Pith. sign in

REVIEW 4 major objections 4 minor 54 references

Quantum Analog-to-Digital Converter for Phase Estimation

T0 review · 4 major / 4 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The first experimental quantum analog-to-digital converter encodes an unknown optical phase into three bits and, after machine-learning denoising, extracts more phase bits per photon than classical interferometry.

desk verdict First hardware step toward digital phase estimation, with a real caveat: the headline advantage after denoising may be an artifact of training on the ideal model. read the letter →

arxiv 2502.07676 v1 pith:DVCNMO6N submitted 2025-02-11 quant-ph

Eugenio Caruccio , Simone Roncallo , Valeria Cimini , Riccardo Albiero , Ciro Pentangelo , Francesco Ceccarelli , Giacomo Corrielli , Roberto Osellame
show 4 more authors
Nicolò Spagnolo Lorenzo Maccone Chiara Macchiavello Fabio Sciarrino
This is my paper · ORCID
classification quant-ph
keywords quantummetrologyphaseestimationanalog-to-digitalconverterdigitalmutualinformationdenoisingautoencoderintegratedphotoniccircuitsGHZstates
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper reports the first experimental realization of a quantum analog-to-digital converter (QADC) for phase estimation: a photonic circuit that takes a continuous optical phase $\phi$ and outputs a three-bit string encoding its binary expansion. It demonstrates that this digital quantum strategy, using a four-photon Greenberger–Horne–Zeilinger entangled state, a two-photon entangled pair, and a single photon, can surpass the standard quantum limit — the per-photon information bound of classical strategies — when performance is measured in recoverable bits, provided a denoising autoencoder first removes the experimental noise. The information-theoretic benchmark matters because quantum sensors that feed digital processors need to know how many significant bits they extract per resource, not just the variance of a continuous estimate. The paper also shows that a feed-forward neural network converts the denoised probabilities into an unbiased phase estimate over the full $2\pi$ range, removing the bias and periodicity that otherwise limit the scheme.

What carries the argument

The load-bearing object is the parallel quantum phase estimation circuit acting as a QADC: probes made of $k=1,2,4$ photons in GHZ states, with the phase encoded by $U(\phi)=|0\rangle\langle 0|+e^{i\phi}|1\rangle\langle 1|$, a CNOT chain realized by Hadamard gates plus parity measurements and $\sigma_z$ corrections, and an inverse quantum Fourier transform implemented without SWAPs through controlled rotations $R_l^{-1}=U(-2\pi/2^l)$. For three probes this yields the binary expansion of $\phi$ in three bits. The noise-mitigation machinery is a denoising autoencoder — an encoder-bottleneck-decoder network with layers of 64, 32, 16, 32, and 64 neurons — trained to map Gaussian-corrupted ideal probabilities back to the clean ideal distribution; it does the work of projecting the real, noisy experimental frequencies onto the ideal-model manifold. The final component is a feed-forward neural network that takes the denoised conditional probabilities, augmented by the probabilities at a fixed shifted phase $\phi+\delta\phi$, and outputs a continuous, unbiased phase estimate.

What would settle it

Retrain the denoising autoencoder with noise generated not from Gaussian corruption but from the independently measured physical error parameters reported in the paper — photon-pair Hong-Ou-Mandel visibilities between about 0.908 and 0.940, $g^{(2)}(0)\approx 5.3\times10^{-3}$, and circuit programming fidelity 0.995 — and recompute the mutual-information curves; if the quantum advantage over the classical strategy disappears or shrinks below the reported level under this physically grounded training set, the central claim rests on the Gaussian-noise assumption rather than on the protocol.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that a quantum phase-estimation circuit can be physically realized as a quantum analog-to-digital converter and can outperform classical interferometry in the number of recovered bits. The device consumes seven detected photons per conversion — a four-photon GHZ state, a two-photon entangled state, and a single photon — and outputs a bit string $\mathbf{b}=(b_1,b_2,b_3)$ such that $\phi = 2\pi(b_1/2 + b_2/4 + b_3/8)$. The paper reports that after a denoising autoencoder (trained on the ideal protocol's conditional probabilities corrupted with Gaussian noise) is applied to the measured seven-bit outcome probabilities, the mutual information $I(\mathbf{m}:\phi)$ of the quantum strategy exceeds that of a classical strategy using seven single photons in a standard interferometer, at every repetition count explored up to $n_{\mathrm{shots}}=5377$. A second neural network maps the denoised probabilities to a phase estimate, correcting the oscillatory bias of the raw digital estimate and giving unambiguous phase resolution across the full $[0,2\pi)$ interval, whereas the classical single-photon strategy resolves only $[0,\pi)$. These results are offered as the first experimental evidence that digital quantum estimation can surpass the standard quantum limit in the sense of bits recovered per photon.

Load-bearing premise

The load-bearing premise is that the real experimental noise looks to the denoising network like bell-shaped random noise added to the ideal probability distribution; if the actual noise has systematic structure (drift, partial distinguishability, source-intensity fluctuations) that the training distribution does not cover, the denoised probabilities and the reported mutual-information advantage would be artifacts of the training target rather than measured properties of the protocol.

Editorial extensions

If this is right

  • Digital quantum estimation is no longer only a theoretical benchmark: a programmable eight-mode integrated interferometer with a quantum-dot photon source is enough to implement a QADC and beat the classical per-photon bit rate.
  • Because the QADC output is a bit string, phase information can be handed directly to classical digital processors or to downstream quantum algorithms that expect a digital phase encoding, without an intermediate analog readout.
  • The denoising step restores the quantum advantage without adding physical resources, so the honest resource cost of the demonstrated advantage is seven detected photons per conversion plus the classical training of the neural networks.
  • The quantum probes distinguish phases across the entire $[0,2\pi)$ range without the periodicity ambiguity of the classical interferometric estimate, which matters for absolute rather than incremental phase sensing.
  • Extending the same architecture to more probes should increase the number of recovered bits, tracking the $\log_2 N$ quantum scaling of digital estimation with total photon number $N$, as long as the noise model used for training remains representative.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The claimed advantage is relative to one specific classical strategy — unentangled single photons processed by standard interferometry; whether the QADC also surpasses adaptive or otherwise optimized classical phase-estimation schemes is not established by this experiment, and testing against those baselines is a natural next step.
  • The training recipe of corrupting ideal probabilities with Gaussian noise is transferable in principle, but its validity is tied to the actual noise statistics of the device; a useful extension is to build the training set from the independently measured physical error sources and check that the advantage survives that harder test.
  • Because the protocol is post-selected on correlations between controlled operations and measurement outcomes, the conversion efficiency (accepted versus emitted photons) is an additional cost that the mutual-information-per-detected-photon figure does not capture.
  • The uniform-prior benchmark used for mutual information is natural but not universal; for a concrete sensing task with a different prior over phases, the same protocol would need to be re-evaluated under that prior to translate bits into a precision advantage.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper reports an experimental implementation of a digital phase estimation protocol, termed a Quantum Analog-to-Digital Converter (QADC), on a reconfigurable integrated photonic platform. The protocol uses a 4-photon GHZ state, a 2-photon entangled state, and a single-photon state to encode an unknown phase into three bits, and it is benchmarked against a classical interferometric strategy using the mutual information between the measurement outcome and the phase. Raw experimental data show a quantum advantage over the classical strategy at moderate numbers of phase repetitions, but the advantage diminishes as the number of repetitions grows. The authors then apply a denoising autoencoder (DAE) trained on ideal, noise-free conditional probabilities corrupted by Gaussian noise, and report that after denoising the quantum advantage persists 'regardless of the number of phase repetitions'. The abstract claims that the protocol is 'capable of surpassing the standard quantum limit'.

Significance. If the central claim is validated, this would be the first experimental realization of a quantum analog-to-digital converter for phase estimation and a demonstration of a quantum advantage in a digital, information-theoretic metrology setting. The experimental platform is state of the art, with careful characterization of state fidelities (0.997/0.948/0.811), HOM visibilities (0.91–0.94), g(2) values, and calibration fidelity (0.995). The paper also makes explicit the theoretical framework of digital quantum estimation and its relation to the standard quantum limit. However, the headline advantage rests on the validity of the DAE denoising step and on the fairness of the resource comparison; as presented, these are not established, so the significance of the claimed result is currently contingent.

major comments (4)
  1. [Experimental results (DAE training) and SI S1] The DAE is trained using ideal conditional probabilities corrupted with Gaussian noise, with targets set to the clean, noise-free probabilities, and is then applied to experimental data. The SI (Figs. S1–S2) identifies the dominant experimental noise sources as partial photon distinguishability, multiphoton components, and circuit-programming errors, which are structured and non-Gaussian. Because the training target is the ideal model, the DAE will tend to project any input toward that ideal manifold. The 'regardless of the number of phase repetitions' advantage in Fig. 4b is therefore a property of the training target rather than of the measured statistics. The paper does not validate the DAE on data generated from the physically calibrated noise model, so it is not established that the DAE removes real noise instead of manufacturing an apparent advantage.
  2. [Discussion and Fig. 4b; SI S2 A] The claim of using 'the same amount of resources (photons and phase shift applications)' is not supported by the experimental procedure. The quantum protocol requires post-selection: as described in SI S2 A, runs are discarded when the controlled operations do not match the parity of the measured bits, and the reported nshots correspond to the number of valid (post-selected) repetitions. The classical strategy has no analogous post-selection. With a 4-photon coincidence rate of ~8 Hz versus ~5 MHz for single photons, the number of input photons consumed per successful quantum trial is far larger than seven. The paper should either count all input photons (including discarded ones) or explicitly justify why counting only successful trials is the appropriate resource measure for surpassing the standard quantum limit.
  3. [Abstract and Fig. 4b; SI Fig. S1] The paper claims to surpass the standard quantum limit, but the experimental comparison in Fig. 4b is made against a specific classical interferometric strategy, not against the SQL bound itself. The SI (Fig. S1) shows that the asymptotic mutual information of this classical strategy lies below the SQL line. Thus, an advantage over this particular classical strategy does not establish that the quantum protocol surpasses the standard quantum limit. The authors should plot the SQL bound and demonstrate that the measured (raw or denoised) quantum mutual information exceeds it, or compare against an optimally chosen classical strategy.
  4. [Fig. 4b and SI S2 C] The mutual information curves in Fig. 4b are presented without uncertainty estimates. The denoised probabilities are outputs of a neural network, so their statistical behavior is not described by the Poissonian counting statistics used in SI S2 C for the raw-data estimator. The claim that the advantage holds 'regardless of the number of phase repetitions' requires a bootstrap or similar analysis that propagates raw-data fluctuations through the DAE, to rule out that the apparent advantage is within statistical error.
minor comments (4)
  1. [Protocol section] The notation for the resource states is inconsistent: the text refers to |GHZ4> and |GHZ2>, while the experimental implementation generates the dual-rail equivalents |0101>+|1010> and |01>+|10>. Please clarify the equivalence explicitly in the main text.
  2. [Fig. 4b caption] The shaded grey region in Fig. 4b is mentioned in the caption but not explained. Please add a sentence stating why the mutual information is not meaningful in that region.
  3. [Discussion] The phrase 'overcoming the mutual information bound of the classical strategies' is ambiguous: it could mean the fundamental SQL or the particular classical strategy implemented. Please specify which bound is overcome in the experimental comparison.
  4. [SI S2 C] The mutual information estimator described is the plug-in estimator, which is known to be biased for finite nshots. Please state that the bootstrapping procedure is applied to the difference between quantum and classical mutual information, not just to each separately.

Circularity Check

1 steps flagged · score 6.0 of 10

The claimed quantum mutual-information advantage at all nshots is carried by a DAE trained on the ideal quantum probabilities, so the advantage is largely a projection of the experimental data onto the training target rather than an independent measurement.

  1. fitted input called prediction [Experimental results, paragraph beginning 'For our application, we train the DAE...' and Fig. 4b; Discussion]
    "For our application, we train the DAE using ideal conditional probabilities corrupted with Gaussian noise, with the targets set as the clean, noise-free probabilities. ... Once trained, the DAE is applied to the experimental probabilities p(m|phi), reconstructed from our raw data, obtaining their denoised version ... After using the denoised probabilities to compute the mutual information, the advantage of the quantum strategy becomes evident regardless of the number of phase repetitions considered."

    The DAE is optimized to output the ideal, noise-free quantum conditional probabilities. The headline advantage is computed from the mutual information of these denoised probabilities. Since the ideal quantum distribution is already known (and used as the training target) to have higher mutual information than the classical one, the post-denoising 'regardless of repetitions' advantage is a consequence of the training target, not of the raw measured statistics. The paper explicitly states that the raw-data advantage diminishes as nshots increases, and the restored advantage appears only after the DAE is applied.

full rationale

The central experimental claim of surpassing the classical mutual-information bound rests on the denoising autoencoder stage. The paper states that the DAE is trained with ideal conditional probabilities corrupted with Gaussian noise and with clean noise-free probabilities as targets, and that the denoised probabilities are then used to compute the mutual information, yielding an advantage at all nshots. This is a fitted-input-called-prediction pattern: the network is explicitly fitted to the ideal quantum model, and the predicted advantage is read off from its output. The raw-data results, by the paper's own admission, show the quantum advantage diminishing with phase repetitions, so the restored advantage is not independently measured but is produced by the DAE. Because the SI lists physical noise sources that are structured and non-Gaussian, the Gaussian-only training does not validate that the DAE removes real noise; applying it to experimental data effectively projects those data toward the ideal theoretical distribution used as the training target. The theoretical framework and the experimental platform are otherwise self-contained, and the self-citations to the digital-estimation literature are not load-bearing in a circular way. However, the central 'overcoming the mutual information bound' result after denoising reduces by construction to the ideal-model property encoded in the DAE targets, warranting a score of 6.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the following unpaid inputs: (i) the digital-estimation benchmark (MI figure of merit, 1/2 log2 N vs log2 N bounds) taken from the authors' own refs [14,21,22]; (ii) the ideal protocol distributions used as DAE training targets; (iii) the Gaussian noise model for corruption, whose validity is assumed rather than demonstrated on real noise; (iv) post-selected resource accounting that ignores the orders-of-magnitude gap between the 4-fold and single-photon rates. No free parameters enter the phase-estimation formula itself (b1b2b3 expansion), but the ML pipeline has unreported or hand-chosen knobs: the Gaussian corruption level, the NN training-noise parameters, and delta_phi=0.44 rad. No new physical entities are postulated.

free parameters (3)
  • Gaussian corruption level for DAE training = not stated
    The DAE is trained on ideal probabilities 'corrupted with Gaussian noise'; the corruption amplitude controls the denoising strength and is not reported. If tuned so that denoised curves match the ideal model, it becomes a fitted knob for the headline result.
  • Noise model parameters for NN training data = not stated
    The feed-forward NN is trained on simulated outcomes with 'noise and variations consistent with experimental conditions'; the specific levels are not given, so the reader cannot tell whether the noise model was adjusted to make the experimental estimates look good.
  • Auxiliary phase shift delta_phi = 0.44 rad
    A fixed, arbitrary shift appended as auxiliary input to the NN to resolve boundary ambiguities; a hand-chosen hyperparameter (stated as 'arbitrary and fixed').
assumptions (5)
  • domain assumption Mutual information is the correct figure of merit for digital estimation, with classical and quantum bounds 1/2 log2 N and log2 N (refs [14,21,22]).
    The entire benchmarking framework comes from the authors' own theory papers; the experiment assumes this framework rather than testing it.
  • domain assumption The ideal conditional probabilities used as DAE training targets are computed from the noiseless protocol and are the right ground truth.
    Used in 'Experimental results': training targets are 'the clean, noise-free probabilities' of the ideal model.
  • ad hoc to paper Real experimental noise on outcome probabilities is well approximated by additive Gaussian corruption of the ideal probabilities.
    The weakest load-bearing premise; without it the DAE output (and the restored quantum advantage) is not a faithful deconvolution.
  • domain assumption Post-selected resource accounting is a valid way to compare quantum and classical strategies on 'the same amount of resources'.
    The 4-photon GHZ branch is created post-selected with ~8 Hz rate; counting only detected probes ignores the discarded runs that also consumed photons and phase shifts.
  • standard math Standard linear-optics identities: parity measurement on GHZ plus sigma_z equals CNOT chain; QFT without SWAPs yields the bit-by-bit readout.
    Quoted in the Protocol section; standard textbook results (ref [30]).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum Analog-to-Digital Converter for Phase Estimation." pith.science (2026). https://pith.science/paper/DVCNMO6N

@misc{pith2026250207676,
  author       = {Pith},
  title        = {Pith review of: Quantum Analog-to-Digital Converter for Phase Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DVCNMO6N}},
  note         = {Machine review of arXiv:2502.07676}
}
read the original abstract

Traditional quantum metrology assesses precision using the figures of merit of continuous-valued parameter estimation. Recently, quantum digital estimation was introduced: it evaluates the performance information-theoretically by quantifying the number of significant bits of the parameter, redefining key benchmarks like the Heisenberg bound. Here, we report the first experimental realization of a Quantum Analog-to-Digital Converter for quantum metrology, that takes an input continuous parameter and outputs a bit string, using an advanced photonic platform, comprising a fully reconfigurable integrated circuit and a quantum dot source of highly indistinguishable photons. We implement a protocol for digital phase estimation that is capable of surpassing the standard quantum limit, through the simultaneous use of different entangled state resources. We tackle experimental imperfections by employing machine learning techniques for noise deconvolution and estimation process refinement. Our protocol is experimentally benchmarked against classical strategies via the number of recoverable bits of the unknown parameter. Our results open new perspectives for future implementation of quantum digital estimation strategies.

Figures

Figures reproduced from arXiv: 2502.07676 by the authors.

Figure 1
Figure 1. QADC phase estimation protocol. a Quantum digital phase estimation protocol consisting of three stages involving 4, 2 and 1 qubit states, respectively. A unitary operation encodes the phase ϕ in each mode, followed by Hadamard gates and a σz oper￾ation, conditioned on the measurement outcomes on the first modes. Depending on the results obtained in the 4-qubit and 2-qubit steps, controlled phase gates R −1 l are tak… view at source ↗
Figure 2
Figure 2. Experimental setup for the QADC. a, Schematic view of the experimental setup comprising the quantum-dot source, a time￾to-spatial demultiplexing module, coupling to the integrated circuit, and the detection stage. Circuit programming and data acquisition are orchestrated by a processing unit. Inset: internal structure of the integrated device, comprising 28 fully reconfigurable Mach-Zehnder inter￾ferometer, acting a… view at source ↗
Figure 3
Figure 3. Calibration of the measurement outcomes. a-c, Plots of the experimental tomographies for the 1-, 2-, 4-qubit states employed for the quantum digital protocol. d-f, Plots of the experimentally reconstructed marginal probabilities on the informative qubits, for the 1-, 2- and 4-photon post-selected measurement outcomes of the digital estimation scheme. the phase estimation process. Specifically, this can be done by tr… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Denoising autoencoder and feed-forward neural network architectures together with experimental results. a, Architecture of the denoising autoencoder (DAE) implemented to mitigate the influence of noise on the conditional probabilities p˜(⃗m|ϕ), reconstructed from exper…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 48 canonical work pages

  1. [1]

    J. M. Robinson, M. Miklos, Y . M. Tso, C. J. Kennedy, T. Both- well, D. Kedar, J. K. Thompson, and J. Ye, Direct comparison of two spin-squeezed optical clock ensembles at the 10- 17 level, Nature Physics 20, 208 (2024)

    work page 2024

  2. [3]

    When the parity gives 0, the fourth photon is in (|0⟩ +eı4ϕ|1⟩)/ √ 2, otherwise apply σz, obtaining the same result

    Ap- ply Hadamard gates to the first three photons, then measure their parity ˆP in the computational basis, which reads 1 or 0 for bit strings that contain an odd or even numbers of 1s, re- spectively. When the parity gives 0, the fourth photon is in (|0⟩ +eı4ϕ|1⟩)/ √ 2, otherwise apply σz, obtaining the same result. This procedure, which can be implement...

    work page

  3. [4]

    When this returns 0, the other pho- ton is projected into (|0⟩ +eı2ϕ|1⟩)/ √ 2, otherwise applyσz

    Apply a Hadamard gate to the first photon, followed by a computa- tional basis measurement. When this returns 0, the other pho- ton is projected into (|0⟩ +eı2ϕ|1⟩)/ √ 2, otherwise applyσz. We implement the QFT† adding a rotationR−1 2 , controlled by the value ofb3 and followed by a Hadamard gate. A measure- ment in the computational basis yieldsb2. 1-pho...

    work page

  4. [5]

    Finally, apply a Hadamard gate and a measurement in the computa- tional basis

    Apply two subsequent rotations, R−1 2 andR−1 3 , controlled by b2 andb3, respectively. Finally, apply a Hadamard gate and a measurement in the computa- tional basis. This gives b1, the most significant bit of the bi- nary expansion ofϕ. The classical counterpart of our protocol consists in an entanglement-free interferometric estimation. Consider 2t− 1 qu...

    work page

  5. [6]

    For each qubit, p(|1⟩|ϕ) = sin ( ϕ2/2 )

    Apply a Hadamard gate, followed by a computational basis measurement. For each qubit, p(|1⟩|ϕ) = sin ( ϕ2/2 ) . The optical phase reads ϕ = 2 arcsin(N1/(2t− 1)), where N1 is the number of 1s in the output bit string. Experimental implementation The experimental apparatus, for the implementation of the digital estimation algorithm, is based on a photonic h...

    work page 2000

  6. [7]

    B. C. Nichol, R. Srinivas, D. Nadlinger, P. Drmota, D. Main, G. Araneda, C. Ballance, and D. Lucas, An elementary quantum network of entangled optical atomic clocks, Nature 609, 689 (2022)

    work page 2022

  7. [8]

    Pedrozo-Pe ˜nafiel, S

    E. Pedrozo-Pe ˜nafiel, S. Colombo, C. Shu, A. F. Adiyatullin, Z. Li, E. Mendez, B. Braverman, A. Kawasaki, D. Akamatsu, Y . Xiao, and V . Vuletic, Entanglement on an optical atomic- clock transition, Nature 588, 414 (2020)

    work page 2020

  8. [9]

    Appel, P

    J. Appel, P. J. Windpassinger, D. Oblak, U. B. Hoff, N. Kjær- gaard, and E. S. Polzik, Mesoscopic atomic entanglement for precision measurements beyond the standard quantum limit, Proceedings of the National Academy of Sciences 106, 10960 (2009)

    work page 2009

Show all 54 references
  1. [10]

    Craigie, E

    K. Craigie, E. Gauger, Y . Altmann, and C. Bonato, Resource- efficient adaptive bayesian tracking of magnetic fields with a quantum sensor, Journal of Physics: Condensed Matter 33, 195801 (2021)

    2021

  2. [11]

    Bonato and D

    C. Bonato and D. W. Berry, Adaptive tracking of a time-varying field with a quantum sensor, Physical Review A 95, 052348 (2017). 7

    2017

  3. [12]

    Liu, Y .-Z

    L.-Z. Liu, Y .-Z. Zhang, Z.-D. Li, R. Zhang, X.-F. Yin, Y .-Y . Fei, L. Li, N.-L. Liu, F. Xu, Y .-A. Chen, and J.-W. Pan, Distributed quantum phase estimation with entangled photons, Nature Pho- tonics 15, 137 (2021)

    2021

  4. [13]

    Liu, Y .-B

    Y .-C. Liu, Y .-B. Cheng, X.-B. Pan, Z.-Z. Sun, D. Pan, and G.- L. Long, Quantum integrated sensing and communication via entanglement, Physical Review Applied 22, 034051 (2024)

    2024

  5. [14]

    B. K. Malia, Y . Wu, J. Mart ´ınez-Rinc´on, and M. A. Kase- vich, Distributed quantum sensing with mode-entangled spin- squeezed atomic states, Nature 612, 661 (2022)

    2022

  6. [15]

    D.-H. Kim, S. Hong, Y .-S. Kim, Y . Kim, S.-W. Lee, R. C. Pooser, K. Oh, S.-Y . Lee, C. Lee, and H.-T. Lim, Distributed quantum sensing of multiple phases with fewer photons, Nature communications 15, 266 (2024)

    2024

  7. [16]

    Malitesta, A

    M. Malitesta, A. Smerzi, and L. Pezz `e, Distributed quantum sensing with squeezed-vacuum light in a configurable array of mach-zehnder interferometers, Physical Review A108, 032621 (2023)

    2023

  8. [17]

    Giovannetti, S

    V . Giovannetti, S. Lloyd, and L. Maccone, Quantum metrology, Physical Review Letters 96, 010401 (2006)

    2006

  9. [18]

    Giovannetti, S

    V . Giovannetti, S. Lloyd, and L. Maccone, Advances in quan- tum metrology, Nature Photonics 5, 222 (2011)

    2011

  10. [19]

    Hassani, C

    M. Hassani, C. Macchiavello, and L. Maccone, Digital quantum estimation, Physical Review Letters 119, 200502 (2017)

    2017

  11. [20]

    Paris and J

    M. Paris and J. Rehacek, Quantum state estimation , V ol. 649 (Springer Science & Business Media, 2004)

    2004

  12. [21]

    Cimini, E

    V . Cimini, E. Polino, F. Belliardo, F. Hoch, B. Piccirillo, N. Spagnolo, V . Giovannetti, and F. Sciarrino, Experimental metrology beyond the standard quantum limit for a wide re- sources range, npj Quantum Information 9, 20 (2023)

    2023

  13. [22]

    B. L. Higgins, D. W. Berry, S. D. Bartlett, H. M. Wiseman, and G. J. Pryde, Entanglement-free heisenberg-limited phase estimation, Nature 450, 393 (2007)

    2007

  14. [23]

    Slussarenko, M

    S. Slussarenko, M. M. Weston, H. M. Chrzanowski, L. K. Shalm, V . B. Verma, S. W. Nam, and G. J. Pryde, Unconditional violation of the shot-noise limit in photonic quantum metrology, Nature Photonics 11, 700 (2017)

    2017

  15. [24]

    S. Zhou, M. Zhang, J. Preskill, and L. Jiang, Achieving the heisenberg limit in quantum metrology using quantum error correction, Nature Communications 9, 78 (2018)

    2018

  16. [25]

    Valeri, E

    M. Valeri, E. Polino, N. Spagnolo, and F. Sciarrino, Photonic quantum metrology, A VS Quantum Science2, 024703 (2020)

    2020

  17. [26]

    X. Lu, W. G ´orecki, C. Macchiavello, and L. Maccone, Number of bits returned by a quantum estimation, Physical Review A 110, 032405 (2024)

    2024

  18. [27]

    G ´orecki, X

    W. G ´orecki, X. Lu, C. Macchiavello, and L. Maccone, Mutual information bounded by fisher information (2024), arXiv:2403.10248 [quant-ph]

    2024 arXiv

  19. [28]

    J. Wang, F. Sciarrino, A. Laing, and M. G. Thompson, Inte- grated photonic quantum technologies, Nature Photonics 14, 273 (2020)

    2020

  20. [29]

    Pelucchi, G

    E. Pelucchi, G. Fagas, I. Aharonovich, D. Englund, E. Figueroa, Q. Gong, H. Hannes, J. Liu, C.-Y . Lu, N. Matsuda, J.-W. Pan, F. Schreck, F. Sciarrino, C. Silberhorn, J. Wang, and K. D. J¨ons, The potential and global outlook of integrated photonics for quantum technologies, N...

    2022

  21. [30]

    Cleve, A

    R. Cleve, A. Ekert, C. Macchiavello, and M. Mosca, Quantum algorithms revisited, Proceedings of the Royal Society A 454, 339 (1998)

    1998

  22. [31]

    Cimini, I

    V . Cimini, I. Gianani, N. Spagnolo, F. Leccese, F. Sciarrino, and M. Barbieri, Calibration of quantum sensors by neural net- works, Physical Review Letters 123, 230502 (2019)

    2019

  23. [32]

    Cimini, E

    V . Cimini, E. Polino, M. Valeri, I. Gianani, N. Spagnolo, G. Corrielli, A. Crespi, R. Osellame, M. Barbieri, and F. Sciar- rino, Calibration of multiparameter sensors via machine learn- ing at the single-photon level, Physical Review Applied 15, 044003 (2021)

    2021

  24. [33]

    Cimini, M

    V . Cimini, M. Valeri, E. Polino, S. Piacentini, F. Ceccarelli, G. Corrielli, N. Spagnolo, R. Osellame, and F. Sciarrino, Deep reinforcement learning for quantum multiparameter estimation, Advanced Photonics 5, 016005 (2023)

    2023

  25. [34]

    Cimini, M

    V . Cimini, M. Valeri, S. Piacentini, F. Ceccarelli, G. Corrielli, R. Osellame, N. Spagnolo, and F. Sciarrino, Variational quan- tum algorithm for experimental photonic multiparameter esti- mation, npj Quantum Information 10, 26 (2024)

    2024

  26. [35]

    M. A. Nielsen and I. L. Chuang, Quantum computation and quantum information (Cambridge university press, 2010)

    2010

  27. [36]

    Somaschi, V

    N. Somaschi, V . Giesz, L. De Santis, J. Loredo, M. P. Almeida, G. Hornecker, S. L. Portalupi, T. Grange, C. Anton, J. Demory, C. Gomez, I. Sagnes, N. D. Lanzillotti-Kimura, A. Lemaitre, A. Auffeves, A. G. White, L. Lanco, and P. Senellart, Near- optimal single-photon sources ...

    2016

  28. [37]

    Gazzano, S

    O. Gazzano, S. Michaelis de Vasconcellos, C. Arnold, A. Nowak, E. Galopin, I. Sagnes, L. Lanco, A. Lema ˆıtre, and P. Senellart, Bright solid-state sources of indistinguishable sin- gle photons, Nature Communications 4, 1425 (2013)

    2013

  29. [38]

    M. Pont, R. Albiero, S. E. Thomas, N. Spagnolo, F. Ceccarelli, G. Corrielli, A. Brieussel, N. Somaschi, H. Huet, A. Harouri, A. Lemaitre, I. Sagnes, N. Belabas, F. Sciarrino, R. Osel- lame, P. Senellart, and A. Crespi, Quantifying n-photon indis- tinguishability with a cyclic ...

    2022

  30. [40]

    Pentangelo, F

    C. Pentangelo, F. Ceccarelli, S. Piacentini, R. Albiero, E. Urbinati, N. Di Giano, S. Atzeni, A. Crespi, and R. Osellame, Universal photonic processors fabricated by femtosecond laser writing, in Integrated Optics: Devices, Materials, and Tech- nologies XXVI, V ol. 12004 (SPIE...

    2022

  31. [41]

    Pentangelo, N

    C. Pentangelo, N. Di Giano, S. Piacentini, R. Arpe, F. Cec- carelli, A. Crespi, and R. Osellame, High-fidelity and polarization-insensitive universal photonic processors fabri- cated by femtosecond laser writing, Nanophotonics 13, 2259 (2024)

    2024

  32. [42]

    W. R. Clements, P. C. Humphreys, B. J. Metcalf, W. S. Kolthammer, and I. A. Walmsley, Optimal design for universal multiport interferometers, Optica 3, 1460 (2016)

    2016

  33. [43]

    P. Kok, W. J. Munro, K. Nemoto, T. C. Ralph, J. P. Dowling, and G. J. Milburn, Linear optical quantum computing with photonic qubits, Reviews of Modern Physics 79, 135 (2007)

    2007

  34. [44]

    M. Pont, G. Corrielli, A. Fyrillas, I. Agresti, G. Carvacho, N. Maring, P.-E. Emeriau, F. Ceccarelli, R. Albiero, P. H. Ferreira, N. Somaschi, J. Senellart, I. Sagnes, M. Morassi, A. Lemaitre, P. Senellart, F. Sciarrino, M. Liscidini, N. Belabas, and R. Osellame, High-fidelity...

    2024

  35. [45]

    Vincent, H

    P. Vincent, H. Larochelle, Y . Bengio, and P.-A. Manzagol, Ex- tracting and composing robust features with denoising autoen- coders, in Proceedings of the 25th international conference on Machine learning (2008) pp. 1096–1103

    2008

  36. [46]

    Bajaj, D

    K. Bajaj, D. K. Singh, and M. A. Ansari, Autoencoders based deep learner for image denoising, Procedia Computer Science 171, 1535 (2020). 8

    2020

  37. [47]

    Sakurada and T

    M. Sakurada and T. Yairi, Anomaly detection using autoen- coders with nonlinear dimensionality reduction, inProceedings of the MLSDA 2014 2nd workshop on machine learning for sen- sory data analysis (2014) pp. 4–11

    2014

  38. [48]

    X. Lu, Y . Tsao, S. Matsuda, and C. Hori, Speech enhancement based on deep denoising autoencoder, inInterspeech, V ol. 2013 (2013) pp. 436–440

    2013

  39. [49]

    Mitarai, M

    K. Mitarai, M. Kitagawa, and K. Fujii, Quantum analog-digital conversion, Physical Review A 99, 012301 (2019)

    2019

  40. [50]

    Schm ¨user and D

    F. Schm ¨user and D. Janzing, Quantum analog-to-digital and digital-to-analog conversion, Physical Review A 72, 042324 (2005)

    2005

  41. [51]

    R. H. Brown and R. Twiss, Lxxiv. a new type of interferometer for use in radio astronomy, The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 45, 663–682 (1954)

    1954

  42. [52]

    Ollivier, S

    H. Ollivier, S. E. Thomas, S. C. Wein, I. M. de Buy Wenniger, N. Coste, J. C. Loredo, N. Somaschi, A. Harouri, A. Lemaitre, I. Sagnes, L. Lanco, C. Simon, C. Anton, O. Krebs, and P. Senellart, Hong-Ou-Mandel interference with imperfect sin- gle photon sources, Physical Review ...

    2021 arXiv

  43. [53]

    Rodari, L

    G. Rodari, L. Novo, R. Albiero, A. Suprano, C. T. Tavares, E. Caruccio, F. Hoch, T. Giordani, G. Carvacho, M. Gardina, N. Di Giano, S. Di Giorgio, G. Corrielli, G. Ceccarelli, R. Osellame, N. Spagnolo, E. F. Galv ¨ao, and F. Sciarrino, Semi-device independent characterization ...

    2024 arXiv

  44. [54]

    Ollivier, S

    H. Ollivier, S. E. Thomas, S. C. Wein, I. M. de Buy Wenniger, N. Coste, J. C. Loredo, N. Somaschi, A. Harouri, A. Lemaitre, I. Sagnes, L. Lanco, C. Simon, C. Anton, O. Krebs, and P. Senellart, Hong-Ou-Mandel interference with imperfect single photon sources, Physical Review Le...

    2021

  45. [55]

    M. Pont, R. Albiero, S. E. Thomas, N. Spagnolo, F. Ceccarelli, G. Corrielli, A. Brieussel, N. Somaschi, H. Huet, A. Harouri, A. Lemaitre, I. Sagnes, N. Belabas, F. Sciarrino, R. Osellame, P. Senellart, and A. Crespi, Quantifying n-photon indistinguishability with a cyclic inte...

    2022

  46. [56]

    Oszmaniec and D

    M. Oszmaniec and D. J. Brod, Classical simulation of photonic linear optics with lost particles, New Journal of Physics20, 092002 (2018)

    2018

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.