PaperlayerView exact revision on arXiv

Source-linked AI summary

Photonic Quantum Metrology

Emanuele Polino, Mauro Valeri, Nicolò Spagnolo, Fabio Sciarrino

arXiv:2003.05821v1quant-ph

TL;DR

The paper addresses how quantum resources and photonic platforms can improve estimation of unknown parameters, including phases and multiple parameters. It reviews foundational bounds, photonic states and strategies, recent advances, and open challenges, including a demonstrated genuine violation of the SQL with a two-photon N00N state. The review concludes that optimal probes and measurements for general noisy multiparameter settings remain unresolved.

  • Problem

    Quantum metrology seeks precision beyond classical strategies for estimating unknown parameters, including simultaneous estimation of multiple parameters and operation in the presence of noise.

  • Method

    The review synthesizes quantum-metrology foundations, photonic state-generation platforms, adaptive strategies, phase estimation, and experimental multiparameter implementations.

  • Results

    ηNV 2N ≈1.23 demonstrated genuine quantum enhancement in 2017 using an N = 2 polarization N00N state, while photonic multiparameter experiments include entangled-versus-separable estimation and joint phase–visibility estimation.

  • Takeaways & Limitations

    Photonic technologies provide platforms for quantum-enhanced metrology across phase, multiparameter, and noisy estimation tasks, but performance depends on experimental resources and strategy design.

  • Takeaways & Limitations

    General recipes for optimal probe states and measurement strategies in multiparameter scenarios, especially with noisy mixed probes, are still lacking.

Abstract

from arXiv · show

Quantum Metrology is one of the most promising application of quantum technologies. The aim of this research field is the estimation of unknown parameters exploiting quantum resources, whose application can lead to enhanced performances with respect to classical strategies. Several physical quantum systems can be employed to develop quantum sensors, and photonic systems represent ideal probes for a large number of metrological tasks. Here we review the basic concepts behind quantum metrology and then focus on the application of photonic technology for this task, with particular attention to phase estimation. We describe the current state of the art in the field in terms of platforms and quantum resources. Furthermore, we present the research area of multiparameter quantum metrology, where multiple parameters have to be estimated at the same time. We conclude by discussing the current experimental and theoretical challenges, and the open questions towards implementation of photonic quantum sensors with quantum-enhanced performances in the presence of noise.

I. INTRODUCTION

Quantum metrology estimates unknown parameters using quantum resources to seek precision beyond classical strategies. The review introduces estimation protocols, photonic platforms, phase estimation, and multiparameter extensions.

  • Motivation: Quantum metrology seeks ultimate estimation precision by exploiting quantum probes, potentially surpassing the classical scaling ∼m^-1/2.The Heisenberg Limit scales as 1/m under the stated conditions.
  • Photonic platforms: Photons are promising probes because they combine high mobility, low environmental interaction, and mature generation, manipulation, and detection technologies.Interferometry is especially relevant because many physical problems can be mapped to phase estimation.
  • Multiparameter estimation: Multiparameter quantum metrology studies simultaneous estimation of multiple unknown parameters and the possibility of quantum enhancement.Open questions include reaching quantum ultimate bounds for all parameters simultaneously.
  • Estimation protocol: The general estimation process prepares a probe, encodes the parameter through an evolution, measures the probe, and estimates the parameter from outcomes.The unitary case gives ρλ = Uλρ0U†λ, while non-unitary maps can also be considered.
  • Estimators: Maximum likelihood estimation selects the parameter value maximizing the likelihood of observed outcomes and is asymptotically unbiased, consistent, and Cramér-Rao efficient.Bayesian estimation and the Method of Moments provide alternative post-processing approaches.

B. Fisher Information and Cramer Rao bound

Fisher Information quantifies information about a parameter in measurement outcomes and supports the Cramér-Rao bound on estimator variance. Its convexity and additivity characterize mixtures and independent probes.

  • Fisher Information: Fisher Information measures the information about an unknown parameter encoded in the output probabilities of a fixed probe-and-measurement process.It captures sensitivity through derivatives of the output probabilities.
  • Properties: Fisher Information is convex for mixed states and additive across ν independently measured probes, with F_tot(λ) = ∑i F_i(λ).Additivity decomposes total information into the contributions of individual probes and measurements.
  • Cramér-Rao bound: The Cramér-Rao bound links Fisher Information to the minimum variance achievable by estimators using ν identical independent probes and measurements.For asymptotically locally unbiased estimators, the bound takes its locally unbiased form.
  • Estimator efficiency: Maximum likelihood and Bayesian estimators are asymptotically efficient, but finite data do not guarantee saturation of the Cramér-Rao bound.Efficiency means saturating the relevant inequality.

C. Quantum Fisher Information and Quantum Cramer Rao Bound

Quantum Fisher Information optimizes Fisher Information over measurements and defines a measurement-independent precision bound. Probe-state optimization and entanglement can yield Heisenberg scaling, while assumptions and implementation resources determine its scope.

  • Quantum Fisher Information: Quantum Fisher Information is the maximum Fisher Information over all POVMs and extends the Cramér-Rao bound to the Quantum Cramér-Rao bound.For a fixed probe state, the resulting bound is independent of the measurement.
  • Optimal measurements: In single-parameter estimation, suitable POVMs can saturate the Quantum Cramér-Rao bound, and local adaptive measurements avoid entangling measurements.Projectors onto symmetric-logarithmic-derivative eigenstates provide an optimal POVM.
  • Standard Quantum Limit: For m classically correlated probes, the uncertainty scales as Δλ ∝ 1/√m, defining the Standard Quantum Limit.This scaling follows from Fisher-information additivity and the central limit theorem.
  • Heisenberg Limit: For maximally entangled probes under parallel linear unitary evolution, Δλ ∝ 1/m, improving precision by √m over the Standard Quantum Limit.This scaling is identified as the Heisenberg Limit.
  • Scope of scaling laws: The Heisenberg scaling applies to parallel strategies with linear unitary evolution; nonlinear interactions or resources such as running time can change the scaling.Multiround protocols can beat the Standard Quantum Limit with non-entangled probes when additional resources are counted.

III. PHOTONIC QUANTUM METROLOGY: SCHEMES AND PLATFORMS

Photonic quantum metrology uses light’s field modes and optical transformations to encode and estimate parameters, especially phase shifts. The review develops discrete- and continuous-variable descriptions of photonic probes.

  • Photonic platforms: Photons are useful metrological probes because they have high mobility, low decoherence, and technologies for generating, manipulating, and detecting encoded quantum states.Their degrees of freedom include mode, polarization, frequency, and time-bin encodings.
  • Discrete variables: A photonic mode is described with annihilation and creation operators, whose number operator defines photon occupation and whose Fock states have fixed photon number.The radiation field across modes can be represented using tensor products of individual mode states.
  • Continuous variables: Quadrature operators are Hermitian field observables that describe photonic states in phase space, particularly continuous-variable states.Rotated quadratures are parameterized by an angle θ.
  • Optical transformations: Linear and bilinear optical transformations implement mode mixing, displacement, and squeezing, with passive elements such as beam splitters and phase shifters conserving photon number.The corresponding evolution can be expressed through a Bogoliubov transformation.
  • Phase estimation: Phase estimation is a paradigmatic photonic task in which the unknown parameter is a phase shift between optical modes.Interferometry is therefore a central scenario for photonic quantum metrology.

A. Phase estimation problem

Phase estimation infers an unknown phase shift between optical modes, commonly using interferometers such as the Mach-Zehnder interferometer. Its beam splitters prepare and measure the probe, while output probabilities encode phase-dependent information.

  • Phase estimation seeks to infer an unknown phase shift φ between two modes or optical paths.
  • A phase shift is represented as unitary evolution generated by the photon-number operator of the affected mode.
  • The standard quantum limit for phase estimation follows from the single-probe number-operator eigenvalue difference hS − hs = 1.
  • Mach-Zehnder interferometer: A Mach-Zehnder interferometer uses two beam splitters separated by a phase shifter, with the first beam splitter preparing the probe and the last contributing to measurement.
  • Mach-Zehnder interferometer: Larger variations in the interferometer’s output fringe pattern yield greater Fisher Information and phase sensitivity.
  • Mach-Zehnder interferometer: Mach-Zehnder transformations can implement arbitrary SU(2) operations, and analogous interferometers use half-wave plates for polarization.

B. States and schemes

This section introduces coherent states as classical benchmarks for optical phase estimation and describes their phase-space properties. Coherent states attain the standard quantum limit but not the Heisenberg limit.

  • Coherent states: Coherent states are eigenstates of the annihilation operator and provide a classical benchmark for quantum phase-estimation schemes.
  • Coherent states: Their photon-number statistics are Poissonian, with mean and variance both equal to |α|^2.
  • Coherent states: Coherent states are minimum-uncertainty states that symmetrically saturate the Heisenberg relation for quadratures.
  • Coherent states: A displacement operation translates a vacuum state in phase space without changing its uncertainties.
  • Coherent states: Coherent states reach the SQL but not the HL in phase estimation, with error scaling 1/|α|.

2. N00N states

N00N states are maximally entangled fixed-number states whose phase response is amplified by the photon number, enabling Heisenberg-limited local estimation. Their preparation and global phase discrimination remain constrained.

  • N00N states: N00N states are maximally entangled multipartite states distributed across two modes with fixed photon number N.
  • Metrological performance: Their number variance is N^2/4, yielding the Heisenberg limit for phase sensitivity.
  • Metrological performance: A relative phase shift φ produces an amplified phase shift Nφ in a N00N state.
  • Generation: For N = 2, indistinguishable photons entering a symmetric beam splitter generate a N00N state deterministically through the Hong-Ou-Mandel effect.
  • Generation: No equivalent deterministic generation scheme exists for N00N states with N > 2.
  • Limitations: N00N states are optimal for small unknown phases but cannot distinguish shifts differing by π/N without prior phase knowledge.

4. Squeezed states

Squeezed states reduce fluctuations in selected quadratures and provide Gaussian resources for quantum metrology. Their preparation and detection use optical nonlinearities, beam splitters, and homodyne measurements.

  • Single-mode squeezing: A squeezed state has quadrature fluctuations below those of the vacuum while saturating the Heisenberg relation.
  • Single-mode squeezing: For squeezing angle θ = 0, one quadrature’s uncertainty decreases by e^|r| while the conjugate quadrature increases by the same factor.
  • Single-mode squeezing: Squeezed vacuum states contain only even photon numbers and have mean photon number ⟨n⟩ = sinh^2|r|.
  • Two-mode squeezing: Two-mode squeezed vacuum states exhibit nonclassical correlations between modes and approach the EPR state as |r| becomes large.
  • Generation: Two-mode squeezed states can be generated by combining squeezed states at a beam splitter or splitting a squeezed state with a polarizing beam splitter.
  • Homodyne measurements: Homodyne detection interferes a target signal with a coherent local oscillator to measure quadratures, with the oscillator phase selecting the measured quadrature.

5. Other states

The review surveys photonic quantum states and encoding degrees of freedom relevant to quantum metrology, including entangled, continuous-variable, and path-based resources.

  • Other states: Holland-Burnett states generalize the Hong–Ou–Mandel effect by interfering two indistinguishable N/2-photon beams at a beam splitter.For N = 2, the resulting state coincides with a N00N state and reaches Heisenberg scaling up to a constant factor.
  • Other states: Dicke states are symmetric superpositions of N qubits containing k excitations, with twin Fock states arising when k = N/2 for even N.Dicke-squeezed states are described as more robust against noise, while multimode photons from collective Dicke superradiance can support metrology.
  • Other states: Entangled coherent states combine vacuum and coherent-state components across two modes and can be generated with a coherent state and a Schrödinger cat state at a beam splitter.These states are identified as suitable for experimental generation and robust with respect to noise.
  • Photonic encodings: Photonic quantum information can use polarization, orbital angular momentum, time-bin, time-frequency, and field-quadrature degrees of freedom.Polarization uses horizontal and vertical states, while orbital angular momentum is associated with spatial modes and quantized angular momentum.
  • Photonic encodings: Path encoding is widely used in quantum metrology, with beam splitters and phase shifters enabling manipulation of spatial modes and construction of interferometers.Interferometric setups form the basis of many quantum metrology tasks.

2. Generation and detection of photons

The review describes photon-pair sources, detectors, and integrated photonic platforms for quantum metrology. It emphasizes that genuine quantum enhancement requires efficient, stable, and scalable implementations that account for losses and detection resources.

  • Photon generation: SPDC and SFWM generate signal-idler photon pairs through nonlinear optical processes subject to energy and momentum conservation.SPDC uses χ2 materials, whereas SFWM occurs in χ(3) nonlinear waveguides; both can produce heralded single photons and entangled states.
  • Photon generation: 33. SPDC and SFWM are probabilistic, with typically low pair-generation probabilities, so deterministic single-photon sources such as quantum dots or color centers are also used.The cited alternatives target on-demand single-photon generation.
  • Integrated platforms: Integrated photonic circuits address bulk-platform limitations by improving stability, scalability, miniaturization, flexibility, cost reduction, standardization, efficiency, and precision.Complex bulk-optical schemes may require hundreds of elements and strict environmental control for accurate phase control.
  • Integrated platforms: Integrated platforms use waveguides, directional couplers, and tunable phase shifts to control interactions between optical modes.Silicon-based, femtosecond-laser-written, III-V, and UV-written platforms are among the technologies discussed.
  • Integrated platforms: The target of quantum integrated photonics is simultaneous on-chip generation, manipulation, and detection of quantum states while supporting multiple photonic degrees of freedom.Integrated sources, detectors, and path, polarization, time, and orbital-angular-momentum components have all seen progress.
  • Phase-sensitive states: Genuine quantum enhancement requires counting all employed photons, including losses and undetected photons, with transmission, detection efficiency, and interference visibility as crucial parameters.Post-selection losses can prevent enhanced sensitivity even when optical elements are lossless.
  • Phase-sensitive states: Experimental phase-estimation schemes have demonstrated super-resolution, but imperfections have prevented effective unconditional super-sensitivity in several realizations.The first unconditional SQL violation used a polarization N00N state with N = 2; the reported value was ηNV 2N ≈1.23.
  • Phase-sensitive states: N00N states are definite-photon-number states measured by single-photon counting, but low generation and detection efficiency limits scalable unconditional advantages at large N.Beyond N = 2, arbitrary-N deterministic generation is unavailable in the reviewed schemes, so higher-photon-number states generally rely on post-selection.

2. Platforms for squeezed states

Photonic squeezed states are generated through atomic, crystal, and optomechanical platforms and applied mainly to interferometric sensing, especially gravitational-wave detection. Experiments demonstrate improved interferometric sensitivity, while practical performance remains constrained by losses, detection efficiency, and photon-number requirements.

  • Metrological applications: Proof-of-principle experiments show that squeezed states beat shot noise and improve interferometer sensitivity.Squeezed states have also been used with homodyne detection to achieve super-sensitivity and super-resolution simultaneously.
  • Generation platforms: Squeezed light is generated using atoms, nonlinear crystals, or optomechanical systems, with nonlinear crystals producing squeezed light through parametric down conversion.The reported maximum squeezing factors are 14.9 dB for atoms, 25 dB for optomechanics, and 19 dB for nonlinear crystals.
  • Gravitational-wave detection: Gravitational-wave detection is a leading application because wave amplitudes near 10^-22 require long interferometers and very low overall noise.Increasing input photons improves signal-to-shot-noise ratio, but thermal mirror displacement increases with laser power.
  • Gravitational-wave detection: Squeezed vacuum has been adopted in GEO600 and tested in LIGO, while further detector improvements appear to depend substantially on squeezing enhancement.Frequency-dependent and EPR-entangled squeezing schemes address frequency-dependent quantum noise and broadband operation.

E. Other schemes and platforms

Alternative photonic platforms use orbital angular momentum, Holland-Burnett and QFT-generated states, nonlinear interferometers, and correlated photons for enhanced sensing and imaging. Adaptive protocols update probe controls from measurement outcomes and can improve convergence toward ultimate precision with limited data.

  • OAM-based estimation: A single photon in a superposition of opposite OAM modes experiences rotation-amplified interference, yielding sensitivity Δθ = 1/(m√ν).The enhancement factor m arises from the superposition of m-quanta of orbital angular momentum, while single-photon probes offer advantages in generation, detection, and loss robustness over N00N states.
  • Nonlinear interferometry: SU(1,1) interferometers replace beam splitters with nonlinear interactions that generate entanglement and can retain quantum-enhanced performance in the presence of losses.Implementations include four-wave mixing and parametric down conversion.
  • Quantum imaging: Ghost imaging reconstructs a sample image from spatial correlations between photon pairs, with spatial resolution obtained from the photon that does not interact with the sample.Related schemes can perform phase and intensity imaging without detecting photons that interacted with the object.
  • Adaptive estimation: Adaptive protocols prepare each probe, let it interact with the unknown parameter, measure the output, and use the result to select controls for the next probe.With limited probe numbers, adaptive strategies can boost convergence toward asymptotic precision bounds.

1. Adaptive Bayesian protocols

Adaptive Bayesian and machine-learning protocols update or optimize feedback strategies to improve photonic phase estimation, with experiments surpassing the SQL and some schemes reaching HL scaling.

  • Bayesian adaptive protocols: Bayesian adaptive protocols update the posterior after measurement outcomes and use it to select subsequent feedback actions.The posterior can guide optimal feedback according to the protocol heuristic.
  • Bayesian adaptive protocols: N = 1,2,4 photon Holland–Burnett sequences surpassed the SQL in an ab-initio phase-estimation experiment.The protocol optimized an expected sharpness function after posterior updates.
  • Bayesian adaptive protocols: Single photons in a multipass polarization interferometer can reach the HL without entanglement using Bayesian estimation and a generalized Kitaev’s algorithm.A separate hybrid approach combined polarization-entangled two-photon states with multipass operation.
  • Machine learning offline estimation techniques: PSO and DE self-learn feedback policies for phase estimation using model-free reinforcement learning and evolutionary search.Their population-based stochastic exploration is described as reducing the probability of becoming trapped at local optima.
  • Machine learning offline estimation techniques: Offline optimization computes phase-shift policies before experiments, while the final estimate is the last adaptive feedback phase Φ_N.PSO maximizes an average sharpness objective related to minimizing Holevo variance.
  • Multiparameter constraints: In multiparameter estimation, the QCRB need not be attainable because optimal measurements for different parameters may not commute.For pure probes with commuting generators, the cited conditions allow QCRB saturation.

B. Multiphase estimation

Multiphase estimation treats relative phases on multiple interferometer paths as simultaneous parameters. Quantum probes can improve joint precision, but attainable bounds depend on probe states, measurements, generator commutation, and entanglement.

  • Multiphase formulation: Multiphase estimation concerns relative phases on d interferometer arms measured against a common reference.The general scheme prepares a probe across d +1 paths, applies phase evolution, measures the output, and estimates all phases.
  • Multiphase formulation: For commuting phase generators O_i, the quantum Fisher information matrix is determined by covariances of the generators.For independent optical modes, O_i = n_i, giving FQ(φ)_ij = 4[⟨n_i n_j⟩−⟨n_i⟩⟨n_j⟩].
  • Quantum enhancement: Optimized fixed-photon-number probe states distributed across d +1 modes can attain a characterized total variance for simultaneous phase estimation.The reference arm contains the final term of the superposition, with N photons occupying that arm.
  • Quantum enhancement: Simultaneous estimation can provide an O(d) variance advantage over optimal separate quantum single-phase estimation.The cited enhancement also occurs for noncommuting unitary generators and with small amounts of loss.
  • Quantum enhancement: Separable probes define a classical sensitivity bound, while indistinguishable photons can surpass it in multimode interferometers.Useful entanglement is identified when estimator variance falls below the separable-state bound.

1. Photonic platforms for multiphase estimation problems

Photonic platforms support multiphase estimation, with integrated circuits providing stable, scalable interferometers and experiments demonstrating quantum advantage. Distributed sensing also shows an entangled improvement, while noise limits achievable precision.

  • Platform motivation: Photonic multiphase estimation remains experimentally underdeveloped despite photonic systems being a natural platform for the problem.The review notes that relatively few experimental realizations had been reported.
  • Integrated platforms: Integrated photonics offers scalable, stable interferometers for estimating relative phases along multiple spatial paths.Its stability helps avoid thermal-fluctuation and mechanical-vibration problems associated with bulk optics.
  • Integrated platforms: The first experimental quantum-enhanced multiphase implementation used a three-mode integrated interferometer with two cascaded tritters and two-photon measurements.The circuit’s Fisher-information matrix exceeded the optimal separable-probe benchmark in some phase settings.
  • Distributed sensing: M = 4-node distributed sensing achieved σ_ent = 0.099±0.003 versus σ_sep = 0.118 ± 0.002 for the estimated average phase.The experiment used squeezed coherent probes with approximately 2.5 photons per mode.
  • Noise and open problems: Photon losses and phase diffusion limit quantum-enhanced phase precision, which can vanish when noise becomes significant.Time-varying systematic errors can require simultaneous estimation of phase and noise through a non-unitary model.

1. Phase and phase diffusion estimation

Photonic multiparameter metrology estimates phase parameters together with diffusion or visibility noise, using quantum probes and measurements to approach fundamental precision bounds. Experiments demonstrate simultaneous estimation, real-time tracking, and applications to biological and chemical processes, while losses and noise remain major challenges.

  • Phase and phase diffusion estimation: The phase-and-dephasing problem has been studied across qubit and multiqubit platforms, including independent and collective dephasing scenarios.The quantum Fisher information matrix is used to characterize the estimation limits for the considered qubit state.
  • Phase and phase diffusion estimation: Phase and dephasing can be estimated simultaneously, with collective measurements improving the attainable precision even for separable qubit probes.For two collectively measured qubit probes, κ ≤1.5, whereas independently measured probes satisfy κ ≤1.
  • Phase and visibility estimation: Simultaneous estimation of optical phase φ and interference visibility v prevents phase bias when visibility is not properly estimated.A Mach–Zehnder interferometer with a polarization N = 2 N00N state uses coincidence measurements and Bayesian learning for biological samples.
  • Phase and visibility estimation: 2.63 for fructose and 0.10 for sucrose were compatible with the likelihood-ratio test’s null hypothesis at a 95% confidence interval.The test assessed whether the covariance matrix of the two estimated parameters saturated the Cramér–Rao bound.
  • Phase and visibility estimation: The same phase-and-visibility scheme monitored real-time visibility v(t) and phase φ(t) during sucrose acid hydrolysis.The chemical reaction changes optical activity from dexorotatory to levorotatory, producing a measurable polarization-phase variation.
  • Challenges and perspectives: Photonic multiparameter sensing remains limited by losses and realistic noise, which can eliminate quantum scaling advantages for sufficiently large photon numbers.Current work therefore seeks robust strategies and improved photonic platforms for more complex noisy scenarios.
Loading 2003.05821v1…