Tag
The paper introduces Riemannian Hodge Message Passing (RHMP), a neural architecture that separates topology and geometry for learning physical fields on meshes, achieving superior performance across multiple benchmarks.