Research map / Architectures and efficiency / RNNs and normalization
Transform Once: Efficient Operator Learning in Frequency Domain
Michael Poli, Stefano Massaroli, Federico Berto, Jinkyoo Park, Tri Dao, Christopher Re, Stefano Ermon
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 |
|---|---|---|---|
| replace FNO with T1+vp | T1 · Navier-Stokes high viscosity | N-MSE: 0.379 -> 0.257 | Improved |
| replace FFNO with T1+vp | T1 · Navier-Stokes high viscosity | N-MSE: 0.997 -> 0.257 | Improved |
| replace FNO with T1+vp | T1 · Navier-Stokes | step time (ms): 31 -> 19 | Improved |
| add variance preserving init to FNO | FNO · Navier-Stokes high viscosity | N-MSE: 0.379 -> 0.351 | Improved |
| add variance preserving init to FNO | FNO · Navier-Stokes low viscosity | N-MSE: 0.328 -> 0.315 | Improved |
| add vp init to T1 (ablation) | T1 · Navier-Stokes high viscosity | N-MSE: 0.491 -> 0.304 | Improved |
| replace k-space model with UNet (ablation) | T1 · Navier-Stokes high viscosity | N-MSE: 0.491 -> 0.295 | Improved |
| combine UNet k-space model with vp init (ablation) | T1 · Navier-Stokes low viscosity | N-MSE: 0.449 -> 0.240 | Improved |
| increase training dataset size | T1 vs FNO · Navier-Stokes nu=1e-4 | N-MSE | Mixed |
| use T1+vp instead of FNO | T1 · Flow around airfoils | training time (hrs): 6.0 -> 1.3 | Improved |
| use T1+vp instead of FNO | T1 · Flow around airfoils | N-MSE: 0.020 -> 0.024 | No clear effect |
| use T1+vp instead of FNO | T1 · ScalarFlow | training time (hrs): 32.4 -> 4.7 | Improved |
Closest papers by meaning
- Beyond Regular Grids: Fourier-Based Neural Operators on Arbitrary Domains (ICML 2024 · similarity 0.86)
- Spectral-Refiner: Accurate Fine-Tuning of Spatiotemporal Fourier Neural Operator for Turbulent Flows (ICLR 2025 · similarity 0.86)
- STNet: Spectral Transformation Network for Solving Operator Eigenvalue Problem (NeurIPS 2025 · similarity 0.86)
- Hierarchical Neural Operator Transformer with Learnable Frequency-aware Loss Prior for Arbitrary-scale Super-resolution (ICML 2024 · similarity 0.86)
- Rethinking Fourier Transform from A Basis Functions Perspective for Long-term Time Series Forecasting (NeurIPS 2024 · similarity 0.86)
- Frequency Domain Reservoir Computing (NeurIPS 2026 · 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.
