AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

GeoPT: Scaling Physics Simulation via Lifted Geometric Pre-Training

Haixu Wu, Minghao Guo, Zongyi Li, Zhiyang Dou, Mingsheng Long, Kaiming He, Wojciech Matusik

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 1 citations (OpenAlex)

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 changedWhereMeasuredEffect
add dynamics-lifted GeoPT pre-training (vs training from scratch)Transolver · DTCHullrelative L2 / data requirement: 60% data requirement reductionImproved
add dynamics-lifted GeoPT pre-trainingTransolver · NASA-CRMrelative L2: moderate improvementImproved
add dynamics-lifted GeoPT pre-trainingTransolver · DrivAerMLdata saving / convergence speed: 40% data saving, 2x faster convergingImproved
add dynamics-lifted GeoPT pre-trainingTransolver · AirCraft, NASA-CRM, DTCHull, Car-Crashdata saving: 20-60% fewer physics samples; up to 2x faster convergenceImproved
pre-train with geometry-only supervision (vector distance) instead of training from scratch (ablation)Transolver · DrivAerMLrelative L2Hurt
condition on Hunyuan3D geometry representation (geometry-only conditioning) (ablation)Transolver · industrial benchmarksrelative L2No clear effect
pre-train backbone with GeoPT instead of using GeoPT features as conditioning (ablation)Transolver · industrial benchmarksrelative L2Improved
increase model depth from 8 to 32 layers with GeoPT pre-training (ablation)Transolver · limited-data industrial simulationrelative L2Improved
reduce the number of unique pre-training geometries (ablation)GeoPT · downstream physics tasksrelative L2Hurt
reduce sampled dynamic trajectories per geometry to 6% (ablation)GeoPT · DrivAerML (fixed inflow)relative L2: 6% comparable to 100 samplesNo clear effect
increase dynamic trajectories per geometry (ablation)GeoPT · AirCraft (varied speed, AoA, sideslip)relative L2Improved
apply GeoPT to radiosity simulationTransolver · Radiosity (Cornell box)MAE: 9.7×10−2 -> 9.0×10−2Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy