). The energy components (exchange-correlation and non-additive kinetic) expressed as explicit density functionals are approximated by means of gradient-free- (local density approximation) of gradient-dependent- (generalized gradient approximation) approximations. The sample of the considered intermolecular complexes was used previously by Zhao and Truhlar to compare the interaction energies derived using various methods based on the Kohn-Sham equations with high-level quantum chemistry results considered as the reference. It stretches from rare gas dimers up to strong hydrogen bonds. Our results indicate that the subsystem-based methods provide an interesting alternative to that based on the Kohn-Sham equations. Local density approximation, which is the simplest approximation for the relevant density functionals and which does not rely on any empirical data, leads to a computational approach comparing favorably with more than twenty methods based on the Kohn-Sham equations including the ones, which use extensively empirical parameterizations. For various types of non-bonding interactions, the strengths and weaknesses of gradient-free and gradient-dependent approximations to exchange-correlation and non-additive kinetic energy density functionals are discussed in detail.