Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Boundary constrained Gaussian processes for robust physics-informed machine learning of linear partial differential equations
David Dalton, Alan Lazarus, Hao Gao, Dirk Husmeier
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 boundary-constrained GP instead of BCNN | GP vs neural network with exact boundary constraints · Poisson-BVP-1 (Nf=4) | erru: 5.4 × 10−11 vs 2.0×10−2 | Improved |
| use BCGP instead of BCNN with a different random restart | BCNN · Poisson-BVP-1 (Nf=4) | erru: 3.2 × 10−3 vs 5.4 × 10−11 | Improved |
| use BCGP instead of BCNN on a pentagon domain with many collocation points | BCNN · Poisson-BVP-2 | erru: 7.2 × 10−5 vs 6.3 × 10−5 | Hurt |
| use BCGP instead of BCNN | BCNN · Poisson-BVP-3 | errθ: 2.2 × 10−2 vs 3.0 × 10−2 | Improved |
| use BCGP instead of BCNN | BCNN · Poisson-BVP-3 | erru: 8.0 × 10−3 vs 1.1 × 10−2 | Improved |
| use BCGP instead of BCNN | BCNN · Poisson-BVP-4 | erru: 1.5×10−3 vs 2.3×10−3 | Improved |
| use BCGP instead of BCNN | BCNN · Poisson-BVP-4 | errθ: 3.7×10−2 vs 1.1×10−2 | Hurt |
| increase collocation points for BCNN to about 50 (from 4) | BCNN · Poisson-BVP-1 | erru: Nf = 4 -> approximately Nf = 50 | Improved |
| add exact boundary constraints (BCGP) to the GP (ablation) | UCGP · Wave-IBVP-1 | errθ: 4.7 × 10−1 -> 1 × 10−2 | Improved |
| add exact boundary constraints (BCGP) to the GP (ablation) | UCGP · Wave-IBVP-1 | erru: 7.9 × 10−2 -> 8.3 × 10−3 | Improved |
| enforce boundary conditions exactly (BCGP) instead of with Nb penalty observations (PCGP) (ablation) | PCGP · Heat-IBVP-1 | errθ | Improved |
| enforce boundary conditions exactly (BCGP) instead of unconstrained GP (UCGP) (ablation) | UCGP · Heat-IBVP-2 | erru / errθ | Improved |
Closest papers by meaning
- Guiding continuous operator learning through Physics-based boundary constraints (ICLR 2023 · similarity 0.88)
- Imposing Boundary Conditions on Neural Operators via Learned Function Extensions (ICML 2026 · similarity 0.87)
- Solving High Frequency and Multi-Scale PDEs with Gaussian Processes (ICLR 2024 · similarity 0.86)
- Physics and Lie symmetry informed Gaussian processes (ICML 2024 · similarity 0.85)
- Solving Differential Equations with Constrained Learning (ICLR 2025 · similarity 0.85)
- Boundary Exploration for Bayesian Optimization With Unknown Physical Constraints (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.
