{"id":"8fad61fb-b423-440a-ba09-ef0834338fda","arxiv_id":"2508.19890","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Non-Gaussianity of a quantum state equals the correlation its two copies generate at a 50:50 beam splitter, enabling a state-agnostic, sample-efficient measurement via SWAP tests.","lead":"This paper shows that a quantum state's non-Gaussianity, a key resource for continuous-variable quantum computing, can be measured by mixing two copies of the state on a 50:50 beam splitter and quantifying how strongly the outputs are correlated. The result yields a state-agnostic, tomography-free certification protocol, and a bound showing the standard Wigner-negativity benchmark is harder to estimate.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Faithfulness of N_C rests on an unstated quantum Darmois–Skitovich theorem, and Theorem 1's 'arbitrary correlation measure' is contradicted by the paper's own separable-output example.","rationale":"The central claim has two pillars: monotonicity under Gaussian channels and faithfulness (zero iff Gaussian). The monotonicity pillar is well supported: Appendix B correctly shows that Gaussian channels can be commuted through the beam splitter to act locally on the outputs, so any total-correlation measure that is monotone under local operations inherits monotonicity. The remaining load-bearing pillar is faithfulness, and it is not proved in the paper. It is delegated to an external theorem whose hypotheses are not stated. If that theorem has unstated regularity conditions, the framework's central diagnostic could fail on valid physical states. The paper also overstates the scope of Theorem 1 by saying 'arbitrary measure of correlation.' The Appendix C example shows a non-Gaussian state whose output is separable, so entanglement-based correlation measures are not faithful; only faithful total-correlation measures such as Rényi mutual information satisfy the claimed equivalence. The pure-state Rényi entropy is not a total-correlation measure on mixed outputs, which compounds the problem for the paper's headline example. These issues do not necessarily sink the framework — they can be fixed by stating the theorem's conditions and restricting C — but they make the current unconditional Theorem 1 and the abstract's 'all such measures' claim inaccurate. This supports the reader's CONDITIONAL verdict rather than demanding rejection, provided the authors qualify the claims and supply the missing verification.","tokens_in":22671,"tokens_out":17636,"duration_ms":221908,"concrete_test":"Obtain the cited Cuesta theorem [31] and check whether it applies to arbitrary density operators with finite mean photon number; if it assumes finite second moments, verify that all finite-energy states satisfy them, and if it assumes purity or full rank, test a rank-deficient non-Gaussian state by computing whether U_BS ρ⊗ρ U_BS† factorizes. Separately, instantiate Theorem 1 with C = entanglement of formation and the Appendix C state ρ = (|α⟩⟨α| + |−α⟩⟨−α|)/2; if N_C(ρ)=0 for this non-Gaussian state, then 'arbitrary measure of correlation' faithfulness is false as written.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The faithfulness side of Theorem 1 is loaded entirely onto the assertion after Eq. (6) that the beam-splitter output is product iff the input is Gaussian, citing [31]. The paper neither states the hypotheses of the quantum Darmois–Skitovich theorem nor proves them for the full class of states considered. If that theorem requires finite second moments, purity, or other regularity conditions, the characterization 'N_C=0 iff Gaussian' can fail for states outside that class. Separately, Theorem 1 states 'C being an arbitrary measure of correlation' and then claims faithfulness. But Appendix C itself constructs non-Gaussian states whose beam-splitter output is separable, so any entanglement-based C — including the pure-state Rényi entropy used as the paper's flagship example — gives N_C=0 for a non-Gaussian state. Faithfulness therefore holds only for C that is a faithful total-correlation measure (e.g., Rényi mutual information), not for arbitrary C. This is a genuine gap in the central claim as stated, even though the monotonicity proof in Appendix B is sound for valid total-correlation measures.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a framework for quantifying non-Gaussianity of a continuous-variable state ρ by the correlation generated when two copies of ρ are mixed on a 50:50 beam splitter, defining N_C(ρ) := C(U_BS ρ⊗ρ U_BS†) for a correlation measure C. The central claim (Theorem 1) is that for any correlation measure C, N_C is monotonic under Gaussian channels and faithful, i.e., N_C(ρ)=0 iff ρ is Gaussian. The authors instantiate the construction with the pure-state Rényi-α entanglement entropy and the mixed-state Rényi-α mutual information, propose a SWAP-test protocol for estimating the Rényi-2 purity with four copies and O(1/ϵ²) samples, and derive a lower bound on the sample complexity of estimating Wigner negativity. Appendix B proves that Gaussian channels can be commuted through the beam splitter so that they act locally on the outputs; Appendix C exhibits non-Gaussian states whose beam-splitter output is separable.","tokens_in":22759,"tokens_out":11911,"duration_ms":144361,"significance":"The idea of characterizing non-Gaussianity through correlation generation at a beam splitter is elegant and provides a useful operational connection, generalizing the Hong–Ou–Mandel effect. For the restricted class of faithful total-correlation measures that are monotone under local operations, the monotonicity argument in Appendix B is sound and the framework is a genuine contribution. The Rényi-2 instance gives a concrete, state-agnostic estimation protocol with constant sample complexity in the ideal setting, which is a practical advantage over full tomography. The analytical examples and the Wigner-negativity sample-complexity comparison are also valuable. However, the theorem as stated is substantially overbroad, and several load-bearing claims need to be tightened before the results can be accepted as stated.","major_comments":[{"comment":"The statement that N_C is faithful for 'C being an arbitrary measure of correlation' is contradicted by the paper's own Appendix C. Equation (C5) constructs a non-Gaussian state ρ = Σ_i p_i D(r_i)ρ_G D(r_i)† whose beam-splitter output is separable. If C is any entanglement-based correlation measure, including the pure-state Rényi entropy E_α of Eq. (7) when used in its natural role, then N_C(ρ)=0 for this non-Gaussian state. Faithfulness therefore holds only for C that is a faithful total-correlation measure (vanishing exactly on product states) and that is monotone under local operations. The theorem should be restated with this class of C; otherwise the central 'iff' claim is false as written.","section":"Theorem 1 / Eq. (6)"},{"comment":"The pure-state measure E_α is not a valid correlation measure for mixed states if Eq. (8) is used as the definition. For a Gaussian thermal state ρ_G, the beam-splitter output is product, ρ_G⊗ρ_G, yet the right-hand side of Eq. (8), namely ∫|χ_ρ(r/√2)|^4/(2π)^m dr, is strictly less than 1, so N_E2(ρ_G)>0. Thus E_2 as defined is nonzero on a Gaussian state and is not a faithful measure. This also breaks the claimed monotonicity under Gaussian channels: a Gaussian noise channel can map a pure input to a thermal state, and the Appendix B proof applies only if C is a local-operation monotone on all states, which E_α is not. The authors should either restrict Theorem 1 to pure states and Gaussian unitaries, or use a genuinely mixed-state total-correlation measure throughout.","section":"Eqs. (7)-(8) and Experimental access"},{"comment":"The assertion that the beam-splitter output is product iff the input is Gaussian is the entire basis for faithfulness, but the paper does not state the theorem from Ref. [31] or verify its hypotheses. The title of Ref. [31] ('A stable quantum Darmois-Skitovich theorem') suggests that stability or regularity conditions may be involved. If those conditions are not satisfied by all states considered, the equivalence N_C=0 iff Gaussian can fail. The manuscript should state the precise theorem used and confirm that every state in the claimed domain satisfies its assumptions.","section":"Faithfulness / Ref. [31]"},{"comment":"The lower bound on the sample complexity of estimating Wigner negativity is advertised as a main result, but the derivation contains uncontrolled approximations. The bound ∥W∥1 ≳ x^{1/6} in Eqs. (H20)-(H29) relies on statements such as 'the term ... does not contribute', 'neglect constant factors', and 'approximate with a constant c'. The plotted scaling in Figure 4 additionally uses a numerically observed x^{1/3} growth. To support the claimed third-root scaling of the sample complexity, the Wigner-negativity lower bound should be proved rigorously, or the scaling should be explicitly labeled as numerical/heuristic.","section":"Appendix H"}],"minor_comments":[{"comment":"Typo: 'states that cen be represented' should be 'states that can be represented'.","section":"Background"},{"comment":"Typo: 'faithfull' should be 'faithful'.","section":"Theorem 1"},{"comment":"Typo: 'psot beam splitter' should be 'post beam splitter'.","section":"Appendix C"},{"comment":"Grammar: 'the number of sample required' should be 'the number of samples required'.","section":"Discussions"},{"comment":"The claim that the protocol is energy-independent should be qualified by the detector-saturation requirement 1−q_{2M}≤ϵ, which is energy-dependent.","section":"Experimental access"}],"recommendation":"major_revision","confidential_remarks":"The manuscript overlaps with independent work [56], which the authors acknowledge. The core construction is promising and likely publishable, but the main theorem overstates the domain of validity and the pure-state measure has a genuine mixed-state gap. These issues are fixable with a careful restatement of the class of admissible correlation measures and the conditions on the Darmois–Skitovich theorem. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is worth a serious look. The core move—defining non-Gaussianity as the correlation generated by sending two copies of a state through a 50:50 beam splitter—is natural and useful, and for the specific measures they actually use (Renyi-2 purity for pure states, Renyi mutual information for mixed states) it works. The monotonicity proof in Appendix B is correct: Gaussian channels commute with the beam splitter and become local operations on the outputs, so any correlation measure that is non-increasing under local operations gives a monotone. The SWAP-test protocol for the pure-state Renyi-2 case is a clean generalization of Hong-Ou-Mandel and the O(1/eps^2) sample count is plausible.\n\nBut Theorem 1, as stated, overclaims. It says N_C is monotone and faithful for “any measure of correlation C.” Their own Appendix C constructs a non-Gaussian mixed state whose beam-splitter output is separable, so any entanglement-based C—including the pure-state Renyi entropy they feature—gives N_C equal to zero for a non-Gaussian state. Faithfulness holds only if C is a faithful total-correlation measure, like the Renyi mutual information. The theorem needs that qualifier.\n\nThe pure-state measure has a second issue: Theorem 1 asserts monotonicity under all Gaussian channels, but a Gaussian channel can take a pure state to a mixed one, and the pure-state Renyi entropy is not defined for mixed outputs. If you extend it using the formula in Eq. (8), a Gaussian noise channel on a Gaussian pure state turns the measure from zero into a positive value, so monotonicity fails. The fix is easy—state that the pure-state measure is monotone under purity-preserving Gaussian channels (e.g., unitaries) and use the mutual information version for general channels—but as written the blanket claim is wrong.\n\nThe faithfulness step also leans entirely on a cited quantum Darmois-Skitovich theorem without stating its hypotheses. I suspect it is fine, but a referee should verify the conditions hold for the full state class considered.\n\nThe Wigner-negativity lower bound is the weakest section. The state-discrimination argument is legitimate, but the quantitative scaling rests on asymptotic approximations (oscillating Airy integrals approximated by constants) that are heuristic. The paper itself calls the x^{1/3} bound numerical, so the claimed “third-root of mean photon number” sample lower bound is not a proven theorem. Soften it or prove it.\n\nThe overlap with Bu-Li is disclosed honestly in the note added, and the citation pattern looks clean. The paper shows clear thinking and the core idea deserves referee time, but it needs revision before acceptance.","headline":"Solid operational framework with a fixable but real overclaim in the main theorem: the pure-state Rényi measure doesn't extend monotonically to mixed outputs, and faithfulness needs the 'faithful total correlation' qualifier.","tokens_in":23432,"tokens_out":4915,"would_cite":true,"duration_ms":55985,"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 claims that a quantum state is non-Gaussian exactly when two of its copies become correlated at a 50:50 beam splitter, and uses that correlation as a faithful, experimentally accessible measure of non-Gaussianity.","keywords":["non-Gaussianity","continuous variables","beam splitter","Rényi entropy","SWAP test","Wigner negativity","Gaussian channels","correlation measures"],"falsifier":"Take a Gaussian thermal state, compute the Rényi-2 purity of one beam-splitter output using Eq. (8); the result is log2(1+2n)>0, nonzero for a Gaussian state. If the paper's monotonicity claim is extended to this measure, a Gaussian channel applied to a pure Gaussian state would move the value from 0 to positive, violating N_C(Φ(ρ)) ≤ N_C(ρ). So the crux is whether the cited Darmois–Skitovich theorem's regularity hypotheses hold for the states in question.","tokens_in":22414,"feed_emoji":"⚛️","tokens_out":11355,"duration_ms":119529,"temperature":0.7,"pith_summary":"Non-Gaussian states are the resource behind quantum advantage in continuous-variable computing, but existing quantifiers are hard to measure. This paper proposes to quantify non-Gaussianity by what it does: mixing two copies of a state on a 50:50 beam splitter produces a correlated output exactly when the input is non-Gaussian, and a product output exactly when the input is Gaussian. The authors prove that any correlation measure applied to the beam-splitter output is a faithful non-Gaussianity monotone under Gaussian channels. For pure states the Rényi-2 version equals the purity of one output mode, which a SWAP test can estimate using four copies and O(1/ε²) samples without knowing the state, independent of mode number and energy. They also prove a lower bound showing that estimating Wigner negativity costs at least a cube-root-of-energy number of samples, so their protocol can be substantially cheaper.","feed_headline":"Non-Gaussianity is exactly the capacity to create correlation at a beam splitter","feed_subtitle":"A SWAP-test protocol measures it with four copies and O(1/ε²) samples, no tomography required.","key_machinery":"The load-bearing object is the 50:50 beam splitter acting on two identical copies: U_BS=exp(iπ/4 Σ [p_A q_B − q_A p_B]). The paper defines N_C(ρ)=C(U_BS ρ⊗ρ U_BS†) with C an arbitrary measure of correlation; the central identity is the characteristic-function factorization χ_{U_BS ρ⊗ρ U†}(r1,r2)=χρ((r1−r2)/√2)χρ((r1+r2)/√2), which yields a product state iff ρ is Gaussian (via the cited quantum Darmois–Skitovich theorem). For pure states, correlation reduces to entanglement, so the Rényi-α entanglement entropy of one output mode—at α=2, the negative log of the reduced-state purity—serves as the measure; the purity is estimated by a SWAP test (standard or destructive, with PNR detectors), requ","core_discovery":"At the paper's center is Theorem 1: for any correlation measure C, N_C(ρ)=C(U_BS ρ⊗ρ U_BS†) is a faithful measure of non-Gaussianity—zero if and only if ρ is Gaussian—and it is monotonic under Gaussian channels. The fact that makes this work is the quantum Darmois–Skitovich theorem, invoked to show that two copies of a state become uncorrelated (in fact, product) after a 50:50 beam splitter exactly when the input is Gaussian. For pure states, the generated correlation is entanglement, so the Rényi-α entropy of one output mode is a non-Gaussianity quantifier; the α=2 case is the purity of the reduced state, which is directly accessible by a SWAP test using four copies and constant depth. For","pith_inferences":["Not pursued in the paper but a direct corollary of the construction: any observed correlation between the two beam-splitter outputs certifies that the input was outside the Gaussian set, so the setup can serve as a state-agnostic non-Gaussianity witness without tomography.","A natural extension the authors leave open is applying the same correlation-generation logic to channels rather than states, e.g., sending a Gaussian probe through an unknown process and measuring the correlation it induces as a quantifier of non-Gaussian operation.","The lower-bound result suggests, though the paper does not design it, that direct estimation protocols for Wigner negativity alone could outperform full state reconstruction; finding such a protocol is an open research direction."],"forward_implications":["If Theorem 1 is right, any measure of correlation—not just entropy—turns into a faithful non-Gaussianity monotone under Gaussian channels, giving a one-line recipe for new measures.","The Rényi-2 instance makes non-Gaussianity experimentally accessible for pure states using only a SWAP test (or a destructive PNR version), with constant sample complexity independent of mode number and energy, in contrast to full state tomography.","The framework unifies and generalizes the Hong–Ou–Mandel experiment: the HOM dip becomes a special case of correlation generation for Fock-state inputs, applicable to arbitrary states.","For mixed states, Rényi-α mutual information is the appropriate correlation measure; the paper shows these are also monotones and gives bounds amenable to future measurement protocols.","Estimating Wigner negativity requires sample complexity at least growing with the cube root of mean photon number for cubic phase states, so the correlation-based measure can be dramatically cheaper."],"supporting_citations":[{"why":"Supplies the quantum Darmois–Skitovich theorem that the beam-splitter output is a product state if and only if the input is Gaussian; the foundation of faithfulness.","marker":"[31]"},{"why":"Provides the standard SWAP-test method for directly estimating state purity, used to access the Rényi-2 measure.","marker":"[42]"},{"why":"Gives the ancilla-free continuous-variable SWAP test with photon-number-resolving detectors, making the purity measurement feasible in optics.","marker":"[46]"},{"why":"Defines the entanglement Rényi-α entropy used to quantify pure-state non-Gaussianity from the beam-splitter output.","marker":"[33]"},{"why":"Defines Rényi-α mutual information and its bounds, the correlation measure proposed for mixed states.","marker":"[34]"},{"why":"Supplies the Wigner-negativity resource-theoretic background and the cubic-phase-state negativity properties used in the sample-complexity lower bound.","marker":"[27]"},{"why":"Provides the continuous-variable state-tomography sample complexity used as the baseline comparison.","marker":"[29]"},{"why":"Shows that a beam splitter generates entanglement only from nonclassical input, the prior idea this framework generalizes.","marker":"[54]"}],"fun_headline_variants":["Non-Gaussianity = beam-splitter correlation","Non-Gaussian iff correlation at beam-splitter","Measure non-Gaussianity via beam-splitter correlation","Correlation at a beam splitter signals non-Gaussianity","Beam-splitter correlation quantifies non-Gaussianity"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The entire faithfulness statement rests on the cited quantum Darmois–Skitovich theorem, whose technical hypotheses are not stated, and the monotonicity proof presumes the pure-state Rényi measure remains a valid correlation measure when a Gaussian channel outputs a mixed state.","fun_headline_variants_meta":{"raw":{"variants":["Non-Gaussianity = beam-splitter correlation","Non-Gaussian iff correlation at beam-splitter","Measure non-Gaussianity via beam-splitter correlation","Correlation at a beam splitter signals non-Gaussianity","Beam-splitter correlation quantifies non-Gaussianity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000892,"raw_usage":{"total_tokens":3684,"prompt_tokens":743,"completion_tokens":2941,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":487,"completion_tokens_details":{"reasoning_tokens":2860}},"tokens_in":487,"tokens_out":2941,"duration_ms":22269,"temperature":1.0,"reasoning_tokens":2860,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T15:27:02.413223+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a Gaussian thermal state, compute the Rényi-2 purity of one beam-splitter output using Eq. (8); the result is log2(1+2n)>0, nonzero for a Gaussian state. If the paper's monotonicity claim is extended to this measure, a Gaussian channel applied to a pure Gaussian state would move the value from 0 to positive, violating N_C(Φ(ρ)) ≤ N_C(ρ). So the crux is whether the cited Darmois–Skitovich theorem's regularity hypotheses hold for the states in question.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the quantum Darmois–Skitovich theorem that the beam-splitter output is a product state if and only if the input is Gaussian; the foundation of faithfulness."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the standard SWAP-test method for directly estimating state purity, used to access the Rényi-2 measure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the ancilla-free continuous-variable SWAP test with photon-number-resolving detectors, making the purity measurement feasible in optics."},{"cited_title":"Wang, L.-Z","cited_arxiv_id":null,"evidence_quote":"Defines the entanglement Rényi-α entropy used to quantify pure-state non-Gaussianity from the beam-splitter output."},{"cited_title":"McKinlay and M","cited_arxiv_id":null,"evidence_quote":"Defines Rényi-α mutual information and its bounds, the correlation measure proposed for mixed states."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that a beam splitter generates entanglement only from nonclassical input, the prior idea this framework generalizes."}],"review_version":1}