All-electron Dynamical Bethe-Salpeter Equation for Extended Systems with Atom-centered Orbital Basis Set

This paper presents an all-electron numerical atom-centered orbital implementation of the dynamical Bethe-Salpeter equation for extended systems, which incorporates dynamical screening effects and is validated through applications to molecular crystals like naphthalene.

The Gist

Imagine you are trying to predict how a crowd of people (electrons) in a stadium (a crystal) will react when a loud cheer (light) goes up. In the world of quantum chemistry, this is called calculating "excited states."

For a long time, scientists have used a popular method called the Bethe-Salpeter Equation (BSE) to solve this. Think of the BSE as a rulebook for how two people in the crowd—a cheerleader and a heckler (an electron and a "hole" where an electron used to be)—interact.

The Problem: The "Instant" vs. "Real-Time" Rule

The standard rulebook assumes that when the cheerleader and heckler interact, it happens instantly. It's like saying, "If I wave my hand, you see it the exact same nanosecond." This is called the static approximation.

However, in reality, there is a tiny, split-second delay. The crowd doesn't react instantly; there's a ripple effect. In physics, this is called dynamical screening. For most materials, this delay is so small we can ignore it. But for certain materials, like organic crystals (think of a block of naphthalene, the stuff in mothballs), this delay is huge. The "ripple" matters. If you ignore it, your prediction of how the material absorbs light is wrong.

The problem is that calculating this "real-time" delay is incredibly expensive. It's like trying to film every single person in the stadium reacting to every single cheer in slow motion. It takes so much computer power that scientists usually can't do it for large, solid materials.

The Solution: A Smarter Shortcut

The authors of this paper, led by Ruiyi Zhou and Yosuke Kanai, have built a new, super-efficient way to calculate this "real-time" delay without needing a supercomputer the size of a city.

They took a clever shortcut method that was previously only available for a specific type of math (using "plane waves," which are like smooth, rolling ocean waves) and translated it into a new language they call Numerical Atom-Centered Orbitals (NAO).

Here is the analogy:

• The Old Way (Plane Waves): Imagine trying to describe the shape of a mountain by measuring the height of the water at every single point on a perfectly flat grid. It's accurate but requires measuring millions of points.
• The New Way (NAO): Imagine describing that same mountain by placing a few specific, detailed sculptures (atoms) on the ground and measuring how they fit together. It's much more efficient for complex shapes like molecules.

The authors successfully taught their "sculpture-based" system how to handle the "real-time delay" (dynamical screening) using a method called the Effective Dielectric Function. Instead of simulating the delay second-by-second, they calculate a single "average delay" value that captures the essence of the interaction perfectly.

The "Symmetry" Trick

Even with their new shortcut, calculating the delay for every single direction in the crystal is still too slow. So, they added a second trick: Symmetry Mapping.

Imagine a snowflake. It has six identical arms. If you know how one arm reacts to heat, you automatically know how the other five react because they are identical. You don't need to test all six.
The authors realized that the crystal they were studying (naphthalene) has similar symmetries. Instead of calculating the interaction for every single point in the crystal's "map" (the Brillouin Zone), they only calculated it for the unique, non-repeating parts (the Irreducible Brillouin Zone). They then used math to "mirror" those results to fill in the rest of the map.

This reduced the amount of work by about 70%, making the calculation fast enough to be practical.

The Proof: Mothball Crystals

To prove their method works, they tested it on crystalline naphthalene.

1. They compared their new "sculpture-based" method against the old "ocean wave" method. The results were almost identical (within a tiny margin of error), proving their translation was successful.
2. They then ran the full "real-time" calculation. They found that including the delay (dynamical screening) changed the color of light the crystal absorbs. Specifically, it shifted the energy of the light absorption by about 0.12 electron-volts.

Why This Matters

This paper doesn't claim to cure diseases or build new batteries today. Instead, it provides a new, faster, and more accurate tool for scientists who study how solid materials (like organic crystals) interact with light.

By making the "real-time" calculation possible for complex, extended systems, they have removed a major roadblock. Now, researchers can study materials with strong "electron-hole" interactions (like those found in organic electronics) with much higher precision than before, without waiting weeks for a computer to finish the math.

In short: They took a very slow, complex calculation, translated it into a more efficient language, and added a "mirror trick" to speed it up, allowing scientists to finally see the subtle, real-time interactions of electrons in solid crystals.

Technical Summary

Technical Summary: All-electron Dynamical Bethe-Salpeter Equation for Extended Systems with Atom-centered Orbital Basis Set

Problem Statement
The Bethe-Salpeter equation (BSE) is the standard approach within many-body perturbation theory for calculating electronic excitations by accounting for particle-hole (exciton) interactions. However, standard BSE calculations typically employ a static approximation for the screened Coulomb interaction kernel, neglecting its frequency dependence. While this approximation is valid for transitions with energies much smaller than the plasma frequency, it becomes questionable for systems with significant excitonic character, such as molecules, low-dimensional systems, and organic solids, where exciton binding energies are large. In these cases, dynamical screening effects can appreciably alter excitation energies and spectral shapes.

Solving the full dynamical BSE for extended systems is computationally prohibitive due to the necessity of dense Brillouin zone (BZ) integration for convergence, particularly when combined with GW calculations for quasiparticle energies. Existing methods to incorporate dynamical effects, such as exact diagonalization in an expanded space or perturbation theory, either suffer from high-order computational scaling or are limited to small systems. Consequently, there is a need for a computationally efficient, accurate method to treat dynamical screening in extended periodic systems, specifically within all-electron frameworks that utilize atom-centered orbital (NAO) basis sets, which are advantageous for chemical applications.

Methodology
The authors formulate and implement an effective dielectric function method for the dynamical BSE, adapting a approach originally developed by Zhang et al. (Phys. Rev. B 107, 235205 (2023)) for plane-wave (PW) bases to an all-electron numerical atom-centered orbital (NAO) framework.

1. Theoretical Framework:

   • The method relies on Shindo's approximation to decouple frequency dependencies in the two-frequency polarization function.
   • Instead of solving the complex coupled eigenvalue problem required for the frequency-dependent BSE, the authors replace the energy difference terms in the screened Coulomb interaction with the exciton binding energy ($E_b$) of the lowest exciton state.
   • This leads to an effective, frequency-independent screened Coulomb operator ($\hat{W}^{\text{eff}}$) derived from an effective inverse dielectric function ($\epsilon^{-1}_{\text{eff}}$).
   • The effective dielectric function is constructed using a plasmon-pole model, where the frequency integration is performed analytically.

2. Implementation in NAO Basis:

   • The authors implement this method within their all-electron BSE@GW code using the resolution-of-identity (RI) technique with atom-centered auxiliary basis functions (ABFs).
   • Unlike the plane-wave formulation which often employs a diagonal approximation (neglecting local-field effects by assuming $G=G'$), the NAO implementation retains the full, dense, non-diagonal dielectric matrix in the auxiliary basis.
   • To handle the non-linear functional transformation of the dielectric response required for the effective model, the authors perform a spectral decomposition of the static inverse dielectric matrix. The transformation is applied to the eigenvalues, and the matrix is reconstructed in the auxiliary basis. This allows for a rigorous treatment of local-field effects.

3. Computational Optimization (IBZ Mapping):

   • To address the bottleneck of dense BZ sampling required for the screened Coulomb interaction $W(q)$, the authors employ a symmetry-adapted strategy.
   • They calculate the static dielectric function and the resulting effective screened interaction only at $q$-points within the irreducible Brillouin zone (IBZ).
   • Matrices for the full BZ are recovered via crystallographic symmetry operations (rotation and phase shifts) applied to the IBZ matrices. This avoids redundant calculations for symmetry-equivalent points.

Key Contributions

• Formulation: The paper presents the first formulation of the effective dielectric function method for dynamical BSE calculations using an all-electron NAO basis set, extending the method beyond its original plane-wave implementation.
• Algorithmic Development: The authors developed a specific algorithm to handle the non-diagonal dielectric matrix in the auxiliary basis, enabling the inclusion of local-field effects which are often neglected in diagonal approximations.
• Symmetry Adaptation: The integration of IBZ-to-full-BZ mapping for the construction of the screened Coulomb interaction significantly reduces the computational cost of the dense $q$-point sampling required for convergence in extended systems.
• Implementation: The method is implemented in the FHI-aims code, building upon their existing all-electron BSE@GW infrastructure.

Results
The method was validated and demonstrated on crystalline naphthalene, an organic crystal known for strong excitonic effects (binding energy $\sim$1 eV).

1. Validation against Plane-Wave Results:

   • The authors compared their NAO-based dynamical BSE results (using a "scissor-shifted" DFT+$\Delta$ approach) against the reference PW-based results by Zhang et al.
   • The optical absorption spectra showed excellent agreement, with peak positions and intensities matching within 30 meV.
   • The computed exciton binding energies differed by less than 30 meV from the reference, validating the accuracy of the NAO implementation.

2. Efficiency of IBZ Mapping:

   • For naphthalene (monoclinic, $C_{2h}$ symmetry), the IBZ mapping reduced the number of required $q$-points from 175 (full BZ) to 52 (IBZ) for a $5\times7\times5$ grid.
   • Tests confirmed that optical absorption spectra obtained via IBZ mapping were identical to those from full BZ sampling, while significantly accelerating the calculation of the screened Coulomb interaction.

3. Dynamical BSE@G$_0$W$_0$ Application:

   • The authors performed a single-shot G$_0$W$_0$ calculation followed by static and dynamical BSE calculations.
   • The inclusion of dynamical screening resulted in a redshift of the excitonic peak by approximately 0.12 eV (from 4.39 eV to 4.27 eV) compared to the static BSE result.
   • This redshift corresponds to an increase in the exciton binding energy, indicating a strengthened electron-hole interaction when dynamical effects are included.
   • The results were consistent with trends observed in perturbation-based studies of small molecules, confirming the method's ability to capture dynamical effects in extended systems.

Significance and Claims
The paper claims that this work successfully bridges the gap between the computational efficiency of effective dielectric function methods and the accuracy of all-electron NAO basis sets for extended systems. By validating the implementation against established plane-wave results and demonstrating its application to a challenging organic crystal, the authors establish a robust framework for studying dynamical screening effects in molecular crystals and other extended materials.

The authors modestly state that while the current implementation is a significant step forward, future work will focus on further enhancing scalability and efficiency for dense BZ sampling. They propose incorporating advanced numerical techniques, such as iterative eigensolvers, coarse-grained k-point grid interpolation, and Wannier interpolation, to extend the applicability of the BSE@GW method to increasingly complex materials. The paper does not claim to have solved all computational challenges but presents a validated, practical approach for tackling dynamical effects in systems where they are most relevant.