发表机构
iTHEMS; RIKEN（iTHEMS; 理化学研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出PR-Smoother，一种保模拟器摊销平滑器，通过仅学习未来条件修正保持指定模拟器显式，实现非高斯后验平滑，并在Lorenz-96及Kolmogorov流中验证多模态捕获与高维扩展能力。
AI 中文摘要
许多物理数据同化（DA）工作流程需要能够表示物理状态变量上的非高斯后验、可扩展到高维模拟器、仅从观测窗口训练、并保持与指定模拟器校准兼容的平滑方法。我们提出了PR-Smoother，一种为这种指定模拟器DA机制设计的保模拟器摊销平滑器。其关键设计原则是在证据下界和变分族中均保持指定模拟器显式存在：PR-Smoother不学习替代动力学或学习轨迹先验，而是仅围绕指定滚动预测学习未来条件修正。这产生了物理轨迹上的显式非高斯平滑分布，并支持仅从观测中联合学习状态、参数和传感器偏差。变分族在确定性和线性高斯极限中包含精确平滑器。实验上，PR-Smoother在4维Lorenz-96中捕获多模态后验，在40维Lorenz-96中在模糊非线性观测和过程噪声下保持准确，并扩展到16,384维Kolmogorov流中的联合状态-参数-偏差推断。
英文摘要
Many physical data assimilation (DA) workflows require smoothing methods that represent non-Gaussian posteriors over physical state variables, scale to high-dimensional simulators, train from observation windows alone, and remain compatible with calibration of the prescribed simulator. We introduce PR-Smoother, a simulator-preserving amortized smoother designed for this prescribed-simulator DA regime. Its key design principle is to keep the prescribed simulator explicit in both the evidence lower bound and the variational family: rather than learning replacement dynamics or a learned trajectory prior, PR-Smoother learns only future-conditioned corrections around the prescribed rollout. This yields an explicit non-Gaussian smoothing distribution over physical trajectories and supports joint state, parameter, and sensor-bias learning from observations alone. The variational family contains the exact smoother in deterministic and linear-Gaussian limits. Empirically, PR-Smoother captures multimodal posteriors in 4-dimensional Lorenz-96, remains accurate under ambiguous nonlinear observations and process noise in 40-dimensional Lorenz-96, and scales to joint state-parameter-bias inference in 16,384-dimensional Kolmogorov flow.
CommentsAccepted at Neurips 2026
