arXiv：cs.LG（机器学习，全量分类）· Yufeng Wang, Parivesh Priye, Lu Wei, Haibin Ling·· 14 小时前AI 评分46
从施加结构与学习物理构建稳定且反事实鲁棒的物理世界模型
Stable and Counterfactually Robust Physical World Models from Imposed Structure and Learned Physics
AI 导读
研究者提出一种物理世界模型，将通用结构硬编码、系统专属物理从数据学习：动力学由学习能量经固定可逆算子生成，能量限定于约束类，单向端口只耗能。在电磁腔、粒子网格与浅水流体中，约九千参数的模型以单套权重区分耗散与非耗散世界达四个数量级，稳定外推至训练时长一百倍，并迁移符号、幅度、速率与重力变化；同等容量但无此结构的模型表现接近随机。
正文
Abstract:A world model learns to forecast how a physical system evolves from recorded trajectories, yet the systems it imitates obey physical laws that are neither fully supplied nor reliably respected. The model may create energy, drift or diverge over long rollouts, and answer a changed law query using the law observed during training. We ask how much general physical structure must be hard coded into a world model, and how much system-specific physics can then be learned from data, for four properties to hold simultaneously: second law compatible dissipation, correct responses to interventions on physical parameters, stability out to one hundred times the training horizon, and robustness to disturbances. The imposed structure is general: dynamics are generated from the gradient of a learned energy through a fixed reversible operator, the energy is restricted to a confining class, a one way port can remove energy but never inject it, the drive channel is known, and the intervened parameter enters through a separable map. The model learns the energy functional, constitutive relations, dissipation rate, and couplings. Across an electromagnetic cavity, a particle in cell grid, and a shallow-water fluid, models with roughly nine thousand parameters recover constitutive functions with unit slope, separate conserving from dissipating worlds by four orders of magnitude using a single set of weights, and transfer changes in sign, magnitude, rate, and gravity to unseen values, where equal-capacity models without the same structure perform at chance or worse. A nonlinear constitutive law is recovered with its curvature preserved and predicts a held-out intervention $2$-$17\times$ better than a converged linear model.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.00280 [cs.LG] |
| (or arXiv:2610.00280v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.00280 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yufeng Wang [view email]
[v1]
Thu, 24 Sep 2026 19:56:30 UTC (2,317 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org