3D equivariant graph neural network designed for efficient molecular modeling and 3D graph analysis.
Unlike traditional models that rely on computationally expensive Clebsch-Gordan Products or irreducible representations, GotenNet uses geometric tensor representations and inner-product-based operations to reduce overhead.
The model maintains strict E(3) equivariance, meaning its predictions remain consistent regardless of how a 3D structure is rotated, translated, or reflected.