Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Physics-informed Kernel Learning
Nathan Doumèche, Francis Bach, Gérard Biau, Claire Boyer
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 |
|---|---|---|---|
| use PIKL kernel estimator instead of vanilla PINNs | PDE solving · Wave equation (clean boundary) | L2 relative error: 4.52 -> 8.70 | Improved |
| use PIKL instead of NTK-optimized PINNs | PDE solving · Wave equation (clean boundary) | L2 relative error: 1.73 -> 8.70 | Improved |
| use PIKL instead of curriculum-trained PINNs | PDE solving · Convection equation (beta=40) | L2 relative error: 5.33 -> 7.31 | Improved |
| change Fourier frequency parameter L from 0.5 to 0.55 (ablation) | PIKL estimator · Wave equation (clean boundary) | L2 relative error: 8.70 -> 9.6 | No clear effect |
| use PIKL instead of traditional solvers (Euler, RK4) with clean boundaries | Wave equation solving · Wave equation (clean boundary) | L2 relative error: 3.8 -> 8.70 | Hurt |
| use PIKL instead of Crank-Nicolson with noisy boundary conditions | Wave equation solving · Wave equation (noisy boundary) | L2 relative error: 2.01 -> 1.87 | Improved |
| use PIKL instead of PINNs with noisy boundary conditions | Wave equation solving · Wave equation (noisy boundary) | L2 relative error: 4.61 -> 1.87 | Improved |
| add the PDE penalty (PIKL) to the Sobolev data-only estimator (ablation) | PIKL estimator · Heat equation, imperfect modeling (d=2) | L2 error | Improved |
| add data-driven term (PIKL) to a strongly PDE-penalized estimator (ablation) | PDE-only estimator · Heat equation, imperfect modeling (d=2) | L2 error | Improved |
| use PIKL instead of OLS tailored to PDE solution basis | Harmonic oscillator · Harmonic oscillator d=1 | L2 error: n^-1.1 | No clear effect |
Closest papers by meaning
- Learning a Neural Solver for Parametric PDEs to Enhance Physics-Informed Methods (ICLR 2025 · similarity 0.89)
- The Challenges of the Nonlinear Regime for Physics-Informed Neural Networks (NeurIPS 2024 · similarity 0.89)
- Understanding Generalization in Physics Informed Models through Affine Variety Dimensions (NeurIPS 2025 · similarity 0.88)
- An operator preconditioning perspective on training in physics-informed machine learning (ICLR 2024 · similarity 0.88)
- Neural Oscillators for Generalization of Physics-Informed Machine Learning (AAAI 2024 · similarity 0.87)
- Physics and Lie symmetry informed Gaussian processes (ICML 2024 · 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.
