Gaussian Process Regression for the Modeling of Metalenses

Metasurfaces have attracted a lot of attention due to their exceptional capability to manipulate the amplitude, phase, and polarization of light. In the domain of imaging, metasurface-based lenses, also called metalenses have become a very promising technology for multifunctional and compact imaging devices. In this work, Gaussian process regression (GPR) models are investigated to model quantities of interest to describe the electromagnetic behavior of metalenses. Single output and multitask (multioutput) GPR models are explored. Pertinent numerical results will be presented to validate the proposed GPR modeling for metalenses.