{"id":"ec2acb50-5483-4bd1-8f91-251d3741ab9d","arxiv_id":"2504.15831","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Third-order partial-transpose moments, read out with few-copy Fourier interferometers and photon counting, certify non-Gaussian continuous-variable entanglement, including all NOON states.","lead":"Entanglement in non-Gaussian quantum light can be certified from a few copies of the state using simple interferometers and photon-counting detectors. This new readout detects families like NOON states that standard continuous-variable entanglement tests miss.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Noisy-copy robustness analysis evaluates the criterion on unequal copies, which is not a valid entanglement test; the infinite-dimensional Descartes step, by contrast, is a presentation gap, not a correctness risk.","rationale":"The reader's flagged weakest assumption (infinite-dimensional Descartes/Newton step in Sec. IIA2) is not the load-bearing risk. The necessary conditions used for detection follow directly from positivity of the partial transpose: for separable states ρ̃ ≥ 0, so its eigenvalues are nonnegative, and Newton's identities acting on the finite traces p2, p3 yield the linear and quadratic bounds without any need for Descartes' rule over an infinite spectrum. The paper's terse appeal to Descartes is a presentational gap, not a source of false positives. The genuinely load-bearing concern is the noisy-copy robustness analysis. Fig. 3b evaluates the third-order criterion on three different states with different parameters. Since the criteria are derived for identical copies, applying them to p2 = Tr{ρ1ρ2} and p3 = Tr{ρ1⊗ρ2⊗ρ3 Π_A⊗Π_B^{-1}} is unjustified. I constructed a concrete counterexample where three separable product states violate the quadratic criterion, demonstrating that unequal-copy violations do not certify entanglement. This does not invalidate the central detection results for NOON, cat, and HHG states, nor the loss and finite-statistics analyses, so the CONDITIONAL verdict stands, but the robustness claim should be reworked around the average state or i.i.d. noise model.","tokens_in":29445,"tokens_out":29229,"duration_ms":274356,"concrete_test":"Redo the noisy-copy analysis using the physically correct average state arρ = ∫ρ(α,τ)p(α)p(τ)dα dτ for the Gaussian noise model, and plot the same detection regions; compare with Fig. 3b. Independently, evaluate the unequal-copy criterion on triples of separable states from the N=1 lossy NOON family (including one copy with α=0 or τ=0) and check whether any region yields p3 < p2^2. If such violations occur, the unequal-copy use of the criterion is invalid and the robustness claim must be revised to rely solely on the average-state or i.i.d. analysis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The core mathematical extension of the third-order PT-moment criteria to CV is sound: for any separable state ρ, the partial transpose is positive semidefinite, so its eigenvalues λ_i ≥ 0. Then e3 = (1 - 3p2 + 2p3)/6 ≥ 0 gives the linear criterion p3 ≥ (3p2 - 1)/2, and the Hankel matrix [[1, p2], [p2, p3]] ≥ 0 gives the quadratic criterion p3 ≥ p2^2, without invoking Descartes' rule in infinite dimensions. The paper's Sec. IIA2 assertion is terse but the necessary direction is safe. However, the noisy-copy analysis in Sec. VB1 (Fig. 3b) computes p2 and p3 from three different states: p2 = Tr{ρ1 ρ2} and p3 = Tr{ρ1⊗ρ2⊗ρ3 Π_A⊗Π_B^{-1}}. The separability bounds (6)-(7) were derived for identical copies, i.e., p_n = Tr{ρ̃^n} of a single state. For unequal copies no such bound exists. Concretely, choose Bob's state fixed and Alice's three single-mode states |0>, -1/2|0> + √3/2|1>, -1/2|0> - √3/2|1>. Each copy is a product state, hence separable, yet p2 = 1/4 and the three-copy cyclic overlap is (-1/2)^3 = -1/8, so p3 = -1/8 < p2^2 = 1/16, violating the quadratic criterion. Thus the 'detected entanglement' regions in Fig. 3b can include false positives, and the claimed robustness to noisy copies is not established by that analysis. The full i.i.d. simulation in Fig. 4 is better posed, but the paper does not draw this distinction.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper extends recently introduced partial-transpose moment (PT-moment) entanglement criteria to continuous-variable systems and proposes a multicopy interferometric readout of the first nontrivial moments p2 and p3 using discrete Fourier transforms and particle-number-resolving measurements. It derives third-order criteria, benchmarks them on Gaussian states, mixed cat states, high-harmonic-generation states, and NOON states, and analyzes robustness to loss, noisy copies, and finite statistics. The central claim is that the criteria certify genuine non-Gaussian entanglement, including all NOON states for all mode numbers N, and remain robust under realistic experimental constraints.","tokens_in":1490,"tokens_out":1506,"duration_ms":112097,"significance":"If the claims hold, this is a significant contribution: PT-moment criteria provide an experimentally concrete route to certifying entanglement in states that escape Gaussian second-moment and entropic criteria. The analytic NOON result for arbitrary N is striking, and the proposed Fourier-interferometer readout is feasible with current technology. The paper contains no fitted free parameters: the benchmark values are computed analytically from the stated state families, and the finite-statistics error model is explicit and testable. The core third-order inequalities (6)-(7) are valid for separable CV states, as they follow from elementary moment inequalities for the positive semidefinite partial transpose, and the multicopy implementation is a natural extension of known purity and Rényi-entropy measurements. However, the noisy-copy robustness analysis in Sec. VB1 evaluates p2 and p3 on unequal copies, for which no separability bound exists, so the robustness claim in its current form is not established.","major_comments":[{"comment":"The noisy-copy analysis computes p2 from copies 1 and 2 and p3 from copies 1, 2, and 3 with independently varying parameters. The inequalities (6)-(7) are only valid for p_n = Tr{ρ̃^n} of a single state ρ; for unequal copies no such separability bound exists. This is not a harmless modeling choice: take Bob's state fixed and Alice's three single-mode states |0>, -1/2|0>+√3/2|1>, and -1/2|0>-√3/2|1>. Each copy is a bipartite product state and hence separable, yet the two-copy overlap is 1/4 and the three-copy cyclic overlap is -1/8, so p3 = -1/8 < p2^2 = 1/16, meaning the quadratic criterion would flag a fully separable ensemble as entangled. The 'detected entanglement' regions in Fig. 3b can therefore contain false positives, and the claimed robustness to noisy copies is not established by this analysis. The same clarification is needed for the full simulation in Sec. VB2 and Fig. 4: if the phase and loss parameters are drawn independently for each copy within a run, the identical-copy assumption of the criterion is violated.","section":"Sec. VB1, Fig. 3b, criteria (6)-(7)"},{"comment":"The extension of Newton's identities and Descartes' rule of signs to the infinite-dimensional partial transpose is asserted in one sentence and is not a standard result: a characteristic polynomial is not generally defined for a trace-class operator with countably many eigenvalues. The third-order criteria (6) and (7) are nevertheless correct for separable CV states, because ρ̃ is positive semidefinite and p_n are moments of a probability distribution. For example, p3 ≥ p2^2 follows from Cauchy-Schwarz, and p3 ≥ (3p2 - 1)/2 follows from λ^3 - (3/2)λ^2 + 1/2 = (λ-1)(λ^2 - λ/2 - 1/2) ≥ 0 for λ ∈ [0,1]. The authors should replace the Descartes-rule justification with such an elementary derivation, or provide a rigorous treatment of the infinite-dimensional case.","section":"Sec. IIA2, Eq. (5)"}],"minor_comments":[{"comment":"The phrase 'detecting genuine non-Gaussian entanglement' could be read as claiming that the criteria certify non-Gaussianity; in fact the same third-order criteria also detect Gaussian entangled states, as shown in Sec. IVB. Please clarify that the criteria certify entanglement, and that the non-Gaussian character is a property of the target states being benchmarked.","section":"Sec. IVD1 / Abstract"},{"comment":"The Hermite-Gauss wavefunction is typeset ambiguously; please add explicit parentheses and define the parameters σ+ and σ- in the text.","section":"Eq. (45)"},{"comment":"The caption should state precisely which quantities are computed analytically and which are sampled, and it should define the boundary and color conventions for the 'detected entanglement' regions.","section":"Fig. 3b"},{"comment":"References [88] and [89] appear to share the same journal volume and article number; please verify that these are two distinct papers with correct bibliographic data.","section":"References"},{"comment":"The comparison between the linear p3-PPT criterion and the Shchukin-Vogel criterion relies on an implicitly defined matrix c via f2(c) = p2(ρ) - (3/2)ρ + 1/2; please state explicitly whether this map is shown to be well-defined for all states or only for the examples considered.","section":"Sec. IVA1"}],"recommendation":"major_revision","confidential_remarks":"The core PT-moment extension and the multicopy readout are sound and novel, and the NOON benchmark is strong. The main obstacle is the unequal-copy analysis in Sec. VB1, which currently invalidates the robustness claim as stated. I believe this is fixable by either (i) restricting the noise model to parameters that are common to all copies within each experimental run and deriving the estimator correctly, or (ii) developing valid bounds for unequal copies, or (iii) substantially weakening the robustness claim and removing Fig. 3b. The Descartes-rule gap is presentation-level, since the needed inequalities have elementary proofs. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core of this paper is worth taking seriously: the multicopy DFT readout for PT-moments in continuous variables is a natural but non-obvious step, and the demonstration that all NOON states are certified by third-order criteria is solid. The Gaussian moment formula, the cat and NOON calculations, and the lossy NOON expressions all look right. The worry about extending Descartes' rule to infinite dimensions turns out to be a non-issue: for separable states the partial transpose is positive semidefinite, so its eigenvalues are nonnegative, and the linear criterion follows from e3 >= 0 without any infinite-dimensional Descartes' rule. That section is terse, but the mathematics is safe.\n\nThe real soft spot is the noisy-copy robustness analysis. The criteria (6)-(7) are derived for moments of a single state's partial transpose, p_n = Tr{ρ̃^n}. In Sec. VB1, and possibly in the full simulation of Fig. 4, the authors evaluate p2 and p3 on three different states. Those quantities are not PT-moments of any state, and the separability bounds do not apply. The stress-test counterexample is decisive: three separable product states for Alice and a fixed Bob state give p2 = 1/4 and p3 = -1/8, violating the quadratic criterion. So Fig. 3b can flag separable states as entangled, and the paper's claim in Sec. VI that false positives can be excluded is too strong. If the authors intended the simulation to represent run-to-run fluctuations with identical copies within each run, they need to say that clearly; as written, \"for each individual copy\" suggests per-copy variation.\n\nThis is a serious flaw in an auxiliary part of the paper, not in the central proposal. The core criteria, the readout scheme, and the identical-copy benchmarks survive. But the robustness claims should be corrected or substantially qualified. I'd send it to peer review—the core contribution is important enough—but the referee should insist on fixing the noisy-copy analysis. Also, no code or data is provided for the numerics; that is minor given the analytic expressions, but worth asking for.\n\nWho this is for: anyone in photonics or ultracold atoms working on CV entanglement detection, and people who care about practical non-Gaussian entanglement witnesses. I would bring it to the reading group; the flawed robustness section is a good case study in how multicopy estimators only mean something when the copies are identical. Recommendation: engage, but with eyes open. The technique is a keeper; the robustness section needs work.","headline":"The multicopy DFT readout of PT-moments is a genuinely useful new CV entanglement detection tool, but the noisy-copy robustness analysis in Sec. VB1 is flawed because it evaluates criteria on unequal copies.","tokens_in":30289,"tokens_out":4370,"would_cite":true,"duration_ms":37540,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Mn","42.50.Ex"],"model":"deepseek-v4-flash","headline":"Third-order partial-transpose moment criteria detect NOON entanglement for every photon number $N$ via three-copy Fourier interferometry.","keywords":["partial transpose moments","continuous variables","entanglement detection","NOON states","multicopy method","Fourier interferometer","photon-number-resolving detection","non-Gaussian entanglement"],"falsifier":"Search over separable continuous-variable states for one whose third elementary symmetric polynomial $e_3$, computed from the PT-moments via Newton's identities, is negative while $p_3 \\ge p_2^2$ and $p_3 \\ge (3p_2-1)/2$ both hold; finding such a state would show that the Descartes-rule extension to infinite dimensions is invalid. A concrete numerical route is to take truncated Fock-space approximations of a candidate separable state and check whether the violation persists as the cutoff grows.","tokens_in":29217,"feed_emoji":"⚛️","tokens_out":9523,"duration_ms":81484,"temperature":0.7,"pith_summary":"This paper establishes that entanglement tests based on moments of the partially transposed state, previously limited to discrete-variable systems, work for continuous-variable optical and atomic systems. The core claim is that any separable state must satisfy the two third-order inequalities $p_3 \\ge p_2^2$ and $p_3 \\ge (3p_2-1)/2$, where $p_n = \\mathrm{Tr}\\{\\tilde{\\rho}^n\\}$ are moments of the partial transpose; violating either one certifies entanglement. The authors propose a concrete readout of $p_2$ and $p_3$ from just two or three copies of the state, passive linear optics, and particle-number-resolving detectors. This detects genuine non-Gaussian entanglement in families that homodyne, entropic, and quadrature-moment criteria miss, including every NOON state for every mode number $N$. They also model losses, copy-to-copy noise, and finite statistics, and simulate a full experiment showing that a Bell-type NOON state is certified within about $10^3$ samples even at 40% loss.","feed_headline":"Three-copy interferometer certifies NOON entanglement at any N","feed_subtitle":"Photon counting on three replicas catches non-Gaussian entanglement that homodyne and entropic tests miss.","key_machinery":"The central object is the partial-transpose moment $p_n = \\mathrm{Tr}\\{\\rho^{\\otimes n}\\, (\\vec{\\Pi}^A_n \\otimes \\overleftarrow{\\Pi}^B_n)\\}$, where the shift operators permute the $n$ copies on each side. Because the shift operator is circulant, the $n$-mode discrete Fourier transform $F(n)$ diagonalizes it into a phase observable $D_n$ that depends only on the output particle numbers, so $p_n$ becomes a weighted average of $n$th roots of unity on the measured photon counts. For $n=2$ this is just a balanced beam splitter followed by parity measurements; for $n=3$ it is a three-beam-splitter, three-phase-shift interferometer for both parties, with Bob measuring the inverse phase weights. The inequalities themselves come from two hierarchies: Hankel-matrix positivity of the Stieltjes moment problem, and Newton's identities combined with Descartes' rule of signs on the characteristic polynomial of $\\tilde{\\rho}$.","core_discovery":"The paper's central claim is that the $p_n$-PPT criteria extend to continuous variables, where $\\tilde{\\rho} = (1 \\otimes T_B)\\rho$ is the partial transpose of a bipartite state on an infinite-dimensional Hilbert space. For third order, separability implies the two conditions in Eq. (6), and for pure states these conditions are also sufficient, since $p_3 = \\sum_i |c_i|^6$ in the Schmidt basis equals $1$ only for product states. For NOON states $|\\psi\\rangle = \\alpha|N,0\\rangle + \\beta|0,N\\rangle$, the paper computes $p_3 = |\\alpha|^6 + |\\beta|^6$, independent of $N$, so every nontrivial NOON state violates the criterion; the violation is largest, $-3/4$, at balanced amplitudes. The argument also shows that the noisy, lossy, finite-statistics version remains a valid witness, with explicit error models and simulated experimental runs.","pith_inferences":["A direct extension not pursued here would apply the same Fourier readout to moments of the realignment map; if it works, bound-entangled Gaussian states could be certified with the same two-or-three-copy resources.","The infinite-dimensional validity question could be tested numerically by truncating the Fock basis at increasing cutoff $d$: if a separable state with negative $e_3$ emerges while the $p_3$ inequalities hold for all cutoffs, the Descartes-rule extension would be refuted.","Because the PT-moment inequalities are basis-independent, one testable prediction is that they certify entanglement for high-$N$ NOON states with a sample budget growing only mildly with $N$ under fixed loss, which would make the method useful for metrology-oriented photonic experiments.","A multimode or multipartite generalization, which the authors sketch, would make the same observable a candidate for detecting genuine multipartite non-Gaussian entanglement in many-body bosonic systems; this is my inference, not a result of the paper."],"forward_implications":["Every pure entangled continuous-variable state is certified by the third-order conditions, because pure-state entanglement is equivalent to $p_3 < 1$.","All NOON states with both amplitudes nonzero are detected, independent of the photon number $N$, closing a gap left by second-moment, entropic, and existing multicopy witnesses.","Third-order PT-moment criteria are robust: for NOON states up to $N=10$, losses up to roughly 20% still permit certification, and no certification is possible beyond 50% loss.","The readout cost is fixed by the copy number rather than the Hilbert-space dimension: two copies for $p_2$ and three for $p_3$, with small sample counts required in the simulated Bell-state experiment.","The method applies to settings without phase-stable local oscillators, such as high-harmonic-generation sources, where homodyne or heterodyne quadrature readout is unavailable."],"supporting_citations":[{"why":"Introduces PT-moment entanglement criteria from randomized measurements and supplies the quadratic $p_3 \\ge p_2^2$ condition.","marker":"[82]"},{"why":"Derives the Newton/Descartes hierarchies and the linear $p_3$ criterion that the paper extends to infinite dimensions.","marker":"[83]"},{"why":"Establishes the optimal third-order PT-moment criterion, which the paper uses for mixed-state benchmarks and comparisons.","marker":"[84]"},{"why":"Provides the multicopy method for measuring polynomial functionals of a state, the basis of the interferometric readout.","marker":"[85]"},{"why":"Shows how discrete Fourier transforms on copies turn cyclic shift observables into particle-number measurements, the template for the $p_2$/$p_3$ circuits.","marker":"[99]"},{"why":"Prior multicopy continuous-variable entanglement witness based on spin observables that detects NOON states only up to $N=2$; the comparison benchmark the paper outperforms.","marker":"[89]"},{"why":"Entropic continuous-variable entanglement criterion used as a baseline; it misses the genuinely non-Gaussian states the new method detects.","marker":"[62]"}],"fun_headline_variants":["Three-copy moment test catches NOON entanglement","Multicopy partial transpose criterion flags non-Gaussian entanglement","Three-copy scheme detects genuine non-Gaussian entanglement","Multicopy moment criterion detects NOON entanglement","Photon counting on three copies certifies NOON states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper assumes, without proof, that a classical sign-counting rule for polynomials still works for the infinitely many eigenvalues, some possibly negative, of the partially transposed state; if that assumption is wrong, the linear criterion can produce false-positive entanglement verdicts.","fun_headline_variants_meta":{"raw":{"variants":["Three-copy moment test catches NOON entanglement","Multicopy partial transpose criterion flags non-Gaussian entanglement","Three-copy scheme detects genuine non-Gaussian entanglement","Multicopy moment criterion detects NOON entanglement","Photon counting on three copies certifies NOON states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000754,"raw_usage":{"total_tokens":3312,"prompt_tokens":863,"completion_tokens":2449,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":479,"completion_tokens_details":{"reasoning_tokens":2373}},"tokens_in":479,"tokens_out":2449,"duration_ms":17141,"temperature":1.0,"reasoning_tokens":2373,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:17:44.792385+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search over separable continuous-variable states for one whose third elementary symmetric polynomial $e_3$, computed from the PT-moments via Newton's identities, is negative while $p_3 \\ge p_2^2$ and $p_3 \\ge (3p_2-1)/2$ both hold; finding such a state would show that the Descartes-rule extension to infinite dimensions is invalid. A concrete numerical route is to take truncated Fock-space approximations of a candidate separable state and check whether the violation persists as the cutoff grows.","supporting_citations":[{"cited_title":"Gärttner, T","cited_arxiv_id":null,"evidence_quote":"Shows how discrete Fourier transforms on copies turn cyclic shift observables into particle-number measurements, the template for the $p_2$/$p_3$ circuits."},{"cited_title":"Giovannetti, S","cited_arxiv_id":null,"evidence_quote":"Prior multicopy continuous-variable entanglement witness based on spin observables that detects NOON states only up to $N=2$; the comparison benchmark the paper outperforms."}],"review_version":1}