Research map / Probabilistic and scientific modeling / Differential equations and PDEs
SVD-NO: Learning PDE Solution Operators with SVD Integral Kernels
Noam Koren, Ralf J. J. Mackenbach, Ruud J. G. van Sloun, Kira Radinsky, Daniel Freedman
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 low-rank SVD kernel with a dense MLP kernel (removal) (ablation) | SVD-NO · five PDE benchmarks | mean L2 error: error rises by 3.02x without the SVD factorization | Improved |
| add low-rank SVD kernel instead of dense MLP kernel: training time per epoch (ablation) | SVD-NO · Diffusion Sorption | training time per epoch: 176.19s -> 2.32s | Improved |
| add low-rank SVD kernel instead of dense MLP kernel: training time per epoch (ablation) | SVD-NO · Shallow Water | training time per epoch: 374.75s -> 22.36s | Improved |
| use SVD instead of Mercer expansion (ablation) | SVD-NO · five PDE benchmarks | mean L2 error: Mercer raises mean error by 3.32x | Improved |
| add Gram-matrix orthogonality penalty (ablation) | SVD-NO · five PDE benchmarks | mean L2 error: without it error rises by 2.97x | Improved |
| add orthogonality penalty (ablation) | SVD-NO · Allen Cahn | mean L2 error (%): 0.84 -> 0.06 | Improved |
| add orthogonality penalty (ablation) | SVD-NO · Diffusion Sorption | mean L2 error (%): 0.11 -> 0.10 | No clear effect |
| increase SVD rank L from 3 to 10 (ablation) | SVD-NO · multiple PDE benchmarks | normalized test loss: L=3 to L=10 substantially improves accuracy; memory grows linearly | Improved |
| use SVD-NO instead of best baseline | SVD-NO · Allen Cahn | mean L2 error (%): 0.08 -> 0.06 (25.0% improvement) | Improved |
| use SVD-NO instead of best baseline | SVD-NO · Darcy Flow | mean L2 error (%): 2.02 -> 2.55 | Hurt |
| use SVD-NO instead of best baseline | SVD-NO · Shallow Water | mean L2 error (%): 0.46 -> 0.37 (17.8% improvement) | Improved |
Closest papers by meaning
- Toward Efficient Kernel-Based Solvers for Nonlinear PDEs (ICML 2025 · similarity 0.89)
- Improved Operator Learning by Orthogonal Attention (ICML 2024 · similarity 0.89)
- Component Fourier Neural Operator for Singularly Perturbed Differential Equations (AAAI 2024 · similarity 0.88)
- Neural Spectral Methods: Self-supervised learning in the spectral domain (ICLR 2024 · similarity 0.87)
- Derivative-enhanced Deep Operator Network (NeurIPS 2024 · similarity 0.87)
- Neural Operators with Localized Integral and Differential Kernels (ICML 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.
