Inverse Design#
The adjoint method is an extremely powerful tool for photonics optimization, allowing gradient computation of a cost function in just two simulation steps, regardless of the number of free parameters. While powerful, its implementation can be quite complex. Tidy3D leverages the Autograd package to automatically differentiate a Tidy3D simulation using the adjoint method, making it straightforward to implement adjoint optimization techniques. This section introduces the adjoint method and the inverse design plugin and provides a comprehensive list of application examples.
- Inverse design plugin
- Inverse design quickstart
- Autograd, automatic differentiation, and adjoint optimization: basics
- Adjoint analysis of a multi-layer slab
- Inverse design optimization of a mode converter
- Multi-objective adjoint optimization
- Inverse design optimization of a waveguide taper
- Inverse design optimization of a compact grating coupler
- Inverse design optimization of a metalens
- Adjoint-based shape optimization of a waveguide bend
- Adjoint optimization of a wavelength division multiplexer
- Diffractive metasurface inverse design with topology optimization
- Adjoint inverse design of a quantum emitter light extractor
- Inverse design optimization of a plasmonic nanoantenna metasurface
- Inverse design optimization of a bilayer grating coupler
- Adjoint optimization of an integrated bandpass filter
- Topology optimization of a waveguide bend
- Inverse design of an apodized grating coupler through shape optimization
- Design and shape optimization of a metalens-assisted waveguide taper
- Inverse design of a GaP photon extractor for nitrogen-vacancy centers in diamond
- Adjoint optimization of a photonic crystal
- Fabrication-Aware Inverse Design (FAID) of a Wavelength Division Multiplexer