Design of Polarization Rotator Metamirror Using Metaparticle Surrogate-Based Optimization
We demonstrate the viability of designing a polarization rotator surface using metaparticle surrogate modeling and genetic algorithm optimization. A kriging surrogate is constructed to model the polarization conversion ratio (PCR) of a polarization rotator metaparticle with three degrees of freedom (geometrical parameters). The input to the kriging model is provided by 125 full-wave simulations sampled using the hyper Latin hypercube sampling algorithm, and surrogate model is used to construct the cost function for the genetic algorithm. As an illustrative example, an optimization target is set for PCR greater than or equal to $0.85$ over the frequency range 10 to 12 GHz. The result from the optimization algorithm is compared to full-wave simulation and very good agreement is observed.