{"id":"4188ca55-b848-489b-b73c-081b3b06652c","arxiv_id":"2505.15635","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A Hamiltonian reformulation of the SU(1,1) interferometer yields Heisenberg scaling for relative phase sensing at phi=pi and log-modified Heisenberg scaling for dynamical phase sensing at theta=0, with a shift-operator readout replacing total photon number.","lead":"A new Hamiltonian-based model of the SU(1,1) interferometer predicts different optimal sensing points and readouts than the standard circuit model. It shows Heisenberg-limited precision for relative phase and a logarithmically modified Heisenberg limit for dynamical phase, with an experimentally testable shift-operator readout.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-mode reduction at Eq. (6) assumes perfect mode indistinguishability (|z1|=|z2|=1 in Eq. (5)); all QFI scalings and the optimal readout are properties of this measure-zero alignment, with no analysis of partial distinguishability.","rationale":"I agree with the reader that the weakest assumption is the exact mode alignment used to reduce the four-mode Hamiltonian to the two-mode model in Eq. (6). All of the paper's quantitative predictions—the Heisenberg scaling for phi at phi=pi, the logarithmically modified Heisenberg scaling for theta at theta=0, and the asymptotic optimality of the observable O—are properties of this reduced model. The assumption is explicitly introduced and is a legitimate idealization, but it sits at a singular point of the parameter space: for any |zi|<1 the model requires auxiliary modes, and the paper provides no continuity or stability argument. This makes the assumption load-bearing for the central claim. My independent check of the internal derivations did not reveal a mathematical error: Eq. (14) reproduces the QFI for the squeezed-state path (14), the covariance-matrix calculation in Section V gives QFI(theta=0)=sinh^4(2g)/g^2, and Lemma V.1 is valid as a Gaussian-state identity. The issue is therefore not soundness of the two-mode model but its domain of validity. The concrete test proposed here would settle whether the scalings survive small departures from perfect alignment; until that is done, the appropriate verdict remains CONDITIONAL. The lack of an experimental implementation for O is a practical limitation but not a correctness risk, so it does not change my read. Since the reader's verdict already captures this concern, I recommend no change.","tokens_in":997,"tokens_out":1250,"duration_ms":352609,"concrete_test":"Generalize Eq. (5) by taking z1 = r e^{-i phi}, z2 = r with 0 < r < 1, and model partial distinguishability by coupling each pair (a1,b2) and (a2,b1) to an auxiliary vacuum mode via a beam-splitter-type overlap (as in Ref. [24]), so that the full dynamics is a four-mode-plus-environment Gaussian unitary. Compute the exact covariance matrix and QFI for phi at phi=pi (theta=0) and for theta at theta=0 (phi=pi) as a function of r for fixed g. If QFI(phi)/E^2 and (ln E)^2 QFI(theta)/E^2 collapse when 1-r exceeds a few percent, the headline scalings are fragile; if they remain O(1) with modified constants as r -> 1, the perfect-alignment idealization is benign.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claims (QFI ~ E^2 for phi at phi=pi, QFI ~ (E/ln E)^2 for theta at theta=0, and asymptotic optimality of O in (17)) are all derived from the two-mode Hamiltonian (6), which is obtained from the four-mode Hamiltonian (4) by setting the distinguishability parameters in (5) to |z1|=|z2|=1, specifically z1=e^{-i phi}, z2=1. This is an extreme point of the allowed parameter space: for |zi|<1, unitarity forces the introduction of auxiliary environment modes (as the paper notes, citing [24]), and the effective dynamics is no longer the two-mode Gaussian evolution analyzed here. The paper gives no argument that the QFI scalings are continuous in |zi| near 1, nor does it bound the effect of small misalignment on the optimal operating points or on the readout O. Because the claimed distinction from the circuit model—including the specific prediction that total photon number readout does not saturate the QCRB—is a property of the exactly aligned two-mode model, an imperfectly aligned experiment could in principle evade all of the paper's quantitative predictions. This is a genuine soft spot, not an internal inconsistency: the paper is explicit about the assumption, but it does not establish that the central scalings are stable under the natural experimental imperfection.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reformulates the SU(1,1) interferometer of Yurke, McCall, and Klauder starting from a four-mode Hamiltonian describing two optical downconversion processes with opposite pump phases. Under the assumption of perfect mode alignment (mode indistinguishability up to a phase), the model reduces to a two-mode Hamiltonian parametrized by the nonlinearity g, the relative phase phi, and the dynamical phase theta. The paper computes the quantum Fisher information (QFI) for phi and theta, identifies optimal operating points (phi = pi and theta = 0, respectively), derives Heisenberg and logarithmically modified Heisenberg scalings for the QFI, constructs an observable O based on weighted shift operators that is claimed to be asymptotically optimal, and compares these results with the standard circuit-based model. The main positive results are the explicit covariance matrix derivation, Lemma V.1 establishing a duality between the relative and dynamical phases, and the analytical QFI formulas.","tokens_in":15761,"tokens_out":7775,"duration_ms":66175,"significance":"If the results hold, the paper provides a first-principles Hamiltonian treatment of the SU(1,1) interferometer with predictions that differ from the circuit-based model: the optimal operating points occur for non-vacuum states inside the interferometer, and the total photon number readout does not saturate the quantum Cramér-Rao bound while the weighted-shift observable O is asymptotically optimal. The derivation is largely analytic and internally consistent; in particular, Appendix A supplies a detailed covariance matrix calculation and Lemma V.1 gives an explicit state-duality proof. The paper also makes falsifiable experimental predictions, which is a strength. The main weakness is that all results are derived under an exact mode-alignment assumption with no analysis of robustness to partial distinguishability.","major_comments":[{"comment":"The two-mode Hamiltonian (6) is obtained by setting the distinguishability parameters to |z1|=|z2|=1 (specifically z1=e^{-iφ} and z2=1). All subsequent claims—the QFI scalings in Eqs. (14) and (28), the optimal operating points φ=π and θ=0, and the asymptotic optimality of the observable O in Eq. (17)—are properties of this perfectly aligned model. The paper notes that |zi|<1 requires auxiliary environment modes, but it gives no argument that the central scalings are continuous in |zi| near 1 and no bound on how small misalignment modifies the QFI and optimal readout. Since the abstract and Discussion present these predictions as \"experimental targets for falsification,\" this missing robustness analysis is load-bearing for the physical claims. Please either extend the analysis to partial distinguishability (e.g., a perturbative calculation in 1-|zi| or a numerical study for representative misalignment) or explicitly restrict the claims to the ideal alignment limit and discuss the experimental relevance of that restriction.","section":"Section III, Eqs. (5)-(6)"},{"comment":"The abstract states that θ=0 is the optimal operating point for sensing the dynamical phase, but the manuscript does not provide an analytic proof that QFI(θ=0) is the global maximum over θ for fixed g. Fig. 2 shows numerical evidence for the values of g considered, and the text says \"we will only note the observation that although Fig. 2 suggests...\" without establishing optimality. This is insufficient to support the unqualified claim in the abstract. Please supply an analytic argument (e.g., monotonicity or concavity properties of QFI(θ) in Domain 1 and Domain 2) or change the claim to a numerically observed optimal operating point.","section":"Section V, Eq. (28) and Fig. 2"},{"comment":"The claim that the total photon number readout O=a†1a1+a†2a2 does not saturate the QCRB at θ→0 in the Hamiltonian model is stated without showing the explicit expression for lim_{θ→0} S_NO(θ). This non-saturation is one of the two main claimed differences from the circuit model, so it should be backed by a concrete calculation (even an asymptotic formula). Please include the computed signal-to-noise ratio at θ=0 or an analytic bound showing it is strictly below QFI(θ=0).","section":"Section V, paragraph after Eq. (23)"}],"minor_comments":[{"comment":"The phrase \"we find in that in the Hamiltonian model\" contains a typo; it should read \"we find that in the Hamiltonian model.\"","section":"Abstract"},{"comment":"The sentence beginning \"Because Therefore,\" is a fragment and appears to be a typographical error; the sentence should be completed or split into two coherent sentences.","section":"Section V, paragraph after Eq. (23)"},{"comment":"The sentence \"Recent proposals for embedded passive interferometers [44,45]\" is incomplete and lacks a verb or continuation; it should be finished or merged with the following sentence.","section":"Section VI, first paragraph"},{"comment":"The claim that the global maximum of QFI(φ) occurs at φ=π is asserted rather than derived; a brief proof or even a sentence noting that Eq. (14) as a function of sin^2(φ/2) is unimodal with endpoints compared at φ=0 and φ=π would remove any ambiguity.","section":"Section IV, Eq. (14)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the technical core appears sound. The main risk is that the ideal-alignment assumption is not addressed, which undermines the stated experimental falsifiability. If the authors provide a robustness analysis or explicitly narrow the claims, the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is worth reading and worth sending to a serious referee. The author reformulates the SU(1,1) interferometer starting from a Hamiltonian for two downconversion processes with opposite pump phases, and shows that under exact mode alignment the model reduces to a two-mode quadratic Hamiltonian. From there he derives QFI scalings: Heisenberg E^2 for the relative phase at phi=pi, and a log-modified Heisenberg scaling (E/ln E)^2 for the dynamical phase at theta=0. He also identifies a weighted shift operator O that asymptotically saturates the QCRB, and shows that the total photon number readout, which is optimal in the circuit model, does not do the job here. Those are real, concrete new results, and the derivations are mostly explicit. Lemma V.1, which establishes that the Hamiltonian-evolved state converges to a two-mode squeezed state in the large-g limit, is a nice piece of work and does the heavy lifting for the dynamical-phase results.\n\nThe main soft spot is exactly what the stress-test note flags: the entire analysis lives on the measure-zero subspace of perfect mode indistinguishability (|z1|=|z2|=1 in Eq. (5)). The paper is upfront about this, but it never quantifies how things change when |zi| drops slightly below 1. That matters because the central contrast with the circuit model—the failure of total photon number readout and the success of O—is a property of the perfectly aligned model. A small misalignment could in principle invalidate all of the quantitative predictions. I don't think this kills the paper; it is a modeling paper and the assumption matches the idealized alignment in the original Yurke et al. setup. But a referee should push for a robustness analysis, even a perturbative one, before people start designing experiments around phi=pi.\n\nTwo smaller issues. The claim that theta=0 is the optimal operating point for dynamical phase appears to rely on Fig. 2, which I cannot check from the manuscript; the analytic argument is only for Domain 2, and the global comparison with Domain 1 is not proven. The experimental implementation of O is not discussed at all—no hint of how one would measure weighted shift operators in practice. Both are addressable. There are also a few typos and two incomplete sentences (end of Section V and Section VI) that should be cleaned up. The citation pattern looks solid, including the author's own earlier work on the Zou-Wang-Mandel effect, which is directly relevant.\n\nWho this is for: anyone working on SU(1,1) interferometry or quantum metrology with Gaussian states. It deserves a serious referee, and I would cite it if I were writing on active interferometers.\n\nMy recommendation: send it to peer review, but with the explicit request that the authors either prove or soften the global optimality claim for theta=0 and add at least a brief discussion of robustness to partial distinguishability.","headline":"A genuinely new Hamiltonian treatment of the SU(1,1) interferometer that finds different optimal operating points and a shift-operator readout; the central limitation is the perfect-alignment assumption, which is explicit but unquantified.","tokens_in":16305,"tokens_out":3220,"would_cite":true,"duration_ms":27346,"reading_group":"yes","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 derives a Hamiltonian model of the SU(1,1) interferometer in which the relative phase is optimally sensed at $\\phi=\\pi$ with quantum Fisher information $E^2$, while the dynamical phase at $\\theta=0$ reaches $(E/\\ln E)^2$.","keywords":["SU(1,1) interferometer","Hamiltonian quantum metrology","quantum Fisher information","Heisenberg scaling","two-mode squeezing","relative phase sensing","dynamical phase sensing","Gaussian states"],"falsifier":"Measure the sensitivity of a vacuum-seeded two-crystal downconversion interferometer at $\\phi=\\pi$ as the pump strength grows. The paper predicts that the weighted shift observable gives Fisher information scaling as $E^2$ while total photon number does not saturate the bound; if total photon-number readout already saturates the bound, or the weighted-shift Fisher information grows slower than $E^2$, the central claim is falsified.","tokens_in":15288,"feed_emoji":"⚛️","tokens_out":9605,"duration_ms":81677,"temperature":0.7,"pith_summary":"The paper re-derives the SU(1,1) interferometer from the Hamiltonian of two optical downconversion processes driven by opposite pump phases, rather than from a sequence of unitary circuit operations. Under exact mode indistinguishability, the four-mode dynamics reduces to a two-mode quadratic Hamiltonian in sp(2,R), whose state is labeled by nonlinearity $g$, relative phase $\\phi$, and dynamical phase $\\theta$. The central result is that the best precision for estimating the relative phase $\\phi$ occurs at $\\phi=\\pi$, where the quantum Fisher information grows as $E^2$ in total energy; for the dynamical phase $\\theta$, the optimum is $\\theta=0$ with scaling $(E/\\ln E)^2$. Unlike in the circuit-based model, the optimal operating points involve a non-vacuum state inside the interferometer, and total photon-number readout does not saturate the quantum Cramér-Rao bound, while a weighted shift observable becomes asymptotically optimal. If correct, this provides a first-principles route to designing quantum optical sensors containing multiple downconversion processes.","feed_headline":"SU(1,1) sensor reaches Heisenberg scaling at $\\phi=\\pi$","feed_subtitle":"A Hamiltonian model predicts $E^2$ relative-phase sensitivity and $(E/\\ln E)^2$ for interaction time.","key_machinery":"The load-bearing object is the two-mode quadratic Hamiltonian $H = \\theta(a_1^\\dagger a_1+a_2^\\dagger a_2) + 2g\\sin(\\phi/2)\\,(i e^{-i\\phi/2} a_1^\\dagger a_2^\\dagger + \\mathrm{h.c.})$, an element of sp(2,R) obtained by setting the mode-distinguishability parameters to $|z_1|=|z_2|=1$. Expressed in the $K_1,K_2,K_3$ generators of su(1,1), this Hamiltonian generates the probe state $e^{-iH}|0\\rangle|0\\rangle$. The argument is carried by exact covariance-matrix calculations for Gaussian states, by the symmetric logarithmic derivative at $\\phi=\\pi$ which motivates the weighted shift observable $O=\\sum_{n=0}^\\infty n(|n+1\\rangle|n+1\\rangle\\langle n|\\langle n|+\\mathrm{h.c.})$, and by the state-duality lemma showing that in the domain $\\lambda^2>\\theta^2$ the evolved vacuum converges to a two-mode squeezed state with squeezing phase $\\pi+\\cos^{-1}(\\theta/\\lambda)$. The covariance-matrix formula $\\mathrm{QFI}(\\theta)=\\tfrac14\\mathrm{tr}[(\\Sigma^{-1}\\partial_\\theta\\Sigma)^2]$ produces the claimed scalings.","core_discovery":"The paper's central claim is that when the SU(1,1) interferometer is treated as actual Hamiltonian dynamics of two four-wave-mixing processes rather than as a product of SU(1,1) gates, the parameters $\\phi$ and $\\theta$ are physically distinct and have different optimal sensing points and scalings. At $\\phi=\\pi$, the relative-phase quantum Fisher information is $E^2$, Heisenberg scaling; at $\\theta=0$, the dynamical-phase quantum Fisher information is $4(E/\\ln 2E)^2$, a logarithmically modified Heisenberg scaling. The paper further claims that at these operating points the total photon-number operator is not optimal; instead, an observable built from weighted two-photon shift operators saturates the quantum Fisher information in the large-$g$ limit. This contrasts with the circuit-based model, where the optimal point $\\theta=0$ is the identity circuit and total photon number saturates the bound. The paper also shows, via symplectic covariance matrices, that the Hamiltonian-evolved vacuum is a two-mode squeezed state in the relevant parameter domain, and it establishes a duality between relative phase and dynamical phase in the large-energy limit.","pith_inferences":["One extension the paper leaves implicit is that the same mode-indistinguishability reduction should apply to any network of downconverters, so the $E^2$ scaling may generalize to multi-crystal sensors beyond the two-process case.","The predicted near-cancellation in the noise of $O^2$ between total intensity and two-photon coherence is a measurable signature that could be tested with photon-number-resolving detection.","If the model is correct, proposed decompositions of this interferometer into quadratic-Hamiltonian circuit elements need to be recompiled, since generic parameter regimes do not correspond to any $\\mathrm{sp}(2,\\mathbb{R})$ Hamiltonian.","Relaxing the distinguishability parameters below $|z_i|=1$ would introduce auxiliary modes and loss; a natural testable prediction is that the Heisenberg scaling degrades smoothly as alignment worsens."],"forward_implications":["Relative-phase estimation at $\\phi=\\pi$ achieves Heisenberg scaling $E^2$, so a vacuum-seeded device reaches the same precision class as nonclassical-input schemes at equal energy.","Dynamical-phase estimation at $\\theta=0$ is not trivial in this model: the state is still entangled and the precision is $(E/\\ln E)^2$, only logarithmically below Heisenberg scaling.","Total photon-number readout, which is optimal in the circuit-based model, is suboptimal in the Hamiltonian model, so experiments should measure the weighted shift observable $O$ instead.","Because the optimal operating points are non-vacuum states inside the interferometer, intensity and precision measurements at these points can discriminate between Hamiltonian and circuit-based dynamics."],"supporting_citations":[{"why":"It introduces the SU(1,1) interferometer and the circuit-based model whose predictions are contrasted with the Hamiltonian model throughout the paper.","marker":"[1]"},{"why":"It presents the nonlinear interferometer circuit model with two downconversion processes that the paper compares against for dynamical-phase sensing.","marker":"[19]"},{"why":"It argues that products of active optical unitaries need not correspond to quadratic Hamiltonians, supporting the paper's dismissal of fine-tuned circuit models.","marker":"[23]"},{"why":"It supplies the relationship between the symplectic group and the pseudounitary group used to write the Hamiltonian in su(1,1) generator form.","marker":"[27]"},{"why":"It provides the covariance-matrix formula for quantum Fisher information that the paper uses to compute the dynamical-phase scaling.","marker":"[43]"},{"why":"It supplies the Gaussian-state covariance-matrix formalism used to derive the covariance matrices and symplectic eigenvalues of the probe states.","marker":"[48]"}],"fun_headline_variants":["Heisenberg scaling for relative phase in SU(1,1) interferometer","At phi=pi, SU(1,1) sensor achieves Heisenberg limit","Two phases, two scalings in Hamiltonian SU(1,1) model","Log-Heisenberg scaling for dynamical phase in SU(1,1)"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole calculation depends on perfect overlap of the two downconversion outputs: the modes pair up so completely that only one relative phase remains. If the interferometer is misaligned, the two-mode Hamiltonian and every precision claim in the paper no longer follow.","fun_headline_variants_meta":{"raw":{"variants":["Heisenberg scaling for relative phase in SU(1,1) interferometer","At phi=pi, SU(1,1) sensor achieves Heisenberg limit","Two phases, two scalings in Hamiltonian SU(1,1) model","Log-Heisenberg scaling for dynamical phase in SU(1,1)"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000772,"raw_usage":{"total_tokens":3476,"prompt_tokens":1061,"completion_tokens":2415,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":677,"completion_tokens_details":{"reasoning_tokens":2333}},"tokens_in":677,"tokens_out":2415,"duration_ms":16247,"temperature":1.0,"reasoning_tokens":2333,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:14:57.223326+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the sensitivity of a vacuum-seeded two-crystal downconversion interferometer at $\\phi=\\pi$ as the pump strength grows. The paper predicts that the weighted shift observable gives Fisher information scaling as $E^2$ while total photon number does not saturate the bound; if total photon-number readout already saturates the bound, or the weighted-shift Fisher information grows slower than $E^2$, the central claim is falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It presents the nonlinear interferometer circuit model with two downconversion processes that the paper compares against for dynamical-phase sensing."},{"cited_title":"Burgarth, P","cited_arxiv_id":null,"evidence_quote":"It argues that products of active optical unitaries need not correspond to quadratic Hamiltonians, supporting the paper's dismissal of fine-tuned circuit models."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the relationship between the symplectic group and the pseudounitary group used to write the Hamiltonian in su(1,1) generator form."},{"cited_title":"Pinel, J","cited_arxiv_id":null,"evidence_quote":"It provides the covariance-matrix formula for quantum Fisher information that the paper uses to compute the dynamical-phase scaling."},{"cited_title":"Serafini, Quantum Continuous Variables: A Primer of Theoretical Methods(CRC Press, 2017)","cited_arxiv_id":null,"evidence_quote":"It supplies the Gaussian-state covariance-matrix formalism used to derive the covariance matrices and symplectic eigenvalues of the probe states."}],"review_version":1}