Инфографический слайд по физическому моделированию

Academic conference overview slide explaining [GeoPT: Scaling Physics Simulation via Lifted Geometric Pre-Training], styled as a clean scientific infographic on a widescreen 16:9 white background with navy-blue typography and accents. Header: - Bold dark-navy title: "[GeoPT: Scaling Physics Simulation via Lifted Geometric Pre-Training]" - Subtitle directly underneath in smaller blue font: "[An intuitive overview of Wu et al. (2026)]" Big idea banner: - Rounded rectangular callout card with a light blue border and pale blue fill, containing a blue circular lightbulb icon on the left. - Text: "Big idea: use abundant 3D geometry and synthetic motion to pre-train neural simulators before using scarce real simulation data." Main pipeline (3 columns x 4 rows connected left-to-right by thick blue arrows): - Column headers: "3D Geometry", "Physics Simulation (examples)", "Fields (pressure / velocity / stress)" - Row 1 (Car): Gray 3D CAD car model → car in aerodynamic airflow with blue streamlines, colorful vortices, and a small "t →" time indicator → car mesh colored with a rainbow heatmap field and a vertical color bar legend labeled "Pressure" (High / Low). - Row 2 (Aircraft): Gray 3D airplane model → aircraft surrounded by turbulent flow streamlines and wake vortices with a "t →" indicator → rainbow heatmap aircraft with a vertical color bar legend labeled "Velocity" (High / Low). - Row 3 (Ship hull): Gray 3D ship hull model → ship hull slicing through simulated ocean water waves and splash dynamics with a "t →" indicator → rainbow heatmap ship hull with a vertical color bar legend labeled "Pressure" (High / Low). - Row 4 (Crash / Deformation): Deformed gray vehicle frame → vehicle crash dynamic impact simulation showing structural buckling, debris, and internal stress visualization with a "t →" indicator → rainbow heatmap deformed frame with a vertical color bar legend labeled "Stress" (High / Low). Bottom explanation strip (3 circular icon-and-caption cards connected by dotted blue arrows): 1. Wireframe cube icon, title "Abundant geometry", caption "Easily available 3D shapes from CAD repositories" 2. Flow lines icon, title "Synthetic dynamics", caption "Scalable simulations produce diverse motion and fields" 3. Neural network graph icon, title "Better physics simulators", caption "Pre-trained models that generalize better with limited real data" Footer: - Centered small gray citation text: "[Source paper: Haixu Wu et al., arXiv:2602.20399 (2026)]" - Thin navy accent line spanning the bottom with circular endpoint dots.