Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Particle-Guided Diffusion Models for Partial Differential Equations
Andrew Millard, Fredrik Lindsten, Zheng Zhao
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 pBS weighting instead of TDS weighting | GEM proposal · Darcy forward | L2 relative error (%): 5.52 -> 4.49 | Improved |
| use pBS weighting instead of TDS weighting | GEM proposal · Poisson inverse | L2 relative error (%): 35.27 -> 22.01 | Improved |
| use pBS weighting instead of TDS weighting | GEM proposal · Gray-Scott RD, noise 0, D fields | error (%): 4.46 -> 1.45 | Improved |
| add SMC pBS with 4 particles to SOSaG | SOSaG · Darcy inverse | error rate (%): 8.00 -> 5.28 | Improved |
| add SMC pBS with 4 particles to SOSaG | SOSaG · Helmholtz forward | L2 relative error (%): 11.71 -> 12.99 | Hurt |
| add SMC pBS with 4 particles to GEM | GEM · Helmholtz forward | L2 relative error (%): 11.22 -> 9.01 | Improved |
| use stochastic second-order SOSaG instead of ODE DiffPDE (single sample) | DiffPDE · Darcy forward | L2 relative error (%): 5.58 -> 4.47 | Improved |
| use GEM instead of ODE DiffPDE (single sample) | DiffPDE · Poisson inverse | L2 relative error (%): 19.88 -> 22.28 | Hurt |
| use SOSaG-pBS instead of GEM-pBS under high observation noise | GEM-pBS · Gray-Scott RD, noise 0.02, u fields | error (%): 16.19 -> 8.55 | Improved |
| use SOSaG-pBS instead of DiffPDE on 3-species RD u fields | DiffPDE · 3-Species RD, noise 0.02, u fields | error (%): 2.24 -> 2.45 | Hurt |
| increase tempering rho from 5 to 50 (ablation) | GEM-pBS · Darcy forward | L2 relative error (%): 6.19% -> 3.58% | Improved |
| increase tempering rho from 5 to 1000 (ablation) | SOSaG-pBS · Poisson inverse | L2 relative error (%): 24.13% -> 20.17% | Improved |
Closest papers by meaning
- Ultra Fast PDE Solving via Physics Guided Few-step Diffusion (NeurIPS 2026 · similarity 0.91)
- DiffusionPDE: Generative PDE-Solving under Partial Observation (NeurIPS 2024 · similarity 0.91)
- Text2PDE: Latent Diffusion Models for Accessible Physics Simulation (ICLR 2025 · similarity 0.88)
- Adaptive Stochastic Coefficients for Accelerating Diffusion Sampling (NeurIPS 2025 · similarity 0.88)
- Tortoise and Hare Guidance: Accelerating Diffusion Model Inference with Multirate Integration (NeurIPS 2025 · similarity 0.88)
- Physics-Informed Diffusion Models (ICLR 2025 · similarity 0.88)
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.
