Inverse Design for Everyone

By Tyler Hughes

In this presentation, we introduce TidyGrad, an innovative inverse design tool integrated with Tidy3D's GPU-accelerated FDTD solver. TidyGrad leverages automatic differentiation to streamline the design of photonic devices, enabling users to define objective functions in Python and compute gradients efficiently using the adjoint method. This approach simplifies the optimization process, allowing for the creation of high-performance, non-intuitive photonic structures. The presentation covers the fundamentals of inverse design, the implementation of TidyGrad, and practical examples demonstrating its application in photonics engineering.

Share On:
Additional information: This Session was updated in Oct 10, 2024