Source-linked AI summary
Quantum computational advantage using photons
Han-Sen Zhong, Hui Wang, Yu-Hao Deng, Ming-Cheng Chen, Li-Chao Peng, Yi-Han Luo, Jian Qin, Dian Wu, Xing Ding, Yi Hu, Peng Hu, Xiao-Yan Yang, Wei-Jun Zhang, Hao Li, Yuxuan Li, Xiao Jiang, Lin Gan, Guangwen Yang, Lixing You, Zhen Wang, Li Li, Nai-Le Liu, Chao-Yang Lu, Jian-Wei Pan
TL;DR
Scaling boson sampling to a computationally interesting regime remained difficult for classical computers, so this experiment used squeezed states in a large, phase-stable interferometer and observed up to 76 photon-clicks in a state space reaching 10^30.
Problem
Scaling boson sampling to a computationally interesting regime remained difficult because the problem size grows exponentially for classical computers.
Method
The experiment used single-mode squeezed states in a phase-stable, fully connected interferometer implementing a random transformation.
Results
Up to 76 photon-clicks were observed, reaching a state-space dimension of 10^30.
Takeaways & Limitations
The experiment reaches a boson-sampling regime with a state space far beyond the directly verified small-photon-number regime.
Takeaways & Limitations
The results at large photon numbers cannot be directly calculated and verified.
Abstract
from arXiv · showhide
Gaussian boson sampling exploits squeezed states to provide a highly efficient way to demonstrate quantum computational advantage. We perform experiments with 50 input single-mode squeezed states with high indistinguishability and squeezing parameters, which are fed into a 100-mode ultralow-loss interferometer with full connectivity and random transformation, and sampled using 100 high-efficiency single-photon detectors. The whole optical set-up is phase-locked to maintain a high coherence between the superposition of all photon number states. We observe up to 76 output photon-clicks, which yield an output state space dimension of $10^{30}$ and a sampling rate that is $10^{14}$ faster than using the state-of-the-art simulation strategy and supercomputers. The obtained samples are validated against various hypotheses including using thermal states, distinguishable photons, and uniform distribution.
Abstract · Main text
The study develops large-scale Gaussian boson sampling with deterministic squeezed-state inputs and a stabilized, fully connected 100-mode photonic interferometer. It demonstrates high-photon-number quantum interference, validates the samples against competing hypotheses, and benchmarks sampling against classical computation.
- Main text: The experiment addresses the main scaling barriers—source purity, indistinguishability, interferometer stability, transmission, detection efficiency, and validation—that had limited computationally interesting boson sampling.Prior boson-sampling demonstrations culminated at 14-photon detection, whereas large-scale GBS requires these conditions simultaneously.
- Main text: The resulting photonic platform is positioned for applications in graph-based problems, point processes, and quantum chemistry, while supporting extensions to other quantum-state families.The paper identifies Fock, cat, and NOON states as examples of alternative state families.
- Main text: Gaussian boson sampling replaces single-photon inputs with deterministically prepared single-mode squeezed states whose output probabilities are governed by the computationally hard Torontonian.The Torontonian is an infinite sum of Hafnians and belongs to the #P complexity class.
- Main text: The experiment reaches a 76-photon coincidence and a state-space dimension of 10^30, establishing large-scale Gaussian boson sampling beyond prior demonstrations.Within 200 s, the experiment records 3,097,810 43-photon coincidence events and one 76-photon coincidence.
- Main text: Measured samples show genuine multiphoton quantum interference, deviating strongly from thermal-state behavior while agreeing with theoretical predictions.The observed distributions also significantly diverge from the tested competing hypotheses and agree with the theoretical prediction.
- Main text: Classical simulation becomes prohibitive at higher photon numbers, increasing from ~0.03 s for 30 photons to ~2 days for 50 photons and peaking near 70 photons.Extrapolating the classical cost to the device’s sample volume gives 8×10^16 s, or 2.5 billion years, for the corresponding computation.
Figure Captions
The captions describe the GBS architecture, squeezed-light sources, phase-locking system, experimental validation, and classical computational-cost estimation. Together, they document a 100-mode, detector-based implementation with characterized source quality and high agreement between measured and theoretical distributions.
- Implementation overview: The GBS device sends 50 single-mode squeezed states into a 100-mode interferometer and samples outputs with 100 single-photon detectors.
- Quantum light sources: The 25 squeezed-light sources have average purity 0.938 and average efficiency 0.628 after a 12-nm filter.
- Phase locking: Active phase locking stabilizes the optical phases from the photon sources to the interferometer using heterodyne detection and real-time PID feedback.
- Classical computational cost: The classical computational-cost figure estimates sampling time on the Sunway TaihuLight supercomputer, with error bars from Poissonian counting statistics.
- Experimental validation: For 23 input sets, the measured two-photon distributions achieve average fidelity 0.990(1) and total variation distance 0.103(1) relative to theory.