Project Grant 2419889

The National Science Foundation (NSF) awarded a $599,474 Project Grant under the Computer and Information Science and Engineering (CFDA 47.070) program to the University of California, Santa Barbara (UCSB) to develop a real-time and energy-efficient neural network-based Partial Differential Equation (PDE) solver on a 2.5D photonic chip. The goal is to create a highly compressed and backward propagation-free training method for large-scale Physics-Informed Neural Networks (PINNs) that can be easily implemented on photonic hardware without using photonic memory. This project, which runs from October 2024 through September 2027, aims to enable the real-time and ultra-low energy solution of complex PDE problems in science and engineering applications such as aircraft design, semiconductor chip development, and autonomous vehicles.