To address the U.S. Air Force need for Developing innovative wave-optics Propagation methods to model laser systems that are faster, efficient and more accurate, Luminit, LLC, and University of Southern California (USC) propose to develop Wave-Optic Propagation Computation Enabled by Machine Learning Algorithms (WOPA). The proposed algorithms will be based on cutting off redundant frequencies upon artificial neural network training techniques. This way, we aim to speed up the algorithms of DFT by using the existing data to educate the system. In Phase I, Luminit and USC work on the feasibility of two different (machine-learning-based) approaches, and comparison to traditional Fourier transform-propagation-inverse transform approach. Phase II will be more concentrated on software issues and application to data from High Energy Laser (HEL) systems. At the end, an innovative software package, utilizing machine learning and neural network algorithms to the ubiquitous wave-optical beam propagations.
