Physics-constrained neural networks for surrogate modeling of lossless periodic structures
For engineers designing lossless periodic structures, this provides a physically consistent surrogate that enables gradient-based inverse design without violating energy conservation.
The paper introduces a physics-constrained neural network (PCNN) that enforces energy conservation as a hard constraint for surrogate modeling of RCWA outputs on Stiefel manifolds, achieving rapid prediction of Jones matrices. The method is demonstrated through inverse design of a diffractive waveguide combiner for augmented reality glasses.
We introduce a physics-constrained neural network (PCNN) for the rapid prediction of rigorous coupled-wave analysis (RCWA) outputs in the form of Jones matrices. Starting from energy conservation in lossless layered periodic structures, we use the fact that RCWA outputs lie on a Stiefel manifold. This energy constraint is enforced as a hard condition by projecting onto the manifold using differentiable symmetric orthogonalization. The resulting surrogate enforces energy conservation by construction while preserving differentiability for gradient-based inverse design. The performance and generality of the proposed approach are demonstrated through the inverse design of a diffractive waveguide combiner for augmented reality glasses.