{"id":"0b3b4364-17c8-4735-9d7d-94d707677220","arxiv_id":"2608.03755","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For any Gaussian single-mode state and any orthogonal copy-mixing matrix fixing the symmetric collective mode, Tr(ρ^⊗k Γ(O)) = (1-λ)^k/det(I_k - λO); violation of this identity certifies non-Gaussianity.","lead":"A theory paper builds non-Gaussianity detectors from symmetries of many identical copies of a quantum light state under passive mode mixing, deriving a closed-form identity every Gaussian state must satisfy. Measuring a multi-copy expectation and the purity, any violation of the identity certifies the state is non-Gaussian.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (153)'s certification rule ignores systematic detector loss and inefficiency, so the claimed 1−δz−δp confidence for the experimental witness is unestablished; the algebraic Gaussian identity itself appears sound.","rationale":"The reader's weakest_assumption pinpoints exactly the most load-bearing gap. I independently checked the algebraic core: Theorem 1/2, the trace identity Eq. (80) via spectral decomposition, the k=3 examples (Fock, cat, mixture), and the k=2 purity reduction all are internally consistent; I found no error in the Gaussian reference derivation. The commutation with displacement requiring Oe1=e1 is sound, and the purity-based fixing of λ is not circular. The remaining weakness is therefore not in the mathematics of the witness but in the experimental certification claim. Equations (143)–(153) treat sampling noise as the only error source, giving a confidence statement 1−δ_z−δ_p. Any physical implementation with loss, mode-dependent efficiency, dark counts, or copy drift converts the estimator into a biased one, and the threshold ε_z+L_λ ε_λ does not include that bias. A false certification of a Gaussian state becomes possible. This is load-bearing because the paper's stated added value is an experimentally accessible certificate with bounded sample complexity. The same kind of gap appears in the multi-mode covariance estimation, but the single-mode loss/detector bias is enough to invalidate the headline experimental claim as written. A targeted simulation on a known Gaussian state with realistic η will settle whether the bias exceeds the tolerance and thus whether the confidence claim survives; until then, the reader's CONDITIONAL verdict is appropriate.","tokens_in":22545,"tokens_out":21052,"duration_ms":245929,"concrete_test":"Simulate the full protocol for a known single-mode Gaussian state (e.g., thermal λ=0.5) at detector efficiency η=0.9 and dark count ν=10^{-3} per mode, with θ=π/(2√3) and the sample sizes required by Eqs. (145),(147) for ε_z=ε_λ=0.05, δ=0.01. Compute the exact expectation of the phase estimator under the loss/dark-count channel, e.g. via the generating function (1−η+η e^{i√3θ})^{n3}(1−η+η e^{-i√3θ})^{n2} averaged over the post-tritter PNR distribution for ρ_λ^{⊗3}, and compare |z_meas−W(λ,θ)| to ε_z+L_λ ε_λ. If the bias exceeds the threshold, the claimed 1−δ_z−δ_p false-positive bound fails; this check settles whether the missing systematic-error analysis is fatal to the experimental claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The protocol's central experimental claim is that z(θ)=Tr(ρ^{⊗3}Γ(e^{θG})) can be estimated unbiasedly by averaging the phase estimator X=exp(i√3θ(n3−n2)) after a Fourier tritter, with confidence governed only by Hoeffding sampling noise (Eqs. (143)–(145)). This is true only if the three copies are i.i.d., the tritter implements exactly Γ(F)†, and PNR detection is lossless, unit-efficiency, and dark-count-free. With photon loss of efficiency η, the measured estimator's expectation becomes Σ_n p(n) (1−η+η e^{i√3θ})^{n3}(1−η+η e^{-i√3θ})^{n2} rather than z(θ); for a Gaussian state this equals neither z(θ) nor any W(λ,θ) in the family, and the difference is not covered by ε_z+L_λ ε_λ in Eq. (153). Hence the false-certification probability can exceed δ_z+δ_p, and the claimed 'experimentally accessible' certificate with O(ε^{-2}) sample complexity is not justified as stated. The same gap applies to the purity SWAP estimate if implemented with loss (purity is underestimated, biasing λ̂). The multi-mode criterion (Sec. V.A) additionally assumes error-free covariance estimation with no propagation, but the single-mode loss bias is the more load-bearing gap because it directly invalidates the central confidence claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a group-theoretic, multi-copy method for detecting non-Gaussianity of bosonic states. The main result is that for every single-mode Gaussian state ρ_G with thermal parameter λ, and every orthogonal copy-mixing matrix O∈O(k) with Oe_1=e_1, the multi-copy expectation value Tr(ρ_G^{⊗k} Γ(O)) equals (1−λ)^k / det(I_k−λO) (Eq. (80)), independent of displacement and squeezing. After fixing λ from the measured purity, a violation of this identity certifies non-Gaussianity. The authors illustrate the witness with Fock states, cat states, and a mixture of coherent states, and propose an interferometric protocol based on a Fourier tritter and photon-number-resolved detection, with Hoeffding-based sample complexity O(ε^{−2}). They also extend the construction to multi-mode systems. The algebraic core appears sound; the experimental confidence analysis, however, currently covers only sampling noise and not systematic loss or detector inefficiency.","tokens_in":22764,"tokens_out":19256,"duration_ms":221294,"significance":"If the result holds, this is a valuable new witness: it is derived from first principles rather than fitted, depends on a single purity-calibrated parameter, and gives a one-sided certificate with a concrete linear-optics protocol. The examples are analytically nontrivial (e.g., Per(O)=5/9 for the n=1 Fock case and the cat-state second derivative) and the multi-copy symmetry perspective connects to Howe duality. The statistical bounds for sampling noise (Eqs. (144)–(147)) are correctly derived for an ideal lossless implementation. The main gap is that the claimed experimental confidence, Eq. (153), is not robust to photon loss, detector inefficiency, or imperfect interferometers, and the multi-mode extension omits error propagation from covariance estimation. These gaps are fixable but currently limit the 'experimentally accessible' claim.","major_comments":[{"comment":"The claimed confidence 1−δ_z−δ_p for the certification rule assumes the estimator X(n2,n3)=exp(i√3θ(n3−n2)) in Eq. (139) is unbiased. With per-photon efficiency η, the measured expectation becomes Σ_n p(n)(1−η+η e^{i√3θ})^{n3}(1−η+η e^{−i√3θ})^{n2}, which is not z(θ) for any state and, for a Gaussian state, is not W(λ,θ) for any λ. This systematic bias is not covered by the threshold ε_z+L_λ ε_λ, so a Gaussian state can be falsely certified with probability exceeding δ_z+δ_p. Please either include loss in the statistical model (e.g., calibrate η, use a loss-tolerant estimator, or add a systematic-error term) or restate the protocol as applying only to a lossless, unit-efficiency, ideal-tritter setting.","section":"Sec. IV.B, Eq. (153)"},{"comment":"The same loss bias affects the SWAP purity estimate: with loss, the measured purity is systematically low, so λ̂=(1−p̂)/(1+p̂) is biased upward. The bound |λ̂−λ|≤2ε_p in Eq. (148) is derived under the assumption that the SWAP estimator is unbiased; with loss it does not hold. Consequently, the error-propagation term L_λ ε_λ in Eq. (153) does not cover the purity-calibration bias. A complete protocol needs a loss-calibrated purity estimator or an explicit worst-case systematic offset.","section":"Sec. IV.B, purity estimation (Eqs. (137)–(148))"},{"comment":"The multi-mode criterion assumes exact knowledge of the covariance matrix and hence of the symplectic eigenvalues λ_r. In an experiment, the covariance matrix is estimated from finitely many quadrature measurements, and the map from covariance matrix to λ_r is nonlinear. No error propagation is provided, so the violation condition in Eq. (188) has no stated confidence. To make the multi-mode witness experimentally meaningful, the authors should include finite-sample error bounds for λ_r or explicitly state that the criterion is algebraic and does not yet carry a statistical guarantee.","section":"Sec. V.A, Eqs. (186)–(188)"}],"minor_comments":[{"comment":"The text says 'Γ′(A)† = −Γ(A)', but it should be '−Γ′(A)' (the prime is missing on the right-hand side).","section":"Sec. II.B, after Eq. (8)"},{"comment":"The line 'Γ(O)aΓ(O)−1 = O^T a = O^T a' contains a duplicated expression; one occurrence should be removed.","section":"Sec. III.A, Eq. (29)"},{"comment":"The text refers to 'Fig. III C' for the comparison of W(r) and f(r), but no figure appears in the manuscript. Please either include the figure or remove the reference.","section":"Sec. III.C, after Eq. (124)"},{"comment":"'applying Hoeffding’s inequality to the real and imaginary parts separately gives yields'—'gives yields' is redundant; use 'gives'.","section":"Sec. IV.B, Eq. (144)"},{"comment":"The sentence beginning 'Let ρ be an unknown d-mode state' is repeated almost verbatim in two consecutive paragraphs; the duplication should be removed.","section":"Sec. V.A, around Eq. (187)"}],"recommendation":"major_revision","confidential_remarks":"The algebraic theorem (Eq. (80)) and the single-mode examples are sound and form a solid core. The revision should focus on the experimental claims: either add a loss/tolerance analysis or explicitly delimit the ideal-device regime. The multi-mode section also needs an error analysis or a caveat. With those changes, the paper would be acceptable; without them, the 'experimentally accessible' conclusion overreaches the derived confidence bounds."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the algebra is right, the experimental confidence argument is not. The paper's real contribution is the O(k−1)-stabilizer witness family and the closed-form Gaussian reference (1−λ)^k/det(I_k−λO), calibrated by purity. I re-derived the k=2 purity limit, the Fock-state permanent, the cat-state second derivative, and the mixture comparison; they all check out. The one-sided detection logic is sound and the λ-fixing is not circular.\n\nThe soft spot is exactly where the stress-test lands. Eq. (153) claims a 1−δ_z−δ_p confidence under Hoeffding bounds alone, but that ignores systematic errors. With photon loss of efficiency η, the measured phase average becomes Σ_n p(n)(1−η+ηe^{i√3θ})^{n3}(1−η+ηe^{−i√3θ})^{n2}, which is neither z(θ) nor W(λ,θ). The false-certification probability can exceed the advertised bound. The purity SWAP is also loss-biased, which shifts λ. So the 'experimentally accessible' claim is not justified as written. The algebraic identity in Eq. (80) does not depend on this, and the multi-copy witness could still be measured in a low-loss regime, but the confidence statement needs a systematic-error budget or a clear caveat.\n\nTwo smaller issues. There is no benchmark against the existing witnesses in [37–48], so the practical advantage is unquantified. The multi-mode extension is a sketch without error propagation. Notational sloppiness in Eqs. (99)–(100) and a missing figure reference should be cleaned up.\n\nWho this is for: people working on non-Gaussianity certification, GKP magic states, or bosonic coding. They get a clean symmetry argument and explicit examples worth stealing. It deserves a serious referee, but not as is: the experimental section needs revision before the main claim can be trusted.\n\nRecommendation: send to peer review, with a request that the authors either include loss in the confidence model or moderate the claim. I'd cite the algebraic part.","headline":"The algebraic witness family is sound and the closed-form Gaussian reference is a genuinely useful addition, but the experimental protocol's confidence claim ignores systematic loss and needs a serious caveat before it can be called experimentally accessible.","tokens_in":23431,"tokens_out":2667,"would_cite":true,"duration_ms":31589,"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":"Every single-mode Gaussian state obeys one fixed multi-copy mixing identity, calibrated by purity alone; a measured violation certifies non-Gaussianity, via a three-copy interferometric protocol with O(1/ε²) samples.","keywords":["non-Gaussianity detection","Gaussian states","multi-copy symmetries","passive linear optics","Howe duality","commutant structure","continuous-variable quantum information","quantum Fisher information"],"falsifier":"Numerically compute Tr(ρ^⊗3 Γ(e^{θG})) in a truncated Fock basis for a displaced squeezed thermal state with known λ and compare it with (1−λ)²/(1−2λcos(√3θ)+λ²) across a sweep of θ — any statistically significant disagreement would refute the algebraic identity. Alternatively, run the three-copy tritter protocol on a pure coherent state, for which the identity predicts the flat value 1 for every θ; a systematic deviation beyond the Hoeffding bound would falsify the claim.","tokens_in":22276,"feed_emoji":"⚛️","tokens_out":12024,"duration_ms":129322,"temperature":0.7,"pith_summary":"Non-Gaussian states are the resource behind many continuous-variable quantum advantages, yet detecting them is hard because the set of Gaussian states is not convex, so no single linear observable cleanly separates the two families. This paper proves that every single-mode Gaussian state obeys a fixed identity when k identical copies are mixed by a passive linear-optical transformation: the expectation value depends only on the state's thermal parameter λ, and λ is fixed by a single purity measurement. A state that violates the identity is therefore certified non-Gaussian. The paper gives a concrete three-copy protocol — a Fourier tritter followed by photon-number-resolved detection — whose sample complexity scales as 1/ε², and extends the identity to multi-mode states through their Williamson thermal parameters.","feed_headline":"Certify non-Gaussian states with a three-copy symmetry test","feed_subtitle":"With a purity measurement and three copies through an interferometer, a phase mismatch certifies non-Gaussianity.","key_machinery":"The carrying mechanism is the commutant of the Gaussian action on k copies. Orthogonal copy-mixing unitaries Γ(O), O ∈ O(k), commute with the tensor product of any single-copy squeezing or rotation — a pairing between the metaplectic group and the orthogonal group known as Howe duality — while the displacement (Heisenberg) part of a Gaussian unitary restricts commutation to matrices satisfying Oe1 = e1, a subgroup isomorphic to O(k−1). This structure makes the Gaussian expectation value a function of λ and O alone, and the spectral decomposition of O turns the trace into a product of geometric series, yielding the closed-form determinant reference (1−λ)^k / det(I_k − λO). In the three-copy c","core_discovery":"The paper's central claim is the identity Tr(ρ_G^⊗k Γ(O)) = (1−λ)^k / det(I_k − λO), valid for every single-mode Gaussian state ρ_G with thermal parameter λ and every orthogonal copy-mixing matrix O that fixes the collective-mode vector e1. The identity is proven from a commutant structure: the symplectic part of any Gaussian unitary commutes with every orthogonal copy-mixing transformation Γ(O), while the displacement part reduces the commuting family to the stabilizer subgroup O(k−1). Because λ is recoverable from the purity p = Tr(ρ²) through λ = (1−p)/(1+p), the right-hand side becomes a computable reference: any measured violation, for any k ≥ 3 and any admissible O, certifies that the","pith_inferences":["Passing the test for a single θ does not certify Gaussianity — the witness is purely one-sided — so in practice an experimenter would sweep θ and scan the O(k−1) family to build evidence, and the achievable gap between a non-Gaussian state and its Gaussian reference is not bounded a priori.","The purity error enters the certification twice, once in the reference value W(λ̂,θ) and once in the threshold via L_λ, so the stated 1−δ_z−δ_p confidence is only as good as the purity measurement's own calibration; a self-tested purity estimate on the same ensemble would be the natural stress test.","Because the examples already detect non-Gaussianity through the second derivative at θ = 0, taking derivatives of the identity at θ = 0 yields a systematic family of polynomial witnesses that avoid tuning the interferometer to special angles.","The construction's cost grows with k (more copies to prepare, count, and pass through a larger interferometer) while the detector family O(k−1) grows too; the paper does not optimize this trade-off, so the choice of k for a given platform remains open."],"forward_implications":["One purity measurement plus one multi-copy phase observable gives a one-sided certificate of non-Gaussianity for unknown single-mode states, with no full state tomography or Wigner-function reconstruction.","The three-copy protocol uses only a Fourier tritter and photon-number-resolved detection, and its estimators are bounded phase factors |X| = 1, so Hoeffding bounds give O(1/ε²) samples to additive error ε, independent of the photon-number distribution.","For multi-mode states the reference factors over modes with Williamson thermal parameters, so the same commutant argument yields a multi-mode witness from covariance-matrix data.","The sensitivity of ρ^⊗k to copy-mixing rotations suggests a quantitative, asymmetry-based measure of non-Gaussianity built from quantum-Fisher-information-type quantities, which the paper leaves for future work."],"supporting_citations":[{"why":"Supplies the Howe-duality multiplicity-free decomposition of the oscillator representation, framing the O(k) commutant used in Theorem 1.","marker":"[54, 55]"},{"why":"Provides the Bargmann–Fock and Heisenberg-group conventions that define the second-quantized transformations and displacement operators.","marker":"[50]"},{"why":"Supplies the complex-basis symplectic representation, Bogoliubov transformations, and the Gaussian-state formalism the paper builds on.","marker":"[52]"},{"why":"Earlier result that the mixed two-coherent-state example must be consistent with, serving as the baseline for Example 3.","marker":"[56]"},{"why":"Photon-number-resolved detector technology that the measurement protocol assumes for reading out the phase estimator.","marker":"[57, 58]"},{"why":"Universal linear-optical multiport decomposition, cited as the basis for realizing the Fourier tritter.","marker":"[59]"},{"why":"Integrated photonic implementation of the three-port Fourier tritter used in the protocol.","marker":"[60]"},{"why":"Standard error-propagation rule used to convert purity-estimate uncertainty into the tolerance on the reference value.","marker":"[61]"},{"why":"Asymmetry and quantum-Fisher-information framework proposed in the conclusion as the basis for a quantitative non-Gaussianity measure.","marker":"[62, 63]"}],"fun_headline_variants":["Three-copy interferometry certifies non-Gaussianity","Purity plus three copies certify non-Gaussianity","Multi-copy symmetry test flags non-Gaussianity","Group-theoretic witness detects non-Gaussian states","Non-Gaussian states caught by multi-copy symmetry"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The certification step assumes every estimator is unbiased: the k copies are independently and identically prepared, the Fourier tritter implements exactly the intended transformation, and photon counting is lossless with unit efficiency, so any systematic bias shifts the violation signal or the purity-derived λ outside the stated 1−δ_z−δ_p confidence interval; the multi-mode witness makes the analogous assumption for covariance-matrix estimation.","fun_headline_variants_meta":{"raw":{"variants":["Three-copy interferometry certifies non-Gaussianity","Purity plus three copies certify non-Gaussianity","Multi-copy symmetry test flags non-Gaussianity","Group-theoretic witness detects non-Gaussian states","Non-Gaussian states caught by multi-copy symmetry"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000829,"raw_usage":{"total_tokens":3472,"prompt_tokens":769,"completion_tokens":2703,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":513,"completion_tokens_details":{"reasoning_tokens":2641}},"tokens_in":513,"tokens_out":2703,"duration_ms":23115,"temperature":1.0,"reasoning_tokens":2641,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T13:24:57.801492+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically compute Tr(ρ^⊗3 Γ(e^{θG})) in a truncated Fock basis for a displaced squeezed thermal state with known λ and compare it with (1−λ)²/(1−2λcos(√3θ)+λ²) across a sweep of θ — any statistically significant disagreement would refute the algebraic identity. Alternatively, run the three-copy tritter protocol on a pure coherent state, for which the identity predicts the flat value 1 for every θ; a systematic deviation beyond the Hoeffding bound would falsify the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Bargmann–Fock and Heisenberg-group conventions that define the second-quantized transformations and displacement operators."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the complex-basis symplectic representation, Bogoliubov transformations, and the Gaussian-state formalism the paper builds on."},{"cited_title":"Goodman, inRepresentations of real and p-adic groups(World Scientific, 2004) pp","cited_arxiv_id":null,"evidence_quote":"Earlier result that the mixed two-coherent-state example must be consistent with, serving as the baseline for Example 3."},{"cited_title":"Fukuda, G","cited_arxiv_id":null,"evidence_quote":"Universal linear-optical multiport decomposition, cited as the basis for realizing the Fourier tritter."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Integrated photonic implementation of the three-port Fourier tritter used in the protocol."},{"cited_title":"Spagnolo, C","cited_arxiv_id":null,"evidence_quote":"Standard error-propagation rule used to convert purity-estimate uncertainty into the tolerance on the reference value."}],"review_version":1}