Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Towards Generalizable PDE Dynamics Forecasting via Physics-Guided Invariant Learning
Siyang Li, Yize Chen, Yan Guo, Ming Huang, Hui Xiong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add frequency enrichment loss (ablation) | iMOOE · DR OOD | nMSE: 5.64e-2 -> 4.23e-2 | Improved |
| add frequency enrichment loss (ablation) | iMOOE · SST and SSE real-world ocean data | nMSE: SST 5.64e-1 -> 5.12e-1 | Improved |
| add pre-computed derivative selection with masks (ablation) | iMOOE · SST | nMSE: 6.07e-1 -> 5.12e-1 | Improved |
| add mask diversity loss (ablation) | iMOOE · DR OOD | nMSE: 4.78e-2 -> 4.23e-2 | Improved |
| provide second-order derivatives in addition to first-order (ablation) | iMOOE · BG OOD | nMSE: 1.16e-2 -> 1.08e-2 | Improved |
| increase operator experts from 1 to 3 (ablation) | iMOOE · DR OOD | nMSE: 5.26e-2 -> 4.17e-2 | Improved |
| increase operator experts from 3 to 4 (ablation) | iMOOE · DR OOD | nMSE: 4.17e-2 -> 5.00e-2 | Hurt |
| wrap FNO in mixture of operator experts (MOOE) (ablation) | FNO · DR OOD mean over 8 envs | nMSE: 7.94e-2 -> 5.16e-2 | Improved |
| add invariant training objective on top of MOOE (ablation) | FNO · DR OOD mean over 8 envs | nMSE: 5.16e-2 -> 4.23e-2 | Improved |
| wrap DeepONet in MOOE only (ablation) | DeepONet · DR OOD mean over 8 envs | nMSE: 6.15e-1 -> 6.10e-1 | No clear effect |
| apply iMOOE to VCNeF (ablation) | VCNeF · DR OOD mean over 8 envs | nMSE: 7.84e-2 -> 5.52e-2 | Improved |
| iMOOE vs best baseline on Heat-Conduction ID | iMOOE · HC ID | nMSE: 3.68e-2 (DPOT) vs 3.92e-2 | Hurt |
Closest papers by meaning
- Stable Long-Horizon PDE Forecasting via Latent Structured Spectral Propagators (NeurIPS 2026 · similarity 0.89)
- Space-Time Continuous PDE Forecasting using Equivariant Neural Fields (NeurIPS 2024 · similarity 0.87)
- P$^2$C$^2$Net: PDE-Preserved Coarse Correction Network for efficient prediction of spatiotemporal dynamics (NeurIPS 2024 · similarity 0.87)
- Boosting Generalization in Parametric PDE Neural Solvers through Adaptive Conditioning (NeurIPS 2024 · similarity 0.87)
- Physics-informed Temporal Alignment for Auto-regressive PDE Foundation Models (ICML 2025 · similarity 0.87)
- Mechanistic PDE Networks for Discovery of Governing Equations (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
