Application Of Generative Neural Networks To The Broadband Flat Lens Problem




Assistant Professor Electrical and Systems Engineering

Project Summary

            This research sought to provide a solution to the impending water crisis which was both clever and deceivingly simple; going off of a phenomenon called “nocturnal cooling” where heat on the surface of the Earth will escape to the cold of space at night, bringing about ice even in temperatures well above freezing, we sought to replicate the process to condense water from the atmosphere in the daytime. To do so, we used devices on the nanoscale called “metastructures”, particularly designed devices which can be designed to manipulate incoming light/electromagnetic energy in a number of counterintuitive and fascinating ways (such as a device getting cooler when it’s taken out of the shade). However, since the manual design of metastructures of any degree of complexity involves a nigh-impossible amalgam of electrodynamic calculations, we sought to instead look to AI (artificial intelligence) methods to instead learn the relationships of various representations of designs of metastructures and how they manipulate light, and then be able to generate from an input desired “result” for the incoming light a metastructure design that can approximately achieve said result. Our hope, in future work, is to pursue this general method in application to a series of fascinating and salient optics problems.

            Throughout my research experience, I gained not only significant technical knowledge concerning optics, electrodynamics, and machine learning, but also about the research process and how to generate novel solutions to rather obscure problems; working in a field with little precedent research, I often ran into issues that had never been addressed before, which forced me to both think on my feet and approach my work in a dually dynamic and invested manner. As a student who had only cursorily dabbled in the prospect of research beforehand and had ambitions of committing to physics research, my time in the Raman Lab both informed and encouraged that prospect, and both gave me the tools and the requisite confidence in myself to pursue further intriguing scientific loose ends in my research work.