Quantum computational advantage with a programmable photonic processor

Published

June 1, 2022

Author(s)

L.S. Madsen, F. Laudenbach, M.F. Askarani, F. Rortais, T. Vincent, J.F.F. Bulmer, F.M. Miatto, L. Neuhaus, L.G. Helt, Matthew Collins, Adriana Lita, Thomas Gerrits, Sae Woo Nam, V.D. Vaidya, M. Menotti, I. Dhand, Zachary Vernon, N. Quesada, J. Lavoie

Abstract

The demonstration of quantum computational advantage is a key milestone in the race to build a fully functional quantum computer. This milestone involves showing that a particular quantum device can perform a well-defined computational task in a manner that exceeds the best available classical algorithms and machines. To date just two quantum hardware platforms have furnished such claims: superconducting circuits [1, 2], and photonics [3, 4]. In both, the computational task involved sampling from a probability distribution that is prohibitively time-consuming to implement classically. In photonics, Gaussian Boson Sampling [5] (GBS) - the task of sampling from the photon number distribution of a Gaussian state - has emerged as the most achievable domain for demonstrating quantum advantage. While the superconducting circuit-based demonstrations were implemented in programmable devices, to date no fully programmable photonic machine has demonstrated quantum computational advantage; photonic-based claims were confined to use gate sequences that were largely fixed into static, randomized patterns. Furthermore, these photonic demonstrations were vulnerable to classical spoofing [6]: the ability for a hypothetical adversary using classical heuristics to produce samples, without direct simulation, that are difficult to distinguish from authentic samples produced by the quantum hardware. In this work we report a demonstration of quantum computational advantage using a fully programmable photonic processor. We carry out GBS with 216 squeezed modes entangled in a Gaussian state [7] with three-dimensional connectivity. Using a single, pulsed squeezed light source as an input, the entangling circuit is implemented by a dynamically programmable three-loop time-domain interferometer, and sampling accomplished using true photon-number-resolving detectors. Our programmable device generates random samples from a distribution that would be intractable to implement classically: it would take over 9000 years for the best available algorithms and supercomputers to produce even a single sample from the same distribution. This runtime advantage over classical techniques is over 700 times as extreme as that reported from earlier non-programmable photonic machines. We first validate our instances of GBS in few-mode and low-photon-number regimes, reaching over 99.8% fidelity with simulations. We then enter the many-mode, high-photon-number regime, and show that the samples from our quantum hardware outperform those from best known classical adversaries, as assessed using linear cross-entropy benchmarking and Bayesian log average scores. Our implementation constitutes the largest GBS experiment to date, with a mean detected photon number of up to 219, in a programmable device with 216 total modes and populated inputs. This work stands as a critical milestone in the rapidly intensifying efforts for the construction of a useful quantum computer, and provides strong validation of key technological features needed to support photonics as a platform toward this goal.

Citation

ACS Nature, Physics

Volume

606

Pub Type

Journals

Download Paper

https://doi.org/10.1038/s41586-022-04725-x

Local Download

Keywords

Photonic processor , quantum computer, Gaussian Boson Sampling, quantum computational advantage, photon-number-resolving detectors

Citation

Madsen, L. , Laudenbach, F. , Askarani, M. , Rortais, F. , Vincent, T. , Bulmer, J. , Miatto, F. , Neuhaus, L. , Helt, L. , Collins, M. , Lita, A. , Gerrits, T. , Nam, S. , Vaidya, V. , Menotti, M. , Dhand, I. , Vernon, Z. , Quesada, N. and Lavoie, J. (2022), Quantum computational advantage with a programmable photonic processor, ACS Nature, Physics, [online], https://doi.org/10.1038/s41586-022-04725-x, https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=933674 (Accessed October 27, 2025)