Research map / Probabilistic and scientific modeling / Differential equations and PDEs
PICProp: Physics-Informed Confidence Propagation for Uncertainty Quantification
Qianli Shen, Wai Hoh Tang, Zhun Deng, Apostolos Psaros, Kenji Kawaguchi
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 meta-model EffiPICProp instead of plain regression on query-point CIs (SimPICProp, lambda=0) (ablation) | EffiPICProp · Pedagogical 1D PDE | CI generalization to unseen locations | Improved |
| SimPICProp (lambda=0) vs brute-force PICProp on query points (ablation) | SimPICProp · Pedagogical 1D PDE | CI match with PICProp | Mixed |
| EffiPICProp instead of brute-force PICProp per point | PICProp · Pedagogical 1D PDE | CI accuracy and compute | Improved |
| PICProp bi-level optimization vs exhaustive search (5,000 BCs) | PICProp · Pedagogical 1D PDE | CI agreement with ES reference | No clear effect |
| Increase ES trials from 1,000 to 5,000 | Exhaustive search · 2D Poisson equation | CI width: 1,000 trials -> 5,000 trials (intervals significantly wider) | Improved |
| EffiPICProp vs ES with 1,000 trials at equal compute | EffiPICProp · 2D Poisson equation | CI optimality (width) | Improved |
| EffiPICProp vs ES with 5,000 trials | EffiPICProp · 2D Poisson equation | CI width | No clear effect |
| EffiPICProp vs ES with 2,500 and 5,000 trials | EffiPICProp · Burgers equation | CI width (conservativeness) | Improved |
| Set lambda=0 (drop second loss term) in EffiPICProp (ablation) | EffiPICProp · Burgers equation | CI generalization | Improved |
| Use lambda>0 (second loss term over domain) in EffiPICProp (ablation) | EffiPICProp · Pedagogical 1D PDE | CI generalization | Improved |
Closest papers by meaning
- Calibrated Physics-Informed Uncertainty Quantification (ICML 2025 · similarity 0.87)
- Uncertainty Quantification for Physics-Informed Neural Networks with Extended Fiducial Inference (NeurIPS 2025 · similarity 0.86)
- Uncertainty Quantification for Forward and Inverse Problems of PDEs via Latent Global Evolution (AAAI 2024 · similarity 0.85)
- Uncertainty Quantification via Stable Distribution Propagation (ICLR 2024 · similarity 0.85)
- S$^{2}$-PINN: Stochastic Separable Physics-Informed Neural Networks (NeurIPS 2026 · similarity 0.84)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.84)
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.
