Application-Driven Benchmarking of the Traveling Salesperson Problem: a Quantum Hardware Deep-Dive

TL;DR

This paper compares various quantum hardware architectures by applying the Traveling Salesperson Problem, providing insights into their practical capabilities and highlighting necessary steps for real-world quantum applications.

Contribution

It offers a comprehensive benchmarking framework for different quantum hardware using a real-world optimization problem, advancing understanding of their practical performance.

Findings

Superconducting hardware shows higher efficiency for TSP algorithms.

Ion trap and neutral atom hardware demonstrate comparable performance levels.

Identifies key steps needed to deploy quantum algorithms in real-world scenarios.

Abstract

The potential analysis of the capabilities of quantum computing, especially before fault tolerance at scale, is difficult due to the variety of existing hardware technologies with a wide spread of maturity. Not only the result of computations, but also the very process of running quantum-enhanced algorithms differ from provider to provider. The study includes a comparative analysis of various hardware architectures with the example of the Traveling Salesperson Problem, a central class of combinatorial optimization. It highlights what steps are necessary to run real-world applications on quantum hardware, showcases how the providers and various technologies differ and presents results in the relative efficiency of exemplary quantum algorithms on neutral atom-based, ion trap and superconducting hardware, the latter including both gate-based and annealing devices. This is an important step in advancing the understanding of quantum computing capabilities from an application standpoint - agnostic to the underlying qubit technology and projecting results into the future to judge what further developments on the application side are necessary.