发表机构
Stanford University; European Organization for Nuclear Research (CERN); SLAC National Accelerator Laboratory（斯坦福大学; 欧洲核子研究组织（CERN）; SLAC国家加速器实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出跨探测器的点云自蒸馏框架，构建Panda V2基础模型，仅用千级标注数据即实现粒子聚类、识别等任务的最优性能，大幅降低高能核物理标注成本。
AI 中文摘要
粒子与核物理领域正日益关注基础模型，但现有方法因依赖探测器特定架构或预训练目标，仍与单个实验深度绑定，限制了其在不同传感模态间的复用性。本文提出点云自蒸馏框架，可生成通用性更强的传感器级预训练方案。研究显示，同一优化后的架构与目标仅需极少改动，即可在三种性质迥异的探测器模态上独立预训练：液氩时间投影室（LArTPC）、对撞机TPC及水切伦科夫探测器。在下游任务适配时仅使用1000张标注图像，Panda V2即可达到或超越需多量级监督训练的专用基础模型基线；在sPHENIX上，其以70倍更少的标注事件达到最先进的粒子聚类性能，同时大幅提升粒子识别能力；在LArTPC数据上，其以多达1000倍更少的标签匹配Panda（arXiv:2512.01324）的粒子重建性能。除重建任务外，简单线性探针还揭示了与粒子因果性和径迹曲率相关的物理意义潜在结构。
英文摘要
Foundation models are increasingly being pursued in particle and nuclear physics, but existing approaches remain strongly tied to individual experiments through detector-specific architectures or pre-training objectives, limiting their reuse across sensing modalities. We show that a point cloud self-distillation framework yields a substantially more general sensor-level pre-training recipe. We show that the same refined architecture and objective can be independently pre-trained with minimal changes on three qualitatively different detector modalities: liquid argon time projection chamber (LArTPC), collider TPC, and water Cherenkov. Using 1,000 labeled images for downstream task adaptation, Panda V2 matches or exceeds specialized foundation-model baselines trained with orders of magnitude more supervision, matching state-of-the-art particle-clustering performance with 70x fewer labeled events on sPHENIX while substantially improving particle identification, and on LArTPC data matching Panda (arXiv:2512.01324) particle reconstruction with up to 1,000x fewer labels. Beyond reconstruction, simple linear probes reveal physically meaningful latent structure associated with particle causality and track curvature.
Comments24 pages, 11 figures, preprint
