Research map / Graph learning / GNN architectures
Transformer with Implicit Edges for Particle-Based Physics Simulation
Yidi Shao, Chen Change Loy, Bo Dai
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 implicit edges (receiver/sender tokens) to vanilla Transformer (ablation) | Transformer · FluidShake | M3SE (1e-2): 2.75 -> 1.52 | Improved |
| add implicit edges to vanilla Transformer (ablation) | Transformer · L-FluidShake | M3SE (1e-2): 8.18 -> 3.17 | Improved |
| add normalization effects in attention (ablation) | TIE (config B) · FluidShake | M3SE (1e-2): 1.52 -> 1.22 | Improved |
| add normalization effects in attention (ablation) | TIE (config B) · L-FluidShake | M3SE (1e-2): 3.17 -> 2.40 | Improved |
| add trainable abstract particles (ablation) | TIE · L-FluidShake | M3SE (1e-2): 2.40 -> 2.16 | Improved |
| add trainable abstract particles (ablation) | TIE · FluidShake | M3SE (1e-2): 1.22 -> 1.30 | Hurt |
| add trainable abstract particles (ablation) | TIE · BoxBath (960,64,cube) | M3SE (1e-2): 1.35 -> 0.92 | Improved |
| add trainable abstract particles (ablation) | TIE · BoxBath (960,41,bunny) | M3SE (1e-2): 1.50 -> 1.39 | Improved |
| add trainable abstract particles | TIE · RiceGrip | M3SE (1e-2): 0.13 -> 0.08 | Improved |
| replace abstract particles with dummy zero particles (ablation) | TIE+ · BoxBath (960,64,cube) | M3SE (1e-2): 1.21 vs 0.92 | Mixed |
| replace abstract particles with dummy zero particles (ablation) | TIE+ · BoxBath (960,41,bunny) | M3SE (1e-2): 1.96 vs 1.39 | Hurt |
| decrease radius R from 0.08 to 0.07 (ablation) | TIE · FluidShake | M3SE (1e-2): 1.22 -> 1.60 | Hurt |
Closest papers by meaning
- PDE-Transformer: Efficient and Versatile Transformers for Physics Simulations (ICML 2025 · similarity 0.86)
- DEL: Discrete Element Learner for Learning 3D Particle Dynamics with Neural Rendering (NeurIPS 2024 · similarity 0.85)
- MaNGO — Adaptable Graph Network Simulators via Meta-Learning (NeurIPS 2025 · similarity 0.85)
- Neural SPH: Improved Neural Modeling of Lagrangian Fluid Dynamics (ICML 2024 · similarity 0.85)
- Graph Mamba Operator: A Latent Simulator for Interacting Particle Systems (NeurIPS 2026 · similarity 0.84)
- Universal Physics Transformers: A Framework For Efficiently Scaling Neural Operators (NeurIPS 2024 · 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.
