Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Probabilistic Retrofitting of Learned Simulators
Cristiana Diaconu, Miles Cranmer, Richard E. Turner, Tanya Marwah, Payel Mukhopadhyay
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 |
|---|---|---|---|
| replace deterministic fine-tuning with CRPS retrofitting | Walrus/Lola/Poseidon · single-system RB, Euler, Shear Flow | rollout CRPS: 20-54% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | Walrus and Lola · Rayleigh-Benard | rollout VRMSE: up to 21.5% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | Walrus and Lola · Euler Multi-Quadrants | rollout VRMSE: up to 29.6% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | Walrus and Lola · Shear Flow | rollout VRMSE: up to 10.1% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | Poseidon · single-system datasets | rollout VRMSE | Mixed |
| replace deterministic fine-tuning with CRPS retrofitting | HalfWalrus · RB and TRL | rollout CRPS: 34-40% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | HalfWalrus · RB and TRL | rollout VRMSE: >13% reduction | Improved |
| replace deterministic fine-tuning with CRPS retrofitting | HalfWalrus · SAR and transition regimes (T=20) | rollout VRMSE | Hurt |
| increase number of ensemble members at inference (ablation) | CRPS-retrofitted Walrus · single-system datasets | rollout VRMSE | Improved |
| extend retrofitting duration (ablation) | CRPS retrofitting · single-system and foundation | rollout metrics | No clear effect |
| reduce training ensemble size (ablation) | CRPS retrofitting · single-system and foundation | rollout metrics | No clear effect |
| use higher noise embedding dimension and dense noise injection (ablation) | CRPS retrofitting · complex datasets | rollout metrics | Improved |
Closest papers by meaning
- Uncertainty Quantification for Forward and Inverse Problems of PDEs via Latent Global Evolution (AAAI 2024 · similarity 0.88)
- Calibrated Physics-Informed Uncertainty Quantification (ICML 2025 · similarity 0.87)
- On conditional diffusion models for PDE simulations (NeurIPS 2024 · similarity 0.87)
- Compute-optimal data scaling for neural surrogates via multi-fidelity training (NeurIPS 2026 · similarity 0.87)
- INC: An Indirect Neural Corrector for Auto-Regressive Hybrid PDE Solvers (NeurIPS 2025 · similarity 0.87)
- Active Learning for Neural PDE Solvers (ICLR 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.
