PreprintCC BY 4.0arXiv
A fast summation method for the DFT-D3 dispersion correction
Abstract
The DFT-D3 dispersion correction is routinely added to machine learning force fields (MLFFs) trained on dispersion-deficient functionals such as PBE. Its environment-dependent pair coefficients, however, break the atom-centered separability that fast summation methods require, forcing practitioners either to truncate D3 or to accept a substantial slowdown. We introduce FourierD3, a method that uses a functional low-rank decomposition to restore this separability and enable particle-mesh evaluation in O(Nlog N) time without a real-space cutoff on the dispersion sum.
Related
- Plasmonic Lithography Optical Proximity Correction Method Based on Decomposed Machine Learning Fast Imaging Model
- FaultScope: A software package for fast error attribution in quantum error correction
- Generic phase correction method based on phase recoding for deflectometry
- PyDoseRT Proton: A GPU Pencil-Beam Engine with a Convolutional Residual-Correction Network for Fast Proton Dose Calculation
- Updated Tkatchenko Scheffler method and assessment of dispersion methods for alkali-containing systems
BibTeX
@misc{valeeva2026fast,
title = {{A fast summation method for the DFT-D3 dispersion correction}},
author = {Victoria Valeeva and Cheuk Hin Ho and Mario Geiger and Franco Pellegrini and Gábor Csányi and Emine Kucukbenli and Christoph Ortner},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2607.15103},
eprint = {2607.15103},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.15103},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A fast summation method for the DFT-D3 dispersion correction
AU - Victoria Valeeva
AU - Cheuk Hin Ho
AU - Mario Geiger
AU - Franco Pellegrini
AU - Gábor Csányi
AU - Emine Kucukbenli
AU - Christoph Ortner
PY - 2026
DA - 2026/10/18
PB - arXiv
DO - 10.48550/arxiv.2607.15103
UR - https://arxiv.org/abs/2607.15103
AB - The DFT-D3 dispersion correction is routinely added to machine learning force fields (MLFFs) trained on dispersion-deficient functionals such as PBE. Its environment-dependent pair coefficients, however, break the atom-centered separability that fast summation methods require, forcing practitioners either to truncate D3 or to accept a substantial slowdown. We introduce FourierD3, a method that uses a functional low-rank decomposition to restore this separability and enable particle-mesh evaluation in O(Nlog N) time without a real-space cutoff on the dispersion sum.
ER -