Examples

The examples map onto DiffTMM's two main use cases — forward simulation of a film stack's optical response, and inverse design of layer thicknesses via differentiable optimization. Each example corresponds to a notebook in the repository root.

For accuracy validation against the reference NumPy TMM library and speed/memory comparisons, see the Benchmarks page.

Forward Simulation

Compute the optical response of a known film stack across angle and wavelength, including real dispersive materials and thick incoherent substrates.

Example	Notebook	Description
Forward Simulation	1_forward_simu.ipynb	Fresnel coefficients vs. angle and wavelength with the isotropic 2×2 solver
Real Materials	3_real_materials.ipynb	Dispersive n(λ), k(λ) materials — AR coatings and surface plasmon resonance
Incoherent Films	4_incoherent_film.ipynb	Thick substrates without unphysical Fabry–Perot ripples

Inverse Design

Optimize the film stack with PyTorch autograd, so layer thicknesses are designed end-to-end to match a target optical response.

Example	Notebook	Description
Inverse Design	2_inverse_design.ipynb	Recover five unknown layer thicknesses to sub-nanometer accuracy