稳定自回归PDE基础模型滚动预测的事件触发上下文修复
Stabilizing Autoregressive PDE Foundation Model Rollouts with Event-Triggered Context Healing
查看机构详情
- University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
- National Center for Supercomputing Applications University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校超级计算应用国家中心）
- Indian Institute of Technology Delhi（德里印度理工学院）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
本文提出事件触发上下文修复（ETCH），通过比较预测与圆柱表面压力测量值，在误差超阈值时用扩散变换器重建上下文，显著降低DPOT和Poseidon-T在RealPDEBench上的预测误差，并减少重建成本。
中文摘要 AI 辅助
自回归PDE基础模型能够实现快速的全场预测，但随着预测状态被递归重用，误差会不断累积。我们引入了事件触发上下文修复（ETCH），该方法将预测结果与来自12个圆柱表面压力测点的测量值进行比较。当两者差异超过阈值时，基于压力条件的沙漏扩散变换器会重建速度-压力上下文，用于后续预测。在43条RealPDEBench圆柱轨迹上，使用相同的重建器和阈值，ETCH将DPOT的速度和压力误差从48.05%和63.71%分别降至6.29%和7.25%，将Poseidon-T的相应误差从74.97%和76.91%分别降至5.16%和6.84%。对于DPOT，仅修正14.56%的时间窗口即可保持与重建每个窗口相近的误差，并在A100 GPU上支持平均实时预测和修正。这些结果表明，稀疏的物理反馈能够稳定跨PDE基础模型的长自回归滚动预测，同时降低重建成本。
英文摘要
Autoregressive PDE foundation models enable fast full-field forecasting but accumulate errors as predicted states are recursively reused. We introduce Event-Triggered Context Healing (ETCH), which compares forecasts with measurements from 12 cylinder-surface pressure taps. When their discrepancy exceeds a threshold, pressure-conditioned hourglass diffusion transformers reconstruct the velocity--pressure context for subsequent predictions. On 43 RealPDEBench Cylinder trajectories, ETCH reduces velocity and pressure errors from 48.05 and 63.71% to 6.29 and 7.25% for DPOT, and from 74.97 and 76.91% to 5.16 and 6.84% for Poseidon-T, using the same reconstructors and threshold. For DPOT, correcting 14.56% of windows maintains errors close to reconstructing every window and supports average real-time forecasting and correction on an A100 GPU. These results show that sparse physical feedback stabilizes long autoregressive rollouts across PDE foundation models while reducing reconstruction cost.
