Summary
How can fully coupled anisotropic layer problems become differentiable and GPU-efficient?
Builds scattering operators from boundary fields instead of internal eigenmodes.
Reports FEM/transfer-matrix agreement and 22.8× faster layer construction at 529 harmonics on the same GPU.
Related modelling approaches
Relation labels come from the method guide and are editorial. Unlabelled links are conceptual connections or comparison methods; the paper does not necessarily implement them.
Commentary
This is a preprint and a layer-construction benchmark, not an end-to-end spin-laser validation.
Especially relevant to anisotropic cavity optics, but carrier-spin statistics and dynamics still need their own model.
Interpretation and connections in this note, separate from the authors’ reported findings.