PaperlayerView exact revision on arXiv

Source-linked AI summary

Programmable four-photon graph states on a silicon chip

Jeremy C. Adcock, Caterina Vigliar, Raffaele Santagati, Joshua W. Silverstone, Mark G. Thompson

arXiv:1811.03023v1quant-ph

TL;DR

Chip-scale photonic quantum computing needs graph states and multi-pair operation, while earlier silicon experiments used only one or two photons. This work generates four-photon star and line graph states on chip and demonstrates a basic measurement-based protocol; the authors identify improved photon throughput as the requirement for further increases in photon number.

  • Problem

    Earlier silicon quantum photonics scaled by increasing the number of modes while using only one or two photons, leaving multi-pair operation an important step.

    Source: p. 1

  • Method

    The authors use a silicon device to generate four photons and prepare star- and line-type graph states, then project qubits to implement a basic measurement-based protocol.

    Source: pp. 1, 3

  • Results

    The device generates both four-photon star and line graph states on chip and implements a basic measurement-based protocol.

    Source: pp. 1, 3

  • Takeaways & Limitations

    The authors state that further increases in on-chip photon number depend on engineering improved photon throughput.

    Source: p. 4

    The authors state that their postselection-reliant approach to sourcing photons and preparing entanglement is not scalable.

    Source: p. 4

Abstract

from arXiv · show

Future quantum computers require a scalable architecture on a scalable technology---one that supports millions of high-performance components. Measurement-based protocols, based on graph states, represent the state of the art in architectures for optical quantum computing. Silicon photonics offers enormous scale and proven quantum optical functionality. Here we report the first demonstration of photonic graph states on a mass-manufactured chip using four on-chip generated photons. We generate both star- and line-type graph states, implementing a basic measurement-based protocol, and measure heralded interference of the chip's four photons. We develop a model of the device and bound the dominant sources of error using Bayesian inference. The two-photon barrier, which has constrained chip-scale quantum optics, is now broken; future increases in on-chip photon number now depend solely on reducing loss, and increasing rates. This experiment, combining silicon technology with a graph-based architecture, illuminates one path to a large-scale quantum future.

Methods

The methods describe the chip setup, calibration, and measurements used to characterize graph-state fidelities and device errors. Bayesian parameter estimation compares model predictions with measured observables, under independent-measurement and constant-prior assumptions.

Methods

  • The experimental setup uses fibre-coupled on-chip waveguides, filtered signal and idler photons, and four off-chip superconducting nanowire detectors.The pump pulses are injected at 1544.40 nm, and the device phase shifters are driven by analogue voltage drivers.Source: Methods · p. 6
  • Thermo-optic phase shifters are calibrated by fitting Mach–Zehnder interferometer fringes and numerically solving for the voltage that dials in a target phase.Loss-matched waveguide taps with 2% transmission allow independent calibration of each on-chip phase shifter.Source: Methods · p. 6
  • Thermal cross-talk is 0.003 rad/mW; estimated average thermo-optic phase errors are 0.12 rad for the star state and 0.065 rad for the line state.The estimates follow measured phase deviations and power dissipation across the stabilizer-measurement configurations.Source: Methods · p. 6
  • The signal-photon path has an estimated insertion loss of 19.3 dB, with device insertion loss measured at 26.1 dB.The estimates use component losses measured on test structures on the same die.Source: Methods · p. 6
  • The equivalent HOM-dip visibility is calculated from fringe maxima and minima as (Nmax − 2Nmin)/Nmax.The conversion accounts for an ideal HOM-fringe maximum being twice the distinguishable background level of an ideal HOM dip.Source: Methods · p. 6
  • State fidelities are computed from the graph-state stabilizer expectations, measured through local Pauli projections and n-fold coincidence counts.The method is used for all reported state fidelities.Source: Methods · p. 6
  • For both star and line states, every two-setting Mermin test composable from stabilizers is measured; the three-setting test sums ⟨g_i⟩ over all 16 graph-state stabilizers.Source: Methods · p. 6
  • Bayesian inference estimates model parameters by comparing predicted observables with measured data and returns parameter distributions under independent-measurement and constant-prior assumptions.The likelihood is built from simulated data frequencies.Source: Methods · p. 7
Loading 1811.03023v1…