Feedback-based quantum optimization

TL;DR

This paper introduces a feedback-based quantum optimization method that improves solution estimates for combinatorial problems without classical optimization, demonstrated on a superconducting quantum processor for MaxCut.

Contribution

It presents a novel feedback-driven approach for quantum optimization that enhances solution accuracy without classical optimization, differing from QAOA.

Findings

Monotonically improving solution estimates with circuit depth

Successful experimental demonstration on superconducting quantum processor

Numerical analyses confirming protocol performance

Abstract

It is hoped that quantum computers will offer advantages over classical computers for combinatorial optimization. Here, we introduce a feedback-based strategy for quantum optimization, where the results of qubit measurements are used to constructively assign values to quantum circuit parameters. We show that this procedure results in an estimate of the combinatorial optimization problem solution that improves monotonically with the depth of the quantum circuit. Importantly, the measurement-based feedback enables approximate solutions to the combinatorial optimization problem without the need for any classical optimization effort, as would be required for the quantum approximate optimization algorithm (QAOA). We experimentally demonstrate this feedback-based protocol on a superconducting quantum processor for the graph-partitioning problem MaxCut, and present a series of numerical analyses that further investigate the protocol's performance.