Related Experiment Video
Updated: Jun 5, 2025

Computation of Atmospheric Concentrations of Molecular Clusters from ab initio Thermochemistry
Published on: April 8, 2020
GPU-Accelerated Solution of the Bethe-Salpeter Equation for Large and Heterogeneous Systems
Victor Wen-Zhe Yu1, Yu Jin2, Giulia Galli1,2,3
1Materials Science Division, Argonne National Laboratory, Lemont, Illinois 60439, United States.
We developed a GPU-accelerated Bethe-Salpeter equation (BSE) method to efficiently calculate excitation energies and optical spectra for materials. This approach enables accurate simulations of large systems, crucial for understanding material properties.
Area of Science:
- Computational Physics
- Materials Science
- Quantum Chemistry
Background:
- Accurate calculation of material properties like excitation energies and optical spectra is essential for materials design.
- Traditional methods for solving the Bethe-Salpeter equation (BSE) are computationally intensive, limiting system sizes.
- Density functional theory (DFT) provides a foundation but requires post-processing for excited-state properties.
Purpose of the Study:
- To present a massively parallel, GPU-accelerated implementation of the Bethe-Salpeter equation (BSE).
- To enable efficient and accurate computation of vertical excitation energies (VEEs) and optical absorption spectra for large condensed and molecular systems.
- To investigate the impact of system size and defect interactions on material properties.
Main Methods:
- Developed a GPU-accelerated BSE solver utilizing density matrix perturbation theory and low-rank decomposition of the screened Coulomb interaction.
- Implemented algorithms to circumvent slow convergence issues with empty/occupied states and large dielectric matrix inversions.
- Exploited the nearsightedness of the density matrix for semiconductors and insulators to reduce computational cost.
- Employed a hierarchical loop and data distribution strategy for scaling calculations to thousands of GPUs.
Main Results:
- Demonstrated the method's efficacy by computing VEEs for spin defects in wide-band-gap materials, requiring supercells up to 1000 atoms for convergence.
- Validated the accuracy of the GW-BSE approach on large systems.
- Analyzed the symmetry breaking of triplet states in a diamond lattice with 1727 atoms due to defect interactions.
Conclusions:
- The developed GPU-accelerated BSE method offers significant computational savings and enables accurate calculations for large, complex material systems.
- Supercell sizes of up to 1000 atoms are necessary for converged VEE calculations in certain materials.
- The study provides insights into defect interactions and their influence on electronic and optical properties, particularly symmetry breaking in triplet states.
More Related Videos
10:52Multiscale Sampling of a Heterogeneous Water/Metal Catalyst Interface using Density Functional Theory and Force-Field Molecular Dynamics
Published on: April 12, 2019
06:37Analyzing Melts and Fluids from Ab Initio Molecular Dynamics Simulations with the UMD Package
Published on: September 17, 2021
Related Concept Videos
Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
The Small x Assumption
Chemical Equilibria: Systematic Approach to Equilibrium Calculations
The first step is to identify all the chemical reactions involved, The...
Calculating Equilibrium Concentrations
A more...