Research map / Probabilistic and scientific modeling / Dynamical systems
From geometry to dynamics: Learning overdamped Langevin dynamics from sparse observations with geometric constraints
Dimitra Maoutsa
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 |
|---|---|---|---|
| add geometry-guided path augmentation (geodesic-constrained bridges) over naive GP estimation | Geometric framework · Hopf (tau=200 x dt) | wRMSE: 0.781 -> 0.210 | Improved |
| geometry-guided augmentation vs GP on Selkov | Geometric framework · Selkov (tau=100 x dt) | wRMSE: 0.550 -> 0.414 | Improved |
| geometry-guided augmentation vs GP on Van der Pol with large interval | Geometric framework · Van der Pol (sigma=0.50, tau=280 x dt) | wRMSE: 1.542 -> 0.865 | Improved |
| geometry-guided augmentation on harmonic trapping system with obstacle | Geometric framework · Harmonic trapping (tau=150 x dt) | wRMSE: 2.348 -> 2.762 | Hurt |
| geometry-guided augmentation vs GP on harmonic trapping | Geometric framework · Harmonic trapping (tau=250 x dt) | wRMSE: 3.733 -> 2.693 | Improved |
| increase noise amplitude sigma (ablation) | Geometric framework · Van der Pol | improvement over naive GP | Hurt |
| two E.M. augmentation iterations instead of naive GP estimate (ablation) | Geometric framework · Van der Pol 5D | wRMSE | Improved |
| replace the Gaussian-RBF-based diagonal metric (Geometric_RBF) with the default metric learning (ablation) | Geometric framework · Van der Pol | wRMSE | No clear effect |
| misestimate augmentation noise amplitude by up to 0.1 (ablation) | Geometric framework · Van der Pol | wRMSE: at most +-0.1 deviation | No clear effect |
| increase inter-observation interval tau | Geometric framework vs Ornstein-Uhlenbeck augmentation · Van der Pol | wRMSE | Improved |
Closest papers by meaning
- Langevin-Informed Transfer Learning: Replacing the Target Samples by Black-Box Feedback (NeurIPS 2026 · similarity 0.89)
- Partially Observed Trajectory Inference using Optimal Transport and a Dynamics Prior (ICLR 2025 · similarity 0.88)
- Parametric model reduction of mean-field and stochastic systems via higher-order action matching (NeurIPS 2024 · similarity 0.87)
- Exact Inference for Continuous-Time Gaussian Process Dynamics (AAAI 2024 · similarity 0.87)
- Markovian Dynamics Enforcer: Feasibility Preserving Correction on Learned Dyanmics Manifolds (NeurIPS 2026 · similarity 0.86)
- Predicting partially observable dynamical systems via diffusion models with a multiscale inference scheme (NeurIPS 2025 · 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.
