Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Spectral-Informed Neural Networks Outperform Spectral methods in High-dimensional PDEs
Tianchi Yu, Ivan Oseledets
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 coefficient decay scaler and basis embedding to SINN (ablation) | SINN · 2D Fourier, s=1.0 | relative L2 error: 6.54E-3 -> 5.55E-4 | Improved |
| Add coefficient decay scaler (with embedding) (ablation) | Modified SINN · 2D Fourier, s=1.0 | relative L2 error: 1.09E-3 -> 5.55E-4 | Improved |
| Add basis embedding (with decay) (ablation) | Modified SINN · 2D Fourier, s=0.6 (missing coefficients) | relative L2 error: 2.06E-1 -> 3.08E-3 | Improved |
| Add coefficient decay (with embedding) (ablation) | Modified SINN · 2D Fourier, s=0.6 | relative L2 error: 3.27E-3 -> 3.08E-3 | No clear effect |
| Use SINN instead of PINN | Poisson equation · d=100 | relative L2 error: 5.50E-1 -> 7.75E-3 | Improved |
| Use SINN instead of PINN | Heat equation · d=100 | relative L2 error: 7.39E-1 -> 3.69E-1 | No clear effect |
| Use SINN instead of DRM | Poisson equation · d=2 | relative L2 error: 2.48E-3 -> 1.61E-4 | Improved |
| Replace SGSM with SGSM+SINN to recover masked coefficients | SGSM · Schrödinger d=8, s=0.9 | relative L2 error: 3.05E-1 -> 1.23E-3 | Improved |
| Use SINN instead of SGSM when all coefficients available (s=1) | SGSM · Poisson/heat/convection | relative L2 error | Hurt |
| Use SINN instead of SGSM with missing coefficients (s<1) | SGSM · Poisson/heat/convection d<=8 | relative L2 error | Improved |
Closest papers by meaning
- Neuro-Spectral Architectures for Causal Physics-Informed Networks (NeurIPS 2025 · similarity 0.90)
- Neural Spectral Methods: Self-supervised learning in the spectral domain (ICLR 2024 · similarity 0.89)
- Efficient Error Certification for Physics-Informed Neural Networks (ICML 2024 · similarity 0.88)
- PINNacle: A Comprehensive Benchmark of Physics-Informed Neural Networks for Solving PDEs (NeurIPS 2024 · similarity 0.87)
- S$^{2}$-PINN: Stochastic Separable Physics-Informed Neural Networks (NeurIPS 2026 · similarity 0.87)
- Hyperbolic-PDE GNN: Spectral Graph Neural Networks in the Perspective of A System of Hyperbolic Partial Differential Equations (ICML 2025 · 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.
