\textbf{\textsc{HepUKE}}: Uncertainty-aware Knowledge Evolution for High Energy Physics Analysis Agents
Abstract
Knowledge bases are essential to agents for high energy physics (HEP), supplying few-shot evidence for previously unseen tasks. Their principal utility lies in strategic guidance for novel analyses rather than ready-made answers or reusable code. Prevailing systems, however, treat all retrieved evidence as equally trustworthy and fail to accrue knowledge from the analyses under assistance. In this paper, inspired by the standard HEP treatment of uncertainty, we propose \textbf{\textsc{\underline{HepUKE}}}: an \textbf{\underline{U}}ncertainty-aware \textbf{\underline{K}}nowledge \textbf{\underline{E}}volution system for \textbf{\underline{HEP}} that repurposes information entropy to unify statistical uncertainty from LLM generation and systematic uncertainty from HEP methodological choices. Upon it, we construct a self-evolving domain-specific-language (DSL) knowledge base and release a large-scale expert-annotated HEP physics-analysis benchmark. Per-evidence confidence is quantified via the information entropy of sampled DSL generations: high-confidence precedents constrain program synthesis, whereas analyses lacking close precedents are validated and written back, sustaining continual knowledge evolution. To handle the highly redundant yet logically stringent nature of HEP knowledge, we introduce an early-stopping controller that fuses a backward stationarity test on the answer posterior with a forward marginal information gain estimate. On the 1,000-pair BESIII PhysicsQA benchmark, HepUKE reduces retrieval rounds by 43.0\% while preserving most answer quality, and on the out-of-domain HotpotQA benchmark, it reduces retrieval rounds by 14.5\% with a similar quality--efficiency trade-off. The code and datasets used in this work are available at https://anonymous.4open.science/r/HepUKE-32FF.
Related
- A physics-informed uncertainty-aware surrogate for p-n junction diodes
- The \textsc{ABC}-equation revisited: missed integrable cases. Part I: classification and (quasi) integrable structures
- PEEL: Physics-Enabled Evidential Learning for Identifiable Uncertainty in CT Imaging
- SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty
- Transitioning from A-level to higher education physics courses: where are the knowledge gaps?
BibTeX
@misc{shang2026textbftextschepuke,
title = {{\textbf\textscHepUKE: Uncertainty-aware Knowledge Evolution for High Energy Physics Analysis Agents}},
author = {Zijie Shang and Ke Li and Junkun Jiao and Shengchun Wang and Changzheng Yuan and Xiongfei Wang and Beijiang Liu and Weimin Song and Xiaoshen Kang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.12862},
eprint = {2610.12862},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.12862},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - \textbf{\textsc{HepUKE}}: Uncertainty-aware Knowledge Evolution for High Energy Physics Analysis Agents
AU - Zijie Shang
AU - Ke Li
AU - Junkun Jiao
AU - Shengchun Wang
AU - Changzheng Yuan
AU - Xiongfei Wang
AU - Beijiang Liu
AU - Weimin Song
AU - Xiaoshen Kang
PY - 2026
DA - 2026/10/09
PB - arXiv
DO - 10.48550/arxiv.2610.12862
UR - https://arxiv.org/abs/2610.12862
AB - Knowledge bases are essential to agents for high energy physics (HEP), supplying few-shot evidence for previously unseen tasks. Their principal utility lies in strategic guidance for novel analyses rather than ready-made answers or reusable code. Prevailing systems, however, treat all retrieved evidence as equally trustworthy and fail to accrue knowledge from the analyses under assistance. In this paper, inspired by the standard HEP treatment of uncertainty, we propose \textbf{\textsc{\underline{HepUKE}}}: an \textbf{\underline{U}}ncertainty-aware \textbf{\underline{K}}nowledge \textbf{\underline{E}}volution system for \textbf{\underline{HEP}} that repurposes information entropy to unify statistical uncertainty from LLM generation and systematic uncertainty from HEP methodological choices. Upon it, we construct a self-evolving domain-specific-language (DSL) knowledge base and release a large-scale expert-annotated HEP physics-analysis benchmark. Per-evidence confidence is quantified via the information entropy of sampled DSL generations: high-confidence precedents constrain program synthesis, whereas analyses lacking close precedents are validated and written back, sustaining continual knowledge evolution. To handle the highly redundant yet logically stringent nature of HEP knowledge, we introduce an early-stopping controller that fuses a backward stationarity test on the answer posterior with a forward marginal information gain estimate. On the 1,000-pair BESIII PhysicsQA benchmark, HepUKE reduces retrieval rounds by 43.0\% while preserving most answer quality, and on the out-of-domain HotpotQA benchmark, it reduces retrieval rounds by 14.5\% with a similar quality--efficiency trade-off. The code and datasets used in this work are available at https://anonymous.4open.science/r/HepUKE-32FF.
ER -