Research map / Architectures and efficiency / RNNs and normalization
Learning to Think in Physics: Breaking Shortcut Learning in Scientific Diffusion via Representation Alignment
Haozhe Jia, Pengyu Yin, Wenshuo Chen, Shaofeng Liang, Lei Wang, Bowen Tian, Xiucheng Wang, Jia Nanqian, Yutao Yue
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 intermediate-layer PDE residual alignment (REPA-P) | PIDM U-Net · Darcy flow generation | data loss: 0.0180 -> 0.0119 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM U-Net · Charge potential generation | physics residual: 0.381 -> 0.128 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM U-Net · Topology optimization ID | CE%: 9.24 -> 4.17 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM U-Net · Topology optimization OOD | CE%: 7.93 -> 5.05 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM U-Net · Turbulent channel flow | PSNR: 37.64 -> 39.95 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM DiT · Darcy flow generation | data loss: 0.0831 -> 0.0352 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM DiT · Turbulent channel flow | PSNR: 38.16 -> 39.65 | Improved |
| add intermediate-layer PDE residual alignment (REPA-P) | PIDM DiT · Topology optimization ID | physics residual: 1.38e-3 -> 1.31e-3 | Improved |
| align at bottleneck instead of output only (ablation) | REPA-P U-Net · Darcy flow generation | data loss: 0.0162 -> 0.0119 | Improved |
| align at bottleneck instead of output only (ablation) | REPA-P U-Net · Charge potential generation | physics residual: 0.188 -> 0.128 | Improved |
| align at output layer only vs bottleneck (ablation) | REPA-P U-Net · Darcy flow reconstruction | physics residual: 0.0173 (bottleneck) vs 0.0142 (output) | Mixed |
| add decoder-position alignment (ablation) | REPA-P U-Net · Topology optimization ID | physics residual: 5.2e-3 -> 6.7e-3 | Hurt |
Closest papers by meaning
- Ultra Fast PDE Solving via Physics Guided Few-step Diffusion (NeurIPS 2026 · similarity 0.89)
- Text2PDE: Latent Diffusion Models for Accessible Physics Simulation (ICLR 2025 · similarity 0.88)
- Physics-Informed Diffusion Models (ICLR 2025 · similarity 0.88)
- SBNO : Schrödinger Bridge Neural Operator for the Forward and Inverse Problems under Missing Information (NeurIPS 2026 · similarity 0.87)
- PDE-Transformer: Efficient and Versatile Transformers for Physics Simulations (ICML 2025 · similarity 0.87)
- DiffusionPDE: Generative PDE-Solving under Partial Observation (NeurIPS 2024 · similarity 0.87)
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.
