Oscillator-based Dynamical Computing Platforms to Solve Combinatorial Optimization

Abstract

Synchronized oscillator networks provide a promising pathway towards realizing application specific analog platforms capable of solving computationally intractable problems in combinatorial optimization. Specifically, electronic oscillators offer a low-power and integrated circuit (IC)-compatible hardware solution for designing such systems. In this work, we will present the opportunities as well as the challenges of realizing oscillator-based Ising machines to solve NP-hard problems like the Maximum Cut (Max-Cut) of a graph. Using experiments and simulations, we will discuss the performance and the scalability trade-offs that need to be considered in designing such platforms. Finally, we will showcase the opportunities in novel materials and beyond-CMOS devices that can be leveraged to augment the hardware implementation and performance of such systems.