RotEqNet: Rotation-equivariant network for fluid systems with symmetric high-order tensors

In the recent application of scientific modeling, machine learning models are largely applied to facilitate computational simulations fluid systems. Rotation symmetry is a general property for most symmetric However, in general, current methods have no theoretical guarantee symmetry. By observing an important contraction and rotation operation on high order tensors, we prove that preserved via tensor contraction. Based this justification, paper, introduce Rotation-Equivariant Network (RotEqNet) rotation-equivariance tensors We implement RotEqNet evaluate our claims with four case studies various The error reduction verified these studies. Results showing superiority compared traditional methods.