Research map / Probabilistic and scientific modeling / Dynamical systems
STFlow: Data-Coupled Flow Matching for Geometric Trajectory Simulation
Kiet Bennema ten Brinke, Koen Minartz, Vlado Menkovski
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 the data-dependent informed random-walk prior instead of an N(0,I) prior (ablation) | STFlow (UNet) · N-body Gravity | FDE: 0.382 -> 0.256 | Improved |
| add the data-dependent informed random-walk prior instead of an N(0,I) prior (ablation) | STFlow (Transformer) · N-body Gravity | FDE: 0.307 -> 0.139 | Improved |
| use fitted random-walk prior instead of random walk with N(0,I) velocities (ablation) | STFlow (UNet) · N-body Gravity | FDE: 0.359 -> 0.256 | Improved |
| use fitted random-walk prior instead of random walk with N(0,I) velocities (ablation) | STFlow (Transformer) · N-body Gravity | FDE: 0.288 -> 0.139 | Improved |
| set velocity variance scale s=4 instead of s=1 (ablation) | STFlow (UNet) · N-body Gravity | FDE: 0.252 -> 0.256 | No clear effect |
| set velocity variance scale s=4 instead of s=1 (ablation) | STFlow (Transformer) · N-body Gravity | FDE: 0.231 -> 0.139 | Improved |
| add the Spatial layer (ablation) | STFlow (UNet) · N-body Gravity | FDE: 1.163 -> 0.256 | Improved |
| add the Temporal layer (ablation) | STFlow (UNet) · N-body Gravity | FDE: 1.651 -> 0.256 | Improved |
| add edge updating after each layer (ablation) | STFlow (UNet) · N-body Gravity | FDE: 0.272 -> 0.256 | Improved |
| add edge updating after each layer (ablation) | STFlow (Transformer) · N-body Gravity | FDE: 0.201 -> 0.139 | Improved |
| replace UNet with Transformer as temporal layer | STFlow · N-body Gravity | FDE: 0.256 -> 0.139 | Improved |
| replace UNet with Transformer as temporal layer | STFlow · N-body Spring | FDE: 0.0008 -> 0.0013 | Hurt |
Later papers that compared with this method
2 later accepted papers report STFlow as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- MoLF: Mixture-of-Latent-Flow for Pan-Cancer Spatial Gene Expression Prediction from Histology (ICML 2026 · better on 3 of 3 shared results)
- HistoPrism: Unlocking Functional Pathway Analysis from Pan-Cancer Histology via Gene Expression Prediction (ICLR 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- MoFlow — MoFlow: One-Step Flow Matching for Human Trajectory Forecasting via Implicit Maximum Likelihood Estimation based Distillation (CVPR 2025) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Geometric Trajectory Diffusion Models (NeurIPS 2024 · similarity 0.87)
- Metric Flow Matching for Smooth Interpolations on the Data Manifold (NeurIPS 2024 · similarity 0.86)
- TracingFlow: A Simulation-Free Trajectory Inference Framework Based on Second-Order Dynamics (NeurIPS 2026 · similarity 0.86)
- First-Order Trajectory Matching: Fast Ensemble Predictions of Chaotic, Turbulent, Stochastic Systems (NeurIPS 2026 · similarity 0.85)
- Diffusion Generative Flow Samplers: Improving learning signals through partial trajectory optimization (ICLR 2024 · similarity 0.85)
- PRECISE: SDE-Consistent Stochastic Sampling for RL Post-Training of Flow-Matching Models (NeurIPS 2026 · 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.
