{"id":"5a78a195-d1bc-43a6-9e5d-96bf4775dab8","arxiv_id":"2506.18303","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"From third-order correlations of randomized measurements, the authors reconstruct ten local invariants and derive a separability criterion that detects Werner states in d=3 for p > 1/cuberoot(10), improving the second-order threshold p > 1/2.","lead":"Randomized local measurements can extract third-order invariants of an unknown quantum state, leading to new separability criteria without full tomography. For Werner states in dimension 3 the third-order test improves the detected entanglement range over second-order purity tests.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The printed invariant-form criterion xS≤2x8 is not equivalent to the proven inequality (47): Eq. (42) and Table II imply xS≤x8, and the printed form fails to detect states the proven criterion detects.","rationale":"After independent re-derivation, Eq. (47) is valid for separable states: apply the reduction criterion to σ=ρ^{T_B}; since σ is separable and Tr_B(σ^2)=Tr_B(ρ^2) for separable ρ, adding the two inequalities yields Eq. (47). The Werner d=3 threshold p>1/∛10 follows from the polynomial in Eq. (49), and the core mathematics is therefore sound. The most load-bearing defect is the mismatch between Eq. (47), which bounds Trρ^3+Tr((ρ^{T_B})^3) by 2x8, and the invariant-form statement xS≤2x8. The Table II diagram mapping implies x9=Trρ^3 and x10=Tr((ρ^{T_B})^3), so Eq. (42)'s identification of x9+x10 with two partial-transpose traces and the factor 2 in the printed criterion are both inconsistent with that mapping. This is not cosmetic: it changes which states the implementable criterion detects, as shown by the qutrit maximally entangled example. Secondary issues, including the d=2 singularity in the Weingarten denominators and the d_A/d_B typo in the Appendix inverse, reinforce the need for a revised version, but the factor-of-two invariant-form error is the single check that should gate acceptance.","tokens_in":19837,"tokens_out":24173,"duration_ms":227850,"concrete_test":"Evaluate the tensor-network diagrams in Table II for the two-qutrit maximally entangled state |Φ⟩=(|00⟩+|11⟩+|22⟩)/√3 and compute xS=(x9+x10)/2 and x8 from the paper's definitions. If xS=67/72 and x8=1/3, then xS≤x8 is violated while the printed xS≤2x8 holds; this confirms that the invariant-form criterion must be corrected to xS≤x8 (or Eq. (42) redefined consistently with Eq. (47)).","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section IV.A proves Eq. (47): for separable ρ, reduction-map positivity applied to ρ and to ρ^{T_B} gives Trρ^3+Tr((ρ^{T_B})^3) ≤ 2x8, using Tr_B((ρ^{T_B})^2)=Tr_B(ρ^2) for separable states; this step is sound. The failure is in the translation to invariants. Eq. (42) defines 2xS=x9+x10=Tr((ρ^{T_B})^3)+Tr((ρ^{T_A})^3), and the text then states Eq. (47) reads as xS≤2x8. But the diagram mapping in Table II gives x9+x10=Trρ^3+Tr((ρ^{T_B})^3), not the sum of two partial-transpose traces; in any case xS as defined in Eq. (42) is not half of the left side of Eq. (47). The provable invariant-form inequality is xS≤x8. With the printed xS≤2x8, the criterion is strictly weaker: for the two-qutrit maximally entangled state, Trρ^3=1, Tr((ρ^{T_B})^3)=31/36, xS=67/72, x8=1/3, so xS≤x8 is violated but xS≤2x8 is satisfied. Thus an experimenter following the printed invariant-form criterion would miss a detection that Eq. (47) guarantees. This is a load-bearing error in the actionable form of the central claim, even though the underlying inequality (47) and the Werner-state benchmark survive once the factor is corrected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an experimentally feasible entanglement-detection scheme based on third-order local invariants accessible via randomized measurements. The authors use Weingarten calculus to reconstruct a set of third-order invariants from averaged correlation statistics for bipartite systems, generalizing the second-order purity reconstruction. They then derive separability criteria from positivity of the reduction map, obtaining the third-order inequality Tr(ρ^3)+Tr((ρ^Γ)^3)≤2Tr(ρ_A Tr_B ρ^2). For Werner states in dimension d=3, this improves the detection threshold from p>1/2 (second-order) to p>1/∛10 (third-order). The paper claims the scheme works for arbitrary local dimensions and discusses two-qubit Bell-diagonal states as a benchmark.","tokens_in":20098,"tokens_out":22349,"duration_ms":196327,"significance":"If the reconstruction and criteria are valid, this is a valuable contribution: it probes third-order spectral invariants without full tomography, goes beyond second-order purity criteria, and gives a concrete falsifiable prediction for Werner states. The derivation is analytic and parameter-free, using standard Weingarten calculus, and the Werner-state threshold is a clean benchmark. However, the operational invariant form of the third-order criterion contains a factor error, and the d=2 case is not covered by the reconstruction, so the manuscript requires revision before the central claims are fully supported.","major_comments":[{"comment":"The translation of inequality (47) into the invariant basis is inconsistent with the definitions in Eq. (42) and the diagram assignments in Table II. The left side of Eq. (47) is Trρ^3+Tr((ρ^Γ)^3), and evaluating the diagrams in Table II for a two-qutrit maximally entangled state gives x9=Trρ^3=1, x10=Tr((ρ^{T_B})^3)=1/9, and x8=1/3; hence the left side is 10/9 while 2x8=2/3, so Eq. (47) is violated. However, with the printed invariant form xS≤2x8 one obtains xS=(x9+x10)/2=5/9≤2/3, so the form is satisfied and the state is not detected. The correct invariant form is xS≤x8. This is load-bearing because the invariant form is the operationally used criterion, even though the Werner-state threshold in Eq. (49) appears to have been evaluated directly from Eq. (47). Please correct the factor and the definition of xS in Eq. (42).","section":"IV.A, Eq. (47) and the following sentence"},{"comment":"The reconstruction of the third-order invariants from the measured vector y(2,3) is singular for local dimension d=2. The Weingarten functions in Eq. (7) and the matrices Q_K in Eq. (41) contain factors (d^2−4) in the denominator, so the advertised inversion x=M^{-1}y is not defined for qubit local dimensions. The abstract claims arbitrary local dimensions, and Section IV.B explicitly discusses two-qubit Bell-diagonal states, but no separate d=2 treatment or explicit limiting argument is provided. The final polynomial Eq. (49) can be evaluated at d=2 by continuity, but the central procedure for obtaining xS from randomized measurements is not justified for d=2. The authors should either restrict the third-order protocol to d_A,d_B≥3 or supply a dedicated qubit analysis.","section":"IV, Eqs. (7), (41), and Appendix A"}],"minor_comments":[{"comment":"The displayed equality appears to contain a misprint: positivity of the reduction map applied to ρ gives Tr(ρ(ρ_A⊗I_B−ρ))=Trρ_A^2−Trρ^2≥0, not Trρ_A^2−Trρ≥0.","section":"IV.A, Eq. (45)"},{"comment":"The line defining x4 and x5 lacks punctuation and is difficult to parse; please separate the definitions clearly.","section":"Eq. (42)"},{"comment":"There are grammatical errors: 'Werner states in d=3 the entanglement is detected' should read 'For Werner states in d=3, entanglement is detected', and the sentence 'All the higher-order criteria using reduction won’t be able to detect entangled states' appears to miss the word 'more'.","section":"Abstract and Conclusions"},{"comment":"Ref. [23] is cited in the text as 'Zoller et al.', but the first author is Brydges et al.; please adjust the citation style.","section":"References"},{"comment":"The expressions ρ_A Tr_B ρ^2 would be clearer with explicit parentheses, e.g., Tr(ρ_A (Tr_B ρ^2)), to avoid ambiguity about the operator ordering.","section":"Eqs. (46) and (47)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope and the core derivation via Weingarten calculus appears sound. The main issues are a correctable factor error in the invariant-form criterion and a missing d=2 treatment; I do not see grounds for rejection, but the invariant inequality must be fixed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper is worth refereeing, but the version as printed has a factor-two error in the actionable form of its central criterion, plus an unhandled qubit limit that the abstract overclaims. The underlying derivation is sound.\n\nWhat's new: the second-order purity criterion and the Haar-averaged randomized measurement scheme are known from van Enk–Beenakker and Brydges et al.; the genuinely new piece is the systematic reconstruction of third-order local invariants via Weingarten calculus and the reduction-map separability criterion Eq. (47), including the partial-transpose combination. I checked the Werner d=3 benchmark: the threshold 1/∛10 follows from Eq. (47) with the definitions in Eq. (42), and the derivation uses no fitted parameters. The 10x10 inversion is explicit and the citation pattern is clean.\n\nSoft spots: the sentence “In terms of (42) it reads as xS ≤ 2x8” is not right. The proven inequality is Tr ρ^3 + Tr((ρ^Γ)^3) ≤ 2x8; with the intended xS = (x9 + x10)/2, this is xS ≤ x8, not xS ≤ 2x8. The printed form is strictly weaker and will miss detections that Eq. (47) guarantees. On top of that, Eq. (42)'s definition of xS as half of Tr((ρ^{T_B})^3) + Tr((ρ^{T_A})^3) does not match the left side of Eq. (47); the invariant labeling needs to be fixed so that x9 + x10 is actually the measurable combination. For what it's worth, the specific counterexample in the note I was sent doesn't hold numerically—I get Tr((ρ^Γ)^3) = 1/9 for the two-qutrit maximally entangled state, not 31/36—but the factor mismatch is real.\n\nSecond issue: the third-order Weingarten denominators contain d^2−4, so the formulas as written are singular for d=2. The abstract says arbitrary local dimensions. The qubit case needs a separate limit or a statement that the third-order criterion is vacuous or derivable there. That is a limitation, not a fatal flaw.\n\nBottom line: the paper is a solid subfield contribution that needs corrections before someone should implement it. The core math holds up; the mistakes are in the translation from inequality to invariants and in the domain of validity. I would send it to a referee, and I'd tell the authors to fix the factor and the d=2 caveat before posting a revised version. I wouldn't cite the current printed criterion.","headline":"Solid third-order randomized-measurement entanglement criterion, but the printed invariant-form inequality has a factor-of-two error and the qubit limit is unhandled.","tokens_in":20709,"tokens_out":13821,"would_cite":false,"duration_ms":122860,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper reconstructs all third-order local invariants from randomized measurements and uses them to prove a separability criterion that detects Werner states in dimension 3 beyond the $p>1/2$ threshold.","keywords":["entanglement detection","randomized measurements","third-order local invariants","Weingarten calculus","separability criteria","Werner states","partial transpose"],"falsifier":"Prepare a two-qutrit Werner state with $p$ in the interval $(10^{-1/3}, 1/2)$, estimate $x_S$ and $x_8$ from randomized measurements, and compare with values from full tomography: the claim predicts $x_S>2x_8$ for every such $p$, so finding a $p$ in that range where the inequality still holds would falsify the criterion's improvement.","tokens_in":19568,"feed_emoji":"⚛️","tokens_out":10104,"duration_ms":94137,"temperature":0.7,"pith_summary":"This paper claims that entanglement can be detected from randomized measurements by using third-order local invariants, quantities built from three copies of the density matrix that are unchanged by local unitary rotations. The authors show that these invariants are fully determined by measured correlation statistics via an invertible linear system, so no full state tomography is needed. They derive a separability criterion, inequality (47), $\\mathrm{Tr}(\\rho^3)+\\mathrm{Tr}((\\rho^\\Gamma)^3)\\le 2\\mathrm{Tr}(\\rho_A\\mathrm{Tr}_B\\rho^2)$, which every separable bipartite state must satisfy. For Werner states in dimension 3, this detects entanglement for $p>10^{-1/3}$, improving the second-order purity criterion's $p>1/2$ threshold. If correct, the protocol offers an experimentally feasible path to entanglement detection beyond spectral criteria.","feed_headline":"Third-order criterion catches more Werner-state entanglement","feed_subtitle":"Randomized local measurements recover higher-order invariants, catching d=3 Werner states below the p=0.5 line.","key_machinery":"The machinery is the Weingarten calculus for Haar-random unitaries applied to the outer-loop average of products of three outcome probabilities. For each subsystem, the permutation group $S_3$ acts on the three copies of $\\rho$; its conjugacy classes — identity, transpositions, and 3-cycles — define linear maps $I_K$, $T_K$, $C_K$ whose traces over the diagrams in Tables I and II are the invariants $\\{x_i\\}$. The matrix $M=SW_A\\otimes SW_B$ connects the measured statistics $\\vec y$ to $\\vec x$, and its explicit inversion (after reducing equal columns and using $y_4-y_5=\\eta(x_4-x_5)$) yields the invariants directly from experimental data.","core_discovery":"The central discovery is that all third-order local invariants accessible from randomized measurements can be reconstructed from the $(3,2)$-twirling data $\\vec y^{(2,3)}$ — three-point outcome correlations averaged over local Haar-random unitaries — by inverting a $10\\times10$ matrix $M=SW_A\\otimes SW_B$. Among the recovered invariants is $x_S=(x_9+x_{10})/2=\\mathrm{Tr}(\\rho^3)+\\mathrm{Tr}((\\rho^\\Gamma)^3)$, and the paper proves $x_S\\le 2x_8$ with $x_8=\\mathrm{Tr}(\\rho_A\\,\\mathrm{Tr}_B\\rho^2)$ for every separable state, giving inequality (47). The proof uses positivity of the reduction map $(I\\otimes R)(\\rho)=\\rho_A\\otimes I_B-\\rho\\ge0$ and adds the criterion for $\\rho$ and $\\rho^\\Gamma$. Benchmarked on Werner states, the criterion detects entanglement for $d=3$ at $p>10^{-1/3}$, compared with $p>1/2$ for second-order correlations.","pith_inferences":["The explicit inversion formulas are written for $d\\ge3$; since the Weingarten denominators contain $d^2-4$, applying the method to qubits (the most common experimental platform) requires a separate limiting or reconstructing procedure that the paper does not spell out.","A natural next step is to search for higher-order invariants from four- and five-point correlations; the paper shows the third-order system reduces cleanly to a $10\\times10$ matrix, and the pattern of equality classes suggests the linear-algebraic structure persists at higher orders.","Because the criterion combines $\\rho$ and $\\rho^\\Gamma$, it may be worth benchmarking against bound-entangled states in $3\\times3$, where partial transposition is positive; the paper does not test this, but the inequality is not manifestly restricted to NPT states.","The use of unitary designs instead of full Haar sampling could reduce the experimental overhead; the paper notes averaging can be simplified to sampling over an appropriate unitary design, so a concrete design construction would make the protocol directly implementable."],"forward_implications":["A separable bipartite state must satisfy $x_S\\le 2x_8$; observing $x_S>2x_8$ from randomized-measurement data certifies entanglement without reconstructing the full state.","In dimension 3 the third-order criterion lowers the Werner-state detection threshold from $p>1/2$ to $p>10^{-1/3}$, a concrete regime where second-order spectral criteria fail.","The same linear-inversion scheme estimates all marginal purities of $N$-partite states with arbitrary local dimensions, extending the earlier randomized-measurement purity protocol.","The criterion is built from positivity of the reduction map, so it is locally unitary invariant and requires no calibration of a specific entangled measurement basis."],"supporting_citations":[{"why":"Supplies the Weingarten integration formula that turns averaged products of local measurement outcomes into invariant traces.","marker":"[27]"},{"why":"Establishes the randomized-measurement purity protocol that this paper generalizes to third order and to unequal local dimensions.","marker":"[23]"},{"why":"Proves positivity of the reduction map, the source of the separability inequality used in the third-order criterion.","marker":"[20]"},{"why":"Provides the basic purity inequality for separable states underlying the second-order criterion and the derivation of Eq. (29).","marker":"[2]"},{"why":"Defines the randomized-measurements framework and the inner-loop/outer-loop experimental procedure adopted here.","marker":"[29]"},{"why":"Shows that moments of a quantum state can be estimated by averaging single-copy measurement outcomes over a random unitary, the conceptual basis of the protocol.","marker":"[22]"}],"fun_headline_variants":["Third-order invariants catch more Werner entanglement","Randomized measurements reveal third-order entanglement criteria","Higher-order invariants detect Werner states below p=0.5","Third-order twirling data boosts entanglement detection","New criterion catches Werner states for p > 10^-1/3"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The protocol's correctness rests on the experimental estimate of the Haar-averaged three-point correlations: if the finite sample of random local unitaries does not faithfully represent the uniform average (or a unitary design), the $10\\times10$ inversion will not return the true invariants; the qubit case $d=2$ is also left without an explicit nonsingular formula.","fun_headline_variants_meta":{"raw":{"variants":["Third-order invariants catch more Werner entanglement","Randomized measurements reveal third-order entanglement criteria","Higher-order invariants detect Werner states below p=0.5","Third-order twirling data boosts entanglement detection","New criterion catches Werner states for p > 10^-1/3"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000561,"raw_usage":{"total_tokens":2627,"prompt_tokens":873,"completion_tokens":1754,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":489,"completion_tokens_details":{"reasoning_tokens":1677}},"tokens_in":489,"tokens_out":1754,"duration_ms":13752,"temperature":1.0,"reasoning_tokens":1677,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:55:16.476657+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Prepare a two-qutrit Werner state with $p$ in the interval $(10^{-1/3}, 1/2)$, estimate $x_S$ and $x_8$ from randomized measurements, and compare with values from full tomography: the claim predicts $x_S>2x_8$ for every such $p$, so finding a $p$ in that range where the inequality still holds would falsify the criterion's improvement.","supporting_citations":[{"cited_title":"Mezzadri, Notices of the American Mathematical Society54, 592 (2007)","cited_arxiv_id":null,"evidence_quote":"Supplies the Weingarten integration formula that turns averaged products of local measurement outcomes into invariant traces."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves positivity of the reduction map, the source of the separability inequality used in the third-order criterion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the basic purity inequality for separable states underlying the second-order criterion and the derivation of Eq. (29)."},{"cited_title":"Życzkowski, K","cited_arxiv_id":null,"evidence_quote":"Defines the randomized-measurements framework and the inner-loop/outer-loop experimental procedure adopted here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that moments of a quantum state can be estimated by averaging single-copy measurement outcomes over a random unitary, the conceptual basis of the protocol."}],"review_version":1}