SRC Spring 2017 Semi-annual Report - 2693.002 - Energy Efficient Computing with Chip-Based Photonics
We here present recent theoretical developments demonstrating an Optical Ising Machine based on a Neural Network architecture. This general approach relies on a Noisy Little Network, and is tailored for a silicon photonics implementation, where massive parallelization of matrix multiplication is enabled. This method allows to find the exact solution for any Ising problem, which can be mapped to many applications in combinatorial optimization problems in biology, medicine, wireless communications, artificial intelligence and social network. We here present theoretical foundations and first performance simulation of our algorithm.