LDA-1/2 for Molecular Systems: A Real-Space Finite-Element Benchmark on the GW100 Set
Abstract
The LDA-1/2 method provides an efficient correction to semilocal density functional theory for improving ionization energies and band gaps, yet its application to molecular systems has remained limited. In this work, we present an all-electron finite-element implementation of LDA-1/2 within the NESSIE electronic-structure framework and apply it to the GW100 molecular benchmark set with systematically controllable numerical accuracy. The self-energy correction is constructed explicitly from neutral and half-ionized calculations for each molecule, avoiding the use of precomputed atomic correction potentials. The real-space finite-element formulation enables systematic convergence with respect to the discretization and provides a controlled assessment of LDA-1/2 performance. For the GW100 set, the present implementation yields a mean absolute error of 0.472 eV and a root-mean-square error of 0.645 eV relative to CCSD(T) reference ionization energies, substantially improving upon conventional LDA and the previously reported LAPW implementation of LDA-1/2, while achieving accuracy comparable to G0W0@PBE. Convergence tests show that third-order finite elements are sufficient to reach or approach chemical accuracy relative to higher-order calculations for the representative systems considered. The resulting corrected Hamiltonian also improves several lower lying valence states relative to LDA, although the improvement becomes less systematic away from the HOMO. This work provides accurate LDA-1/2 benchmark data for molecular systems and establishes a rigorous finite-element foundation for future molecular GW calculations.
Related
- The efficiency–coverage trade-off of scientific priors in Bayesian search: a registered benchmark on metallic multi-element elasticity data
- Small-Strain Tensile Characterization and Finite Element Calibration of Sodium Alginate-Based Biomimetic Artificial Muscle Strips
- Accuracy, Reliability, and Efficiency of a MACE Machine-Learning Interatomic Potential for Aspirin: A Reproducible Benchmark for Molecular Simulation
- Finite element modeling of transient hydrogen trapping in a network of randomly sized voids
- Finite Element Modeling of Ceramic Green Part Warping Induced by Shrinkage During the Stereolithography Printing Process
BibTeX
@misc{li2026lda12,
title = {{LDA-1/2 for Molecular Systems: A Real-Space Finite-Element Benchmark on the GW100 Set}},
author = {Dongming Li and Niamh Matthews and Qingchuan Sang and Eric Polizzi},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.27643},
eprint = {2608.27643},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.27643},
}
RIS
TY - UNPB
TI - LDA-1/2 for Molecular Systems: A Real-Space Finite-Element Benchmark on the GW100 Set
AU - Dongming Li
AU - Niamh Matthews
AU - Qingchuan Sang
AU - Eric Polizzi
PY - 2026
DA - 2026/08/27
PB - arXiv
DO - 10.48550/arxiv.2608.27643
UR - https://arxiv.org/abs/2608.27643
AB - The LDA-1/2 method provides an efficient correction to semilocal density functional theory for improving ionization energies and band gaps, yet its application to molecular systems has remained limited. In this work, we present an all-electron finite-element implementation of LDA-1/2 within the NESSIE electronic-structure framework and apply it to the GW100 molecular benchmark set with systematically controllable numerical accuracy. The self-energy correction is constructed explicitly from neutral and half-ionized calculations for each molecule, avoiding the use of precomputed atomic correction potentials. The real-space finite-element formulation enables systematic convergence with respect to the discretization and provides a controlled assessment of LDA-1/2 performance. For the GW100 set, the present implementation yields a mean absolute error of 0.472 eV and a root-mean-square error of 0.645 eV relative to CCSD(T) reference ionization energies, substantially improving upon conventional LDA and the previously reported LAPW implementation of LDA-1/2, while achieving accuracy comparable to G0W0@PBE. Convergence tests show that third-order finite elements are sufficient to reach or approach chemical accuracy relative to higher-order calculations for the representative systems considered. The resulting corrected Hamiltonian also improves several lower lying valence states relative to LDA, although the improvement becomes less systematic away from the HOMO. This work provides accurate LDA-1/2 benchmark data for molecular systems and establishes a rigorous finite-element foundation for future molecular GW calculations.
ER -