Geometry evolution of microwave filters exploiting self-adjoint sensitivity analysis

We present an adjoint sensitivity approach for S-parameter tuning and performance optimization of microwave filters using HFSS. The method is applicable with any EM solver which can provide S-parameter response adjoint sensitivity with respect to the design dimensions. The sensitivities of the S-parameters relative to the vertices of different discontinuities are utilized to allow the shape of the structure to evolve thus producing new geometries. Our approach is illustrated through a reentrant cavity filter problem using the design dimension sensitivities.