Quantum-inspired and Quantum Optimization on a Superconducting Quantum Processor

Abstract

Quantum and Quantum-inspired optimization represent rapidly growing fields that combine classical optimization techniques with either quantum-inspired ideas or quantum hardware to address complex optimization problems. This thesis provides an overview of quantum-inspired optimization as well as quantum optimization, including the theoretical underpinnings of both processes on hardware and in software. In particular, this thesis considers a specific, practically relevant problem, a BMW production planning problem, and evaluates the performance of quantum-inspired optimizers. This evaluation is implemented by comparing the performance of a family of quantum-inspired optimizers with that of several common black-box combinatorial methods. We find that the use of important operations research techniques including the incorporation of domain-specific information as well as state-space pruning improves the performance of all solvers. In addition, we find that in a majority of tested cases, quantum-inspired methods tie or improve upon the results of their conventional counterparts, albeit by small margins, particularly in regimes of moderate state-space size. This thesis demonstrates that quantum-inspired optimization can outperform many conventional optimization methods in some cases, motivating future use and study of quantum-inspired protocals as well as implementation of fully-quantum optimization techniques.