Research map / Probabilistic and scientific modeling / Differential equations and PDEs
OrthoSolver: A Neural Proper Orthogonal Decomposition Solver For PDEs
Ying Pang, Jingyuan Wang, Jiahao Ji, Fanhao Mu
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 mutual-information objective (ablation) | OrthoSolver · PDEBench 1D Navier-Stokes | relative L2 error: 0.0334 -> 0.0157 | Improved |
| add orthogonality regularization (ablation) | OrthoSolver · PDEBench 2D Navier-Stokes | relative L2 error: 0.0159 -> 0.0055 | Improved |
| add reconstruction constraint (ablation) | OrthoSolver · PDEBench 2D Navier-Stokes | relative L2 error: 0.0079 -> 0.0055 | Improved |
| add orthogonality regularization to reduce mode correlation (ablation) | OrthoSolver · all 7 datasets | average inter-mode correlation: 0.7832 -> 0.0631 | Improved |
| increase number of modes from 1 to 4 (ablation) | OrthoSolver · PDEBench 2D Navier-Stokes | relative L2 error: 0.0335 -> 0.0055 | Improved |
| increase number of modes from 4 to 6 (ablation) | OrthoSolver · PDEBench 2D Navier-Stokes | relative L2 error: 0.0055 -> 0.0170 | Hurt |
| use OrthoSolver instead of Transolver | neural PDE solver · PDEBench 2D Navier-Stokes | relative L2 error: 0.0282 -> 0.0055 | Improved |
| use OrthoSolver instead of best baseline | neural PDE solver · PDEBench 2D DiffReac | relative L2 error: 0.0189 -> 0.0172 | Mixed |
| increase number of modes from 4 to 6 (ablation) | OrthoSolver · PDEBench 2D DiffReac | relative L2 error: 0.0172 -> 0.0197 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Transolver — Transolver: A Fast Transformer Solver for PDEs on General Geometries (ICML 2024) · better on 7 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Unisolver: PDE-Conditional Transformers Towards Universal Neural PDE Solvers (ICML 2025 · similarity 0.86)
- OrthoReg: Orthogonal Regularization for Hybrid Symbolic-Neural Dynamical Systems (NeurIPS 2026 · similarity 0.86)
- Physics-informed Reduced Order Modeling of Time-dependent PDEs via Differentiable Solvers (NeurIPS 2025 · similarity 0.86)
- Toward Efficient Kernel-Based Solvers for Nonlinear PDEs (ICML 2025 · similarity 0.86)
- Vectorized Conditional Neural Fields: A Framework for Solving Time-dependent Parametric Partial Differential Equations (ICML 2024 · similarity 0.85)
- Improved Operator Learning by Orthogonal Attention (ICML 2024 · similarity 0.85)
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.
