Robust microwave design optimization using manifold mapping with adjoint sensitivity

We introduce a robust and reliable algorithm for computationally efficient design optimization of microwave structures evaluated with electromagnetic (EM) simulations. Our approach utilizes low-fidelity model constructed from coarse-discretization EM simulation of the structure of interest as well as cheap derivative information obtained through adjoint sensitivity. Adjoint sensitivity which is exploited threefold: (i) to reduce the misalignment between the low- and high-fidelity models through input space mapping, (ii) to construct a first-order consistent surrogate model through manifold mapping, (iii) to optimize the surrogate through trust-region based algorithm. The comprehensive use of adjoint sensitivity results in a very low design cost and excellent convergence capability of the proposed algorithm. Efficiency of our approach is demonstrated through the design of an ultra-wideband antenna and a microstrip bandpass filter.