Bulletin of the American Physical Society
APS March Meeting 2020
Volume 65, Number 1
Monday–Friday, March 2–6, 2020; Denver, Colorado
Session A58: DFT and Beyond IFocus
|
Hide Abstracts |
Sponsoring Units: DCP DCOMP DPOLY DCMP Chair: Jianwei Sun, Tulane Univ Room: Mile High Ballroom 3B |
Monday, March 2, 2020 8:00AM - 8:36AM |
A58.00001: The Predictive Power of Exact Constraints and Appropriate Norms in Density Functional Theory Invited Speaker: John P. Perdew Approximations to the density functional for the exchange-correlation energy of Kohn-Sham theory can be constructed by fitting to data sets for bonded systems, or by satisfying exact mathematical constraints on the functional and appropriate norms (such as the uniform electron gas), or both. Fitting is interpolative, and can predict accurately for systems similar to those in the data sets, while satisfying exact constraints makes a functional more widely predictive over the immense space of possible molecules and materials. For example, the SCAN (strongly constrained and appropriately normed) functional [1] satisfies all 17 exact constraints that a computationally-efficient meta-generalized gradient approximation can, and has been remarkably successful for some complex materials (like liquid water [2]) and some strongly-correlated materials (like the cuprate high-temperature superconducting materials [3,4]). Prospects for an improved SCAN and for the correction of its residual self-interaction error will also be discussed. |
Monday, March 2, 2020 8:36AM - 8:48AM |
A58.00002: Assessment of DFAs in predicting the magnetization of transition metal solids Alberto Vela, Karla A Botello Mancilla, Angel M Albavera Mata, Samuel Trickey, Jose L Gazquez It has been recently reported that SCAN over stabilizes higher magnetic moments for the itinerant magnets Fe, Co and Ni.[1-4] Here we present and discuss the performance of a variety of generalized and meta-generalized gradient approximations, including our recently proposed locally parameterized GGAs, on the magnetic properties of a set composed of 3d, 4dand 5dmetals, a total of 29 solids. Detailed analyses of the itinerant magnets further show different behaviors between the band structures calculated with SCAN versus those with PBE [5] and SCAN-L [6,7] at the magnetization saturation. |
Monday, March 2, 2020 8:48AM - 9:00AM |
A58.00003: The Importance of Smoothness with Exact Constraints in Functional Design James Furness, Aaron Kaplan, John P. Perdew, Jianwei Sun There are two core goals in exchange-correlation functional design: accuracy and efficiency. Whilst modern semi-local functionals have greatly improved accuracy, this has often been at the cost of efficiency. Our previous work [1], and that of others [2], has shown the iso-orbital indicator α as a key component in this sensitivity, compounded by sharply varying interpolation functions. In a recent publication [2] Bartók and Yates propose a regularised SCAN (rSCAN) that resolves these numerical problems for the SCAN functional [3] at the expense of breaking some exact constraints. The uniform electron gas and slowly varying density limits are broken, along with incorrect coordinate scaling of the regularised iso-orbital indicator αr. |
Monday, March 2, 2020 9:00AM - 9:12AM |
A58.00004: Analysis of over-magnetization of elemental transition metal solids from the SCAN and three related density functionals. Daniel Mejia-Rodriguez, Sam Trickey Recent investigations have found that the strongly constrained and appropriately normed (SCAN) meta-generalized gradient approximation exchange-correlation functional significantly over-magnetizes elemental Fe, Co, and Ni solids [1]. The difficulty does not occur in SCAN-L (deorbitalized SCAN) [2]. The problem have been traced to the tendency to favor integer occupations, similar to a DFT+U treatment [3]. We discuss what properties of the SCAN, revSCAN and rSCAN functional forms are responsible for such errors, and why the SCAN-L functional does much better in this respect [2]. |
Monday, March 2, 2020 9:12AM - 9:24AM |
A58.00005: Failure of the SCAN functional for itinerant magnets and the reasons David Singh, Yuhao Fu The SCAN functional fails dramatically in its description of itinerant magnets including the elements Fe, Ni and Co. We report that SCAN is closer in behavior to functionals that yield localized behavior, such as hybrid functionals, than other semilocal functionals that are tested. The results are understood in terms of a tendency to differentiate orbitals, favoring integer occupation, which is necessary for a correct description of atomic systems, but inappropriate for the open shell metallic ferromagnetic metals. This has to do with the fitted nature of SCAN where it reproduces selected “norms”. |
Monday, March 2, 2020 9:24AM - 9:36AM |
A58.00006: Assessment of SCAN and regularized SCAN functional with and without self-interaction correction Alan Salcedo, Yoh Yamamoto, Tunna Baruah, Rajendra Zope The Strongly Constrained and Appropriately Normed (SCAN) is a non-empirical meta-GGA functional that satisfies all the known 17 exact constraints that a meta-GGA functional can. Its numerical implementation is challenging and its use requires extremely fine numerical grids. The numerical issues become even more pronounced for the Perdew-Zunger (PZ) self-interaction correction (SIC) method as it requires evaluating the functional and its derivatives with orbital density that varies rapidly compared to total density. Bartok and Yates recently proposed a modification to SCAN, called regularized SCAN (rSCAN) to remove some of the numerical instabilities of SCAN. We have implemented rSCAN functional in our Fermi-Lowdin Self-Interaction Correction code and performed assessment of rSCAN for various molecular properties using standard datasets. We find rSCAN gives similar results to SCAN despite violation of some of the constraints and requires comparatively coarser grids than SCAN. However, like SCAN, rSCAN also shows numerical instabilities in PZ-SIC calculations where orbital densities are used. This problem is due to the use of variable α in these functionals. A solution to this numerical instability is suggested and implemented. |
Monday, March 2, 2020 9:36AM - 9:48AM |
A58.00007: Systematizing Approximate Density Functional Design Aaron Kaplan, James Furness, Jianwei Sun, John P. Perdew A meta-GGA with well-rounded accuracy and efficient numerical performance for all many-electron systems has proved to be elusive. TPSS, while accurate for many systems, never reaches predictive accuracy. SCAN [1], motivated by a need to systematize development of approximate density functionals, is often predictive, but falters for metallic systems [2]. In this talk, I’ll discuss the need to further systematize the construction of approximate density functionals. The design of possible successors to SCAN, with results for transition metals and weakly-bonded systems, as well as improvements in meta-GGA numeric efficiency from new iso-orbital indicators [3] will also be discussed. |
Monday, March 2, 2020 9:48AM - 10:00AM |
A58.00008: Evaluating exchange correlation performance on structural prediction from GGA to metaGGA Pedram Tavadze, Matthieu J Verstraete, Aldo H Romero The use of density functional theory in materials prediction and structural characterization is a very well-established field. We use this methodology in a high throughput framework, where thousands of materials are analyzed and classified in databases. This step has become quite standard and many databases do not discuss their most basic approximation: which exchange correlation functional has been used. While much work has been done on the functional dependency of the cell parameters and ground state space group, little is known for the internal degrees of freedom. A crystal structure can be reduced to the cell parameters and the Wyckoff positions (WPs), some of which can add additional degrees of freedom to the crystal structure. In this work we analyze the effect on the internal degrees of freedom of more than 1500 structures, both metallic and semiconductor materials from completely local functionals as LDA up to meta-GGAs using the recently introduced SCAN. |
Monday, March 2, 2020 10:00AM - 10:12AM |
A58.00009: Single Hamiltonian for self-interaction corrected DFT with Fermi-Lowdin orbitals Tunna Baruah, Mark Pederson
|
Monday, March 2, 2020 10:12AM - 10:24AM |
A58.00010: Self-interaction correction and dielectric properties of molecules Kushantha Withanage, Alexander I Johnson, Juan Peralta, Yoh Yamamoto, Rajendra Zope, Tunna Baruah, Koblar Alan Jackson We recently used the Fermi-Löwdin orbital implementation of the Perdew-Zunger self-interaction correction (FLO-SIC) to study the impact of self-interaction error on the prediction of dipole moments of molecules and polarizabilities for atoms (DOI: 10.1103/PhysRevA.100.012505). Using FLO-SIC in conjunction with the local spin density approximation (LSDA), the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation, and the strongly-constrained and appropriately normed (SCAN) meta-GGA, we have found that application of SIC generally improves predictions, but typically results in overshooting reference values. In this talk, we examine this overcorrection and discuss simple approaches to overcome it. |
Monday, March 2, 2020 10:24AM - 10:36AM |
A58.00011: Fermi-Löwdin self-interaction correction for ground and excited states Santosh Adhikari, Rajendra R Zope, Carlos Diaz, Adrienn Ruzsinszky Self-interaction error (SIE) is strongly present in ground-state density functional theory (DFT) and propagates into time-dependent DFT as well. A significant requirement of the self-interaction correction (SIC) is the localization of the orbitals, as required by size-consistency. In 2014, Pederson and collaborators recommended a computationally much more attractive version of the earlier Perdew-Zunger (PZ) SIC [1]. This SIC utilizes Fermi orbitals, greatly restricting the number of possible unitary transformations of the occupied Kohn-Sham orbitals. This work is a straightforward continuation of the Fermi orbital PZ-SIC to test crucial physical properties relevant for both ground, and excited states, with the potential to eliminate SIE for excitation energies. |
Monday, March 2, 2020 10:36AM - 10:48AM |
A58.00012: Assisting the performance of self-interaction-corrected SCAN meta-GGA functional using an orbital scaling approach Yoh Yamamoto, Rajendra Zope, Tunna Baruah Semilocal density functional approximations (DFAs) suffer from self-interaction (SI) error which limits their applicability. The Perdew-Zunger SI correction (PZSIC) improves the description of properties such as barrier heights or dissociation energies where SI error is dominant but it degrades already good performance of DFAs when applied to GGA and meta-GGA functionals. We implemented the orbital scaling approach of Vydrov et al.[1] that scales down orbital wise PZSIC in the FLOSIC code and use it with SCAN functional to assess its performance for a wide range of properties such as total energies, ionization potentials, and electron affinities for atoms, and atomization energies, dissociation and reaction energies, and reaction barrier heights of molecules. Our results show that orbital scaling recovers the performance of SCAN that is lost with SIC. We suggest a modification to recover -1/r asymptotic behavior of the potential lost by the scaling scheme of Vydrov et al. We also discuss the cases where the orbital scaling is not successful. General strategies of applying the scaling and how to improve its performance will be discussed in this talk. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700