{"id":"10744003-5a8d-482c-ad95-ddfe944f6783","arxiv_id":"2506.20627","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A stabilized GKP qunaught state used with small-big-small feedback estimates two quadrature displacements nearly at the quantum Cramer-Rao bound and beats Gaussian sensing limits under realistic noise.","lead":"This paper shows that a stabilized GKP state can act as a continuously reusable displacement sensor, estimating both position and momentum shifts in one shot with precision near the fundamental quantum limit. It also shows numerically that such a sensor can beat the best Gaussian-state strategies even with realistic decoherence.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Noisy Gaussian-prior advantage relies on an unverified quadrature-decoupling approximation; a full two-quadrature Fisher-information check could change the conclusion.","rationale":"The paper's strongest claim is the numerical demonstration that the sBs protocol approaches the multivariate QCRB and beats the Gaussian limit with realistic noise. The most load-bearing vulnerability is not the noiseless center-point sensitivity, which is computed exactly at p0=0 and is robust to decoupling, but the Gaussian-prior MSE results (Figs. 4 and 6) that support the 'unconditionally surpass' claim. These results are obtained by constructing estimators from p(b^q|q0,0) while assuming the q-quadrature bitstring statistics are independent of p0. This assumption is stated explicitly in Appendix D and is verified only partially in Appendix E.3 via sensitivity checks at a few nonzero p0 values, not via a full prior-averaged joint analysis. Under noise, qubit relaxation and cavity dephasing can break the symmetry that justifies the decoupling, and the outcome-dependent feedback may propagate errors between quadratures. If the true likelihood depends on p0, the Bayesian estimators are miscalibrated and the reported MSE could be optimistic. The proposed Fisher-information test is direct and feasible: it computes the exact joint likelihood for a moderate number of bits and compares the information content and cross-coupling against the single-quadrature approximation. This settles whether the decoupling approximation is quantitatively safe in the noisy regime that underlies the headline claim. I credit the paper for providing complete code, clear derivations of the QCRB and Holevo bound, and for identifying the approximation in the appendix; the issue is not an internal inconsistency but an unverified numerical assumption in the regime where the strongest claim is made. The reader already judged the paper CONDITIONAL, and this concern reinforces that judgment without moving it to rejection, so I recommend keeping the verdict unchanged.","tokens_in":35351,"tokens_out":12222,"duration_ms":152281,"concrete_test":"Compute the exact 2x2 classical Fisher information matrix of the full 2T-bit outcome distribution for T=8 at several representative displacements (e.g., (0,0), (0.1l,0.1l), (0.2l,0.2l)) using the noisy Kraus operators with η=1, Δ=0.3, including the Lindblad noise during big displacements and enumerating all 2^16 outcomes (feasible with the sBs basis of Ref. [91] or by trajectory sampling). Compare the diagonal entry F_qq(q0,p0) with the single-quadrature value F_qq(q0,0) and check the normalized off-diagonal |F_qp|/√(F_qq F_pp). If either deviates by more than ~10%, the decoupling approximation used to construct the estimators is violated and the MSE results in Fig. 6 are not representative of the full two-quadrature problem.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The core claim that the protocol 'unconditionally surpasses the Gaussian limit with prior information' under realistic noise is backed by simulations that treat the two quadratures as independent: the Bayesian MSE for the q quadrature (Fig. 6b) is computed from p(b^q|q0,0), i.e., with the p displacement fixed to zero (Appendix D, Step 2). This is the same approximation the authors acknowledge in Appendix D for the noiseless case, but in the noisy case no analytic argument is given and the only check reported (Appendix E.3) compares sensitivities for a few nonzero p0 values, not the full prior-averaged MSE used in the central claim. If qubit relaxation or cavity dephasing couples the quadratures through the outcome-dependent feedback and error propagation, then the true marginal p(b^q|q0) = ∫dp0 Σ_{b^p} p(b^q,b^p|q0,p0) p(p0) can differ from p(b^q|q0,0), the estimators built from the latter become miscalibrated for p0≠0, and the actual joint MSE may be larger than the single-quadrature result. Because the noisy advantage over the Gaussian limit (Fig. 6b) is finite rather than asymptotic, even a modest coupling could erode or eliminate the claimed 'unconditional' beating. The noiseless local sensitivity at q0=p0=0 is not affected by this issue, but the Gaussian-prior and noise claims are.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes a metrology protocol that repurposes the small-big-small (sBs) stabilization protocol for finite-energy GKP qunaught states to estimate two quadrature displacements in a single shot. The authors show numerically, using Fock-basis simulations with code availability, that the noiseless sensitivity approaches the multivariate quantum Cramér-Rao bound (e.g., 13.4 dB vs 13.6 dB two-mode squeezing equivalent at 10 bits per quadrature), that the protocol is backaction evading, and that under a Lindblad noise model with experimentally measured lifetimes it can surpass the Gaussian limit of displacement sensing with prior information. The main claims are supported by direct simulation of the sBs Kraus operators; the analytical probability law in Eq. (1) is fitted and used only for qualitative discussion.","tokens_in":35667,"tokens_out":7475,"duration_ms":75343,"significance":"If the noisy advantage withstands the quadrature-decoupling approximation, the protocol would be a valuable example of reservoir engineering for quantum metrology, with potential applications in force sensing and waveform estimation. The paper provides reproducible numerical code and uses a concrete, experimentally grounded noise model. The noiseless saturation of the QCRB and the backaction-evading property are demonstrated convincingly. The main caveat is that the noisy advantage over the Gaussian limit is computed under a decoupling approximation that has not been fully validated in the noisy regime, and the noise model itself neglects errors in the small displacement and rotation operations.","major_comments":[{"comment":"The central noisy claim that the protocol 'unconditionally surpasses the Gaussian limit' (Section III D, Fig. 6b) is computed by approximating the q-quadrature marginal p(bq|q0) with p(bq|q0, p0=0), i.e., by fixing the p displacement to zero while averaging over p-outcomes (Appendix D, Step 2). In the noiseless case this decoupling is justified analytically (Appendix C), but no analogous justification is given for the noisy case. The only check reported (Appendix E.3) compares sensitivities at a few fixed nonzero p0 values, not the full prior-averaged Bayesian MSE used for the Gaussian-limit comparison. Since the noisy advantage is finite, any decoherence-induced coupling between the quadratures—through outcome-dependent feedback or correlated error propagation—could increase the true joint MSE and erase the claimed advantage. The authors should either validate the decoupling for the noisy regime by computing the full two-quadrature probabilities p(bq,bp|q0,p0) for representative noise parameters and evaluating the joint Bayesian MSE, or qualify the 'unconditional' claim.","section":"Section III D, Appendix D Step 2, Fig. 6b"},{"comment":"The realistic-noise simulations assume that all qubit rotations and small conditional displacements are noiseless, with Lindblad decoherence acting only during the big conditional displacements. While the authors argue that these operations are fast compared to decoherence timescales, the small displacements are executed many times per run (twice per bit), and errors in them could accumulate over the T rounds of the metrology protocol. Because the noisy performance in Fig. 6 is the basis for the claim of surpassing the Gaussian limit under 'realistic noise,' the simulation should either include these error channels (or provide an estimate of their effect) or the claim should be restricted to the stated idealized operations.","section":"Section III D"}],"minor_comments":[{"comment":"The phrase 'solid dashed lines' appears twice (referring to the Cramér-Rao bound curves in Fig. 3a); this should be simply 'dashed lines'.","section":"Section III B"},{"comment":"Since a1 and a2 are fitted parameters, the authors should state explicitly in the main text that the quantitative results (QCRB approach, noisy MSE) are obtained from exact Fock-basis simulations and do not depend on this fit.","section":"Equation (1)"},{"comment":"The Gaussian limit in Eq. (A1) is given for the total two-quadrature MSE, but Fig. 4 plots only the q-quadrature MSE; the text should explain how the single-quadrature Gaussian limit is derived from Eq. (A1) to avoid confusion.","section":"Figure 4 and Eq. (A1)"},{"comment":"The sentence 'the fidelities without recovery no longer are the same' should be rewritten for clarity (e.g., 'the fidelities obtained without the recovery displacement are no longer identical').","section":"Appendix E.5"},{"comment":"The discussion of the mean photon number inset (Fig. 2) appears in Section III A, but the sentence beginning 'As shown there, the mean photon number remains bounded' could benefit from an explicit reference to the inset to help the reader locate it.","section":"Section III A"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid protocol proposal with reproducible numerics. The main risk is the unvalidated quadrature-decoupling approximation in the noisy regime, which directly affects the headline claim of unconditionally surpassing the Gaussian limit. I would recommend that the authors be asked to provide a direct numerical check of the joint two-quadrature MSE under noise, or to reformulate the claim as conditional on the decoupling approximation. The self-citation to the sBs protocol (Ref. [45]) is appropriate and not a concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Best read this one before the rush. The paper shows that the sBs GKP stabilization protocol can be used as a continuous, backaction-evading two-quadrature displacement sensor. The headline numbers are good: at Delta=0.3 and 10 bits per quadrature, sensitivity is within 0.2 dB of the QCRB, and with realistic noise parameters it numerically beats the Gaussian limit without postselection or entanglement. The novelties are real: the continuous feedback version with two-quadrature estimation and the noisy violation of the Gaussian limit are not in the cited work, including Ref. [59] which targets shot-number scaling instead of single-shot sensitivity.\n\nThe main results are backed by explicit Fock-basis simulations, and the code is available. The appendices are careful, especially the derivation of the Holevo bound and the analysis of error sources. The discussion of state preparation versus acquisition noise is useful.\n\nThe soft spots are two. First, the noisy Gaussian-prior claim depends on the approximation p(bq|q0) ≈ p(bq|q0,0), i.e. that the q-quadrature bit probabilities are independent of p0. The paper justifies this by checking sensitivities at a few nonzero p0 values (Appendix E.3), but that is not a full marginalization. If qubit relaxation or cavity dephasing couples the quadratures through feedback, the true marginal over p0 could be different, and the finite MSE advantage could shrink. The authors state the assumption, but for a claim that says 'unconditionally surpass', they should either prove the decoupling analytically or compute the full two-quadrature distribution and show the prior-averaged MSE directly. This is an addressable issue, not a fundamental flaw.\n\nSecond, Eq. (1) contains fitted constants a1, a2. That's fine for intuition, but it is not a derivation. The main numerics don't rely on it, so this is minor. Also, the abstract's 'unconditionally' is a bit stronger than the demonstrated range of priors and noise levels; the paper is otherwise accurate about the range.\n\nOverall, the noiseless performance is well-supported, and the noisy result is plausible but not fully nailed down. This deserves a serious referee; I would send it to review and ask the authors to close the quadrature-decoupling gap before acceptance.","headline":"A solid, useful metrology proposal whose main noise claim rests on a quadrature-independence approximation that should be rigorously checked before publication.","tokens_in":36167,"tokens_out":4571,"would_cite":true,"duration_ms":48991,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the small-big-small GKP stabilization protocol, with measurement in place of reset, estimates both quadrature displacements of a single bosonic mode with single-shot sensitivity approaching the multivariate quantum…","keywords":["GKP codes","qunaught states","displacement sensing","multiparameter quantum metrology","quantum Cramer-Rao bound","backaction evading measurements","reservoir engineering","bosonic quantum error correction"],"falsifier":"Compute or measure the full joint bitstring distribution $p(b_q, b_p \\mid q_0, p_0)$ for $\\Delta = 0.25$ and $T \\geq 10$: if the product approximation $p(b_q \\mid q_0, 0)p(b_p \\mid 0, p_0)$ yields sensitivities that differ from the full simulation by more than the paper's reported margins, the claimed near-QCRB performance is called into question.","tokens_in":35122,"feed_emoji":"⚛️","tokens_out":6099,"duration_ms":64530,"temperature":0.7,"pith_summary":"The paper proposes using the small-big-small (sBs) stabilization protocol for GKP grid states as a two-parameter displacement sensor. Each qubit measurement in the stabilization cycle yields one bit about either the position or momentum displacement of an unknown kick, while the same cycle returns the cavity to the sensor state, so the device can run continuously without reset. In the noiseless case the sensitivity approaches the multivariate quantum Cramer-Rao bound: with an envelope parameter $\\Delta = 0.3$ and 10 bits per quadrature it reaches 13.4 dB of two-mode squeezing, 0.2 dB below the quantum limit. With realistic noise parameters the protocol still beats the Gaussian limit of displacement sensing with prior information, without postselection or entanglement. The authors claim this makes reservoir-engineered bosonic stabilization useful for force sensing, waveform estimation, and quantum channel learning.","feed_headline":"GKP sensor reads both quadratures nearly at the quantum limit","feed_subtitle":"A stabilized grid state keeps sensing after each shot, within 0.2 dB of the quantum bound.","key_machinery":"The central object is the small-big-small (sBs) stabilization protocol, a trotterized reservoir-engineering scheme whose unitaries $\\hat{U}_{q,\\Delta}$ and $\\hat{U}_{p,\\Delta}$ alternate small control displacements, qubit rotations, and one large control displacement. For metrology, the qubit reset is replaced by measurement and feedback, so the Kraus operators of each subround act as modular measurements of one quadrature while the small displacements provide stabilization. The paper shows that the averaged measurement probabilities take the approximate form $\\bar{p}_{g/e,x}(q_0, p_0, T) \\simeq \\tfrac{1}{2}[1 \\pm e^{-a_1\\Delta^2}\\sin(l c_\\Delta x_0 e^{-a_2\\Delta^2 (T-1)})]$, where the exponential damping encodes the backaction that erases the displacement while returning the state to the grid. Because the two quadrature measurements are nearly independent, the multivariate estimation problem separates into two single-parameter problems, and this separation is what lets the protocol approach the multivariate quantum Cramer-Rao bound.","core_discovery":"The central claim is that displacement estimation can be merged with GKP stabilization so that every measurement performs two jobs at once: extracting a bit of information about the initial displacement and pumping the oscillator back toward the qunaught grid state. Concretely, after an unknown displacement $(q_0, p_0)$ is applied to a finite-energy qunaught state, repeated sBs rounds with qubit measurement and feedback produce a bitstring whose likelihood is nearly separable between the two quadratures. The authors find that the resulting sensitivity approaches the quantum bound $1/\\sqrt{4\\bar{n}+2}$: for $\\Delta = 0.3$ and 10 bits per quadrature, the sensitivity equals that of two-mode squeezed vacuum states with 13.4 dB of squeezing, only 0.2 dB shy of the 13.6 dB quantum limit, and one bit per quadrature already beats coherent-state heterodyne detection. Under Gaussian priors the protocol surpasses the Gaussian limit with three to four bits per quadrature in the numerically explored envelope range, and two bits in the infinite-energy limit. With realistic noise modeled from measured superconducting-circuit lifetimes, the sensitivity approaches 10 dB and the Gaussian limit is still beaten unconditionally.","pith_inferences":["An extension the authors leave implicit is that pairing sBs bit acquisition with a faster, purely dissipative recovery step could improve the information rate per unit time while keeping the same single-shot sensitivity.","The photon-efficiency advantage over two-mode squeezed vacuum states suggests a concrete testable extension: multimode GKP states may give a constant-factor photon saving when learning random displacement channels, beyond the exponential advantage already known for entangled Gaussian strategies.","Because the sensor is backaction evading and continuously self-resetting, it is a natural candidate for weak stochastic waveform and amplitude detection, such as dark-matter searches, though the authors note that amplitude estimation remains open.","A direct experimental check of the near-product form $p(b_q, b_p \\mid q_0, p_0) \\simeq p(b_q \\mid q_0, 0) p(b_p \\mid 0, p_0)$ would test the load-bearing separation on which the estimator construction relies."],"forward_implications":["The sensor is backaction evading and can operate continuously without reset, making it suited to detecting itinerant signals such as fluctuating forces.","With 10 bits per quadrature the sensitivity reaches 13.4 dB of two-mode squeezing at $\\Delta = 0.3$, within 0.2 dB of the quantum limit.","Under Gaussian priors the protocol beats the Gaussian limit of displacement sensing without postselection or entanglement, even with realistic noise.","The scheme is platform independent and can be implemented in any qubit-oscillator system, including trapped ions and mechanical oscillators.","The results imply a lower bound on stabilization speed: a slower sBs protocol would allow the sensitivity to surpass the quantum bound, so metrology constrains how slowly the grid state may be stabilized."],"supporting_citations":[{"why":"Defines the small-big-small stabilization unitaries that the metrology protocol repurposes.","marker":"[45]"},{"why":"Introduces the GKP grid code and the qunaught state used as the sensor state.","marker":"[34]"},{"why":"Establishes the single-mode displacement-sensor framework and the quantum Cramer-Rao bound for two-quadrature estimation.","marker":"[53]"},{"why":"Gives the Gaussian limit of displacement estimation with prior information that the protocol claims to beat.","marker":"[54]"},{"why":"Supplies the optimal two-mode squeezed-vacuum benchmark expressed in dB of squeezing.","marker":"[48]"},{"why":"Provides Caves' definition of backaction evading measurement used to characterize the sensor's continuous operation.","marker":"[51]"},{"why":"Supplies the multiparameter quantum estimation formalism and the Holevo-bound discussion used to locate the protocol's performance.","marker":"[66]"},{"why":"Provides the experimentally measured coherence times used as the realistic noise reference in the numerical simulations.","marker":"[38]"}],"fun_headline_variants":["GKP grid state sensor nears quantum limit for displacements","Stabilized GKP beats Gaussian sensing limit unconditionally","Displacement sensing with GKP states: 0.2 dB from quantum bound","Backaction-evading GKP sensor reads both quadratures near bound"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results assume that the q and p quadrature measurements are nearly independent and that all noise acts only during the large conditional displacements, so if those separations fail at smaller $\\Delta$, longer $T$, or stronger noise, the claimed closeness to the quantum bound and the beating of the Gaussian limit would degrade.","fun_headline_variants_meta":{"raw":{"variants":["GKP grid state sensor nears quantum limit for displacements","Stabilized GKP beats Gaussian sensing limit unconditionally","Displacement sensing with GKP states: 0.2 dB from quantum bound","Backaction-evading GKP sensor reads both quadratures near bound"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000193,"raw_usage":{"total_tokens":1330,"prompt_tokens":902,"completion_tokens":428,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":518,"completion_tokens_details":{"reasoning_tokens":354}},"tokens_in":518,"tokens_out":428,"duration_ms":4847,"temperature":1.0,"reasoning_tokens":354,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:44:39.649806+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute or measure the full joint bitstring distribution $p(b_q, b_p \\mid q_0, p_0)$ for $\\Delta = 0.25$ and $T \\geq 10$: if the product approximation $p(b_q \\mid q_0, 0)p(b_p \\mid 0, p_0)$ yields sensitivities that differ from the full simulation by more than the paper's reported margins, the claimed near-QCRB performance is called into question.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the small-big-small stabilization unitaries that the metrology protocol repurposes."},{"cited_title":"Demkowicz-Dobrzanski, U","cited_arxiv_id":null,"evidence_quote":"Introduces the GKP grid code and the qunaught state used as the sensor state."},{"cited_title":"Lachance-Quirion, M.-A","cited_arxiv_id":null,"evidence_quote":"Gives the Gaussian limit of displacement estimation with prior information that the protocol claims to beat."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the optimal two-mode squeezed-vacuum benchmark expressed in dB of squeezing."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the multiparameter quantum estimation formalism and the Holevo-bound discussion used to locate the protocol's performance."},{"cited_title":"Demkowicz-Dobrza´ nski, J","cited_arxiv_id":null,"evidence_quote":"Provides the experimentally measured coherence times used as the realistic noise reference in the numerical simulations."}],"review_version":1}