Tag
PGD-NO is a neural operator that precomputes geometry decomposition to achieve linear memory scalability, enabling high-fidelity physics simulations on meshes exceeding 10 million nodes and overcoming the single-node memory bottleneck.