Comparative analysis of methods for calculating Hubbard parameters using cRPA

Indukuru Ramesh Reddy, M. Kaltak, Bongjae Kim · 2025

Read Comparative analysis of methods for calculating Hubbard parameters using cRPA by Indukuru Ramesh Reddy, M. Kaltak, Bongjae Kim (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-11142.txt.

Introduction

Simulating materials with strong electron interactions needs a key number, called U, that measures how much electrons repel each other. Indukuru Ramesh Reddy, Kaltak, and Bongjae Kim compare ways of computing that number in a 2025 paper, focusing on a widely used screening method. The hard cases are materials where the interesting electron bands tangle with others, so the target space must be carved out by projection.

They test several projection and screening choices on two families of oxides, one with clean bands and one with tangled bands. Choices about the target space shift U a great deal. The authors map which methods suit which materials.

Catalog record

CREATORIndukuru Ramesh Reddy, M. Kaltak, Bongjae Kim
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE47 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-11142

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Comparative analysis of methods for calculating Hubbard parameters using cRPA Indukuru Ramesh Reddy, M. Kaltak, Bongjae Kim Abstract In this study, we present a systematic comparison of various approaches within the constrained random-phase approximation (cRPA) for calculating the Coulomb interaction parameter U U . While defining the correlated space is straightforward for disentangled bands, the situation is more complex for entangled bands, where different projection schemes from hybridized bands to the target space can yield varying sizes of interaction parameters. We systematically evaluated different methods for calculating the polarizability functions within the correlated space. Furthermore, we analyze how different definitions of the correlated space, often constructed through Wannierization from Kohn-Sham orbitals, defines the orbital localization and play a crucial role in determining the interaction parameter. To illustrate these effects, we consider two sets of representative correlated d d -orbital oxides: Li M M O 2 ( M M = V–Ni) as examples of isolated d d -electron systems and Sr M M O 3 ( M M = Mn, Fe, and Co) as cases of entangled d d -electron systems. Through this systematic comparison, we provide a detailed analysis of different cRPA methodologies for computing the Hubbard parameters. Preprint: APS/123-QED I Introduction The density functional theory (DFT) is one of the most widely employed theoretical approaches for providing a realistic description of materials. However, an important shortcoming of this method lies in its treatment of correlated systems, where the innate mean-field approach of DFT is not capable to accurately capture correlated electron motions. For a better description of the strongly correlated systems without losing the material specificity, the incorporation of the many-body Hubbard model into DFT, using both static and dynamical treatment of the Hubbard parameter U , has been implemented within the framework of DFT+ U [ 1 , 2 ] and DFT+DMFT [ 3 , 4 ] , and produced great success in various applications [ 5 , 6 , 7 , 8 ] . The accuracy of these methods critically depends on the precise quantification of the interaction parameter, U . Several first-principles approaches have been proposed to determine parameter-free interaction parameters, including the constrained local density approximation (cLDA) [ 9 , 10 , 11 , 12 ] , linear response [ 13 , 14 ] ,constrained random phase approximation (cRPA) [ 15 , 16 ] , pseudohybrid Hubbard density functional [ 17 ] , and a hybrid method between cLDA and cRPA [ 18 ] . Among these methods, cRPA has been actively employed in the correlated electron community due to its close connection to Hubbard model approaches. The U parameter in cRPA is obtained for a well-defined set of correlated bands. Key features of cRPA include the ability to select a low-energy subspace and identify individual interaction matrix elements (on-site, inter-site, interorbital, intra-orbital, and exchange) [ 19 , 1 , 21 ] ; the ability to analyze individual screening channels [ 22 ] ; and its formulation to calculate frequency-dependent U(\omega) [ 23 , 24 , 25 , 26 , 27 , 28 ] . Such features make cRPA particularly suitable for DFT extensions such as DFT+DMFT [ 29 , 30 ] . Despite the advantages of the cRPA approach, a few considerations must be addressed. In the cRPA procedure, one starts with a DFT calculation, and the Kohn–Sham bands are projected onto the correlated space, which then serves as the basis for computing the Coulomb interaction parameters. Two key issues in this process are the projection procedure itself and the definition of the correlated bands . As one starts from the hybridized Kohn-Sham orbitals, projection into a few bands, or localized basis, is required. A typical practice is the projection of the Kohn–Sham wavefunctions onto Wannier functions (WFs), a Bloch sum of orbitals [ 31 ] , mostly conveniently the maximally localized Wannier functions (MLWFs) [ 2 , 33 ] . However, there are multiple ways to perform these projections, and they have not been systematically compared. This oversight is significant because different projection schemes yield different local basis, resulting in different Hubbard interaction parameters, hence the electronic structures of a system [ 34 , 35 ] . Another issue is the definition of the correlated space. In practice, constructing WFs for the correlated bands is not straightforward, as these bands are typically defined by their proximity to the Fermi level. The choice of the energy window used to define the correlated space and the subsequent construction of MLWFs is not unique, especially for hybridized bands. For example, in transition metal oxide materials, d -bands are generally considered as the correlated bands; however, the bands near the Fermi energy often have significant contributions from ligand O- p orbitals, resulting in various approaches to constructing MLWFs for the system [ 36 , 37 ] . High-density k -point sampling enhances the accuracy of cRPA calculations but simultaneously increases computational demands. To accelerate k -point convergence, one might consider employing the derivative of the wavefunction; however, the validity of this approach has not been thoroughly investigated. In this work, we systematically compare various methodologies within the cRPA framework for calculating the Hubbard parameter U . Our analysis focuses on the impact of different projection schemes and the correlated space definitions on the calculated interaction parameters. We demonstrate that different ways of defining the local orbitals can play a crucial role in determining the localization of correlated orbitals, particularly in systems with hybridized bands. To illustrate these effects, we investigate two representative families of d -orbital oxides: Li M O 2 ( M = V–Ni) as prototypical isolated d -electron systems and Sr M O 3 ( M = Mn, Fe, and Co) as examples of entangled d -electron systems. Our comprehensive study provides detailed insights into the optimal selection of cRPA methodologies to accurately determine the interaction parameters in strongly correlated materials. II Theoretical framework II.1 Brief review on cRPA The starting point in cRPA is the choice of an effective low-energy Hilbert space, i.e., the target space, such as the d -orbitals in typical transition metal complexes. This target space is usually defined as the correlated space. The method then excludes the screening channels of the correlated space from the total polarizability function P . The rest space polarizability ( P^{r} ) is defined as P^{r}=P-P^{C} , where P^{C} is the polarizability contribution from the screening within correlated space. The partially screened Coulomb interaction kernel U is then calculated by solving the Bethe-Salpeter equation: U^{-1}=[U^{bare}]^{-1}-P^{r} , where U^{bare} is the unscreened Coulomb interaction kernel. And the Coulomb interaction matrix elements among the obtained Wannier orbitals are then evaluated using the following equation: \begin{split}U_{ijkl}(\omega)=\iint\mathrm{d}\mathbf{r}\mathrm{d}\mathbf{r}^{\prime}\mathcal{W}_{i}^{*}(\textbf{r})\mathcal{W}_{k}^{*}(\textbf{r}^{\prime})\\ U(\textbf{r},\textbf{r}^{\prime},\omega)\mathcal{W}_{j}(\textbf{r})\mathcal{W}_{l}(\textbf{r}^{\prime})\end{split} (1) where \omega is the frequency and \mathcal{W}(\textbf{r}) represent the correlated orbitals, which are the MLWFs. Although cRPA is a general approach, its application to materials is often complicated by issue of defining the correlated orbitals, which are employed to obtain the polarizability of the correlated space, P^{C} . The d bands that span the correlated subspace often mix with s -states [ 2 ] , and in some compounds, extended p - states of ligands, making it difficult to uniquely identify the d - d transitions to calculate P^{C} .

Read the complete text