Energy-Specific Bethe-Salpeter Equation Implementation for Efficient Optical Spectrum Calculations

Christopher Hillenbrand,Jiachen Li,Tianyu Zhu

2024-11-01

Abstract:We present an energy-specific Bethe-Salpeter equation (BSE) implementation for efficient core and valence optical spectrum calculations. In energy-specific BSE, high-lying excitation energies are obtained by constructing trial vectors and expanding the subspace targeting excitation energies above the predefined energy threshold in the Davidson algorithm. To calculate optical spectra over a wide energy range, energy-specific BSE can be applied to multiple consecutive small energy windows, where trial vectors for each subsequent energy window are made orthogonal to the subspace of preceding windows to accelerate the convergence of the Davidson algorithm. For seven small molecules, energy-specific BSE combined with $G_0W_0$ provides small errors around 0.8 eV for absolute and relative $K$-edge excitation energies when starting from a hybrid PBEh solution with 45\% exact exchange. We further showcase the computational efficiency of this approach by simulating the N $1s$ $K$-edge excitation spectrum of the porphine molecule and the valence optical spectrum of silicon nanoclusters involving 5,000 excited states using $G_0W_0$-BSE. This work expands the applicability of the $GW$-BSE formalism for investigating high-energy excited states of large systems.

Materials Science,Chemical Physics,Computational Physics

What problem does this paper attempt to address?

The core problem that this paper attempts to solve is to improve the efficiency and accuracy of optical spectrum calculations in high - energy excited states (such as core - excited states) and over a wide energy range. Specifically, the authors propose an energy - specific Bethe - Salpeter equation (BSE) - based method, namely the energy - specific GW - BSE method, for efficiently calculating the optical spectra of the core and valence bands.

### Main Problems and Solutions

1. **Prediction Difficulties of High - Energy Excited States**:
- **Problem**: The traditional GW - BSE method uses the standard Davidson algorithm and has difficulty accurately predicting high - energy excited states, especially core - excited states measured in X - ray absorption spectra (XAS), because the required subspace is very large.
- **Solution**: By constructing test vectors for excitation energies above a predefined energy threshold and expanding the subspace in the Davidson algorithm, the problem of high - energy excited states is effectively solved.

2. **Efficiency Problem of Optical Spectrum Calculations for Large - Scale Systems**:
- **Problem**: For large - scale systems (such as nanoclusters), calculating optical spectra over a wide energy range requires simultaneously solving a large number of excited states, resulting in a significant decrease in computational efficiency.
- **Solution**: Apply the energy - specific BSE method to multiple small energy windows, with the test vectors of each subsequent energy window orthogonal to the true eigenvectors of the previous window, in order to accelerate the convergence of the Davidson algorithm.

### Method Overview

- **Energy - Specific BSE Method**:
- Construct test vectors for excitation energies above a predefined energy threshold.
- Use the Davidson algorithm to gradually expand the subspace.
- Calculate the optical spectra over a wide energy range step by step through multiple small energy windows, with the test vectors of each window orthogonal to the eigenvectors of the previous window.

- **Computational Efficiency Improvement**:
- Reduce the computational resources required for each iteration through the sliding - window technique.
- Further improve computational efficiency by using approximate methods such as Cholesky decomposition.

### Experimental Verification

- **K - Edge Excitation Energies of Small Molecules and Porphyrin Molecules**:
- The K - edge excitation energies of 7 small molecules and porphyrin molecules were calculated by the energy - specific GW - BSE method. The results show an error of approximately 0.8 eV, and the computational cost is comparable to that of low - energy excited - state calculations.

- **Valence - Band Spectra of Silicon Nanoclusters**:
- The valence - band spectra of the silicon nanocluster Si₃₅H₄₄ were calculated by applying the sliding - window technique, involving 5,000 excited states with a maximum energy of up to 11.5 eV. The entire calculation took approximately 6 hours and 40 minutes on a 48 - core CPU node, with a maximum memory usage of 117 GB.

### Conclusion

The energy - specific GW - BSE method proposed in this paper performs excellently in calculating high - energy excited states and optical spectra over a wide energy range. It not only improves the computational accuracy but also significantly enhances the computational efficiency. This method expands the application range of the GW - BSE formalism in the study of high - energy excited states of large - scale systems.