Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Benchmarking the plasmon-pole and multipole approximations in the Yambo Code using the GW100 dataset

M. Bonacci, D. A. Leon, N. Spallanzani, E. Molinari, D. Varsano, A. Ferretti, C. Cardoso

TL;DR

This study benchmarks G0W0 quasiparticle calculations in the Yambo code against the GW100 dataset, focusing on two frequency treatments: GN-PPA and MPA. By comparing ionization potentials and electron affinities to reference GW100 results and other codes, it shows that GN-PPA achieves ~190 meV and MPA ~143 meV mean deviations, with MPA providing accuracy close to full-frequency methods at a fraction of the computational cost. The work employs automated high-throughput workflows and robust convergence extrapolations to ensure reproducibility, and discusses remaining discrepancies arising from finite-size effects and starting-point limitations. Overall, GN-PPA and especially MPA are validated as reliable and efficient options for large-scale molecular GW calculations in the Yambo framework, guiding future developments toward full-frequency GW on GPU architectures.

Abstract

Verification and validation of electronic structure codes are essential to ensure reliable and reproducible results in computational materials science. While density functional theory has been extensively benchmarked, systematic assessments of many-body perturbation theory methods such as the GW approximation have only recently emerged, most notably through the GW100 dataset. In this work, we assess the numerical accuracy and convergence behavior of the GW implementation in the yambo code using both the Godby-Needs plasmon-pole model and the recently introduced multipole approximation. Quasiparticle energies are compared against GW100 reference data to evaluate the performance, numerical stability, and consistency of these approaches.

Benchmarking the plasmon-pole and multipole approximations in the Yambo Code using the GW100 dataset

TL;DR

This study benchmarks G0W0 quasiparticle calculations in the Yambo code against the GW100 dataset, focusing on two frequency treatments: GN-PPA and MPA. By comparing ionization potentials and electron affinities to reference GW100 results and other codes, it shows that GN-PPA achieves ~190 meV and MPA ~143 meV mean deviations, with MPA providing accuracy close to full-frequency methods at a fraction of the computational cost. The work employs automated high-throughput workflows and robust convergence extrapolations to ensure reproducibility, and discusses remaining discrepancies arising from finite-size effects and starting-point limitations. Overall, GN-PPA and especially MPA are validated as reliable and efficient options for large-scale molecular GW calculations in the Yambo framework, guiding future developments toward full-frequency GW on GPU architectures.

Abstract

Verification and validation of electronic structure codes are essential to ensure reliable and reproducible results in computational materials science. While density functional theory has been extensively benchmarked, systematic assessments of many-body perturbation theory methods such as the GW approximation have only recently emerged, most notably through the GW100 dataset. In this work, we assess the numerical accuracy and convergence behavior of the GW implementation in the yambo code using both the Godby-Needs plasmon-pole model and the recently introduced multipole approximation. Quasiparticle energies are compared against GW100 reference data to evaluate the performance, numerical stability, and consistency of these approaches.
Paper Structure (11 sections, 12 equations, 3 figures, 5 tables)

This paper contains 11 sections, 12 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Computational Approach
  3. The GW100 dataset
  4. DFT setup
  5. The quasiparticle equation
  6. GW self-energy evaluation
  7. Convergences and workflow
  8. Results
  9. Ionization Potentials
  10. Electron Affinities
  11. Conclusions

Figures (3)

  • Figure 1: Convergence tests for the LiF molecule. Left panel: Effect of E$_{GT}$ (see text) on the convergence acceleration of the HOMO with respect to empty states summation (the value at N$_b$=1000 for E$_{GT}$=0.25 Ha is set to zero in the plot). Right panel: HOMO energy versus the PW cutoff $G_{cut}$ in the screening matrix, for different number of empty states, N$_b$. The result obtained with N$_b$=16 Ry $G_{cut}$=36 Ry is set to zero. The values N$_b$ = 3, 9, 12, 14, 16 Ry correspond to 1000, 5000, 7000, 9000 and 11000 bands.
  • Figure 2: Violin plots representing the distribution of the IP deviation between yambo PPA and MPA results and other GW codes. The width of each curve reflects the density of data points. In the MPA case, the vertical black box represents the inter-quartile range, and the white horizontal line indicates the median.
  • Figure 3: Deviation of the quasiparticle electron affinity (EA) between the yambo and the WEST linearized extrapolated results.