GeoPT: Scaling Physics Simulation via Lifted Geometric Pre-Training
This provides a scalable solution for industrial physics simulation in domains like automotive and aerospace, though it is an incremental improvement over existing pre-training approaches.
The paper tackles the problem of scaling neural physics simulators by addressing the bottleneck of high-fidelity training data generation, proposing GeoPT which uses synthetic dynamics to pre-train on geometry, resulting in 20-60% reduction in labeled data requirements and 2x faster convergence across industrial benchmarks.
Neural simulators promise efficient surrogates for physics simulation, but scaling them is bottlenecked by the prohibitive cost of generating high-fidelity training data. Pre-training on abundant off-the-shelf geometries offers a natural alternative, yet faces a fundamental gap: supervision on static geometry alone ignores dynamics and can lead to negative transfer on physics tasks. We present GeoPT, a unified pre-trained model for general physics simulation based on lifted geometric pre-training. The core idea is to augment geometry with synthetic dynamics, enabling dynamics-aware self-supervision without physics labels. Pre-trained on over one million samples, GeoPT consistently improves industrial-fidelity benchmarks spanning fluid mechanics for cars, aircraft, and ships, and solid mechanics in crash simulation, reducing labeled data requirements by 20-60% and accelerating convergence by 2$\times$. These results show that lifting with synthetic dynamics bridges the geometry-physics gap, unlocking a scalable path for neural simulation and potentially beyond. Code is available at https://github.com/Physics-Scaling/GeoPT.