Optimal design of dielectric flat lens utilizing Bayesian optimization

In recent years, various kinds of optimal design approaches are intensively studied for exploring novel and higher performance microwave and optical devices beyond human knowledge. In this article, we first apply a design approach based on Bayesian optimization (BO) to design of dielectric flat lens which is expected to be used in microwave and millimeter wave region because of its excellent features, such as thin and light weight. BO is able to efficiently search optimal values of design variables compared with population based optimization, without requiring sensitivity analysis which is mandatory in the gradient method.