Epistemic uncertainty as a diagnostic probe for deep learning Hamiltonians
Nanoscale · 2026
DOI: 10.1039/d6nr02280d ↗Kurz gefasst
Fachartikel in der Zeitschrift Nanoscale, 2026. Verfasst von Tianhao Su und 3 weiteren Autor:innen. Beteiligte Einrichtung: Material Genome Institute, Institute for Quantum Science and Technology, Physics Department, Shanghai University, Shanghai 200444, China. shunbohu@sh…. Die Arbeit nennt 31 Literaturangaben. Gefördert von: National Key Research and Development Program of China.
Abstract (Originalsprache)
Deep ensemble disagreement, quantified by ensemble variance, correlates strongly with prediction error and serves as a practical DFT-free diagnostic of model reliability, underscoring the need to establish applicability boundaries before deployment.
Autor:innen und Institutionen
- Material Genome Institute, Institute for Quantum Science and Technology, Physics Department, Shanghai University, Shanghai 200444, China. shunbohu@shu.edu.cn.
- Institute for the Conservation of Cultural Heritage, Institute for Quantum Science and Technology, Key Laboratory of Silicate Cultural Heritage Conservation (Shanghai University), Ministry of Education, Shanghai 200444,…
Förderung
- National Key Research and Development Program of China
Zitieren
APA
Su, T., Gao, Z., Hu, S., & Zhang, T. Y. (2026). Epistemic uncertainty as a diagnostic probe for deep learning Hamiltonians. Nanoscale. https://doi.org/10.1039/d6nr02280d
Vancouver
Su T, Gao Z, Hu S, Zhang TY. Epistemic uncertainty as a diagnostic probe for deep learning Hamiltonians. Nanoscale. 2026. doi:10.1039/d6nr02280d
BibTeX
@article{Su2026,
author = {Su, Tianhao and Gao, Zeyun and Hu, Shunbo and Zhang, Tong-Yi},
title = {Epistemic uncertainty as a diagnostic probe for deep learning Hamiltonians},
journal = {Nanoscale},
year = {2026},
doi = {10.1039/d6nr02280d}
}
EUVIMED