Adjoint-Based Inverse Design with Transfer Matrices

Inverse design (InvDes) with gradient-based optimization has become a practical method to design high-performance physical structures that are nearly impossible to produce with human intuition. In particular, InvDes of nanophotonic devices with adjoint method and finite-difference electromagnetic simulation has made significant strides towards solving real-world issues and advancing fundamental science. Here we demonstrate adjoint-based InvDes with the transfer matrix method (TMM). Thanks to the broad applicability of TMM, we show that we can design non-intuitive, high-performance layered structures for photonics, acoustics, thermal, and quantum applications.