From Cheap Geometry to Expensive Physics: Elevating Neural Operators via Latent Shape Pretraining

Open in new window