Validation of a noisy Gaussian boson sampler via graph theory

TL;DR

This paper evaluates the Borealis Gaussian Boson Sampler's performance and validation in noisy conditions, leveraging graph theory to assess its potential for quantum advantage and graph-related computational problems.

Contribution

It introduces a validation protocol for noisy Gaussian Boson Sampling devices using graph theory, linking quantum sampling to graph isomorphism and similarity problems.

Findings

Borealis performs reliably under experimental noise.

Graph theory methods effectively validate Gaussian Boson Sampling.

Potential applications in graph isomorphism and similarity detection.

Abstract

Quantum photonic processors are emerging as promising platforms to prove preliminary evidence of quantum computational advantage towards the realization of universal quantum computers. In the context of non-universal noisy intermediate quantum devices, photonic-based sampling machines solving the Gaussian Boson Sampling problem currently play a central role in the experimental demonstration of a quantum computational advantage. In particular, the recently developed photonic machine Borealis, a large-scale instance of a programmable Gaussian Boson Sampling device encoded in the temporal modes of single photons, is available online for external users. In this work, we test the performances of Borealis as a sampling machine and its possible use cases in graph theory. We focused on the validation problem of the sampling process in the presence of experimental noise, such as photon losses, that could undermine the hardness of simulating the experiment. To this end, we used a recent protocol that exploits the connection between Guassian Boson Sampling and graphs perfect match counting. Such an approach to validation also provides connections with the open question on the effective advantage in using noisy Gaussian Boson Sampling devices for graphs similarity and isomorphism problems.