WISP: Learning spin-resolved electron localization from atomic environments
Abstract
The electron localization function (ELF) reveals bonding and interstitial localization in crystals, but obtaining it usually requires an electronic-structure calculation. Here we develop WISP, Wavefunction-free Inference of Scalar Properties, in which a nonlinear decoder predicts spin-resolved ELF from encoded atomic environments at arbitrarily chosen positions. Signed site moments supply the magnetic information, and two evaluations of a shared network with opposite moments enforce exact spin-channel exchange. We obtain a structure-weighted paired-channel mean absolute error (MAE) of 0.01065 for the model with far-field input on the complete grids of 21,922 held-out structures from a frozen MatPES-derived PBE dataset. With moments predicted from geometry by a separate network, we obtain an average-field MAE of 0.00974. The predictions retain the spin contrast of the magnetic examples and resolve the interstitial localization for electride screening. We propose using predicted ELF to select structures for further electronic-structure calculations.
Related
- Visualizing and Quantifying Atomic Contributions to Raman Intensities governed by Spatially-Resolved Atomic Interferences
- Accurate Atomic Correlation Energies and Virial Trends for Correlation-Consistent Effective Core Potentials III: Quantitative Separation of Fixed-Node Biases and Localization Errors in Cs-Rn
- Integrating Ion Mobility–Resolved Tandem Mass Spectrometry and Microcrystal Electron Diffraction for Structural Elucidation
- Atomic-Scale Mapping of an Emergent Interfacial Two-Dimensional Electron Gas at Anatase/LaAlO3
- Geometry‐Based Neural‐Network Prediction of Electron Localization Function Topology in Dense Hydrogen
BibTeX
@misc{ellis2026wisp,
title = {{WISP: Learning spin-resolved electron localization from atomic environments}},
author = {Austin Ellis},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15011092/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15011092/v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - WISP: Learning spin-resolved electron localization from atomic environments
AU - Austin Ellis
PY - 2026
DA - 2026/10/27
PB - ChemRxiv
DO - 10.26434/chemrxiv.15011092/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15011092/v1
AB - The electron localization function (ELF) reveals bonding and interstitial localization in crystals, but obtaining it usually requires an electronic-structure calculation. Here we develop WISP, Wavefunction-free Inference of Scalar Properties, in which a nonlinear decoder predicts spin-resolved ELF from encoded atomic environments at arbitrarily chosen positions. Signed site moments supply the magnetic information, and two evaluations of a shared network with opposite moments enforce exact spin-channel exchange. We obtain a structure-weighted paired-channel mean absolute error (MAE) of 0.01065 for the model with far-field input on the complete grids of 21,922 held-out structures from a frozen MatPES-derived PBE dataset. With moments predicted from geometry by a separate network, we obtain an average-field MAE of 0.00974. The predictions retain the spin contrast of the magnetic examples and resolve the interstitial localization for electride screening. We propose using predicted ELF to select structures for further electronic-structure calculations.
ER -