{"id":"f71006e6-6386-40a7-b510-29a434d5706e","arxiv_id":"2412.19540","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Adaptive local measurement strategies verify the three-qubit GHZ-W entangled subspace with sample complexity about 2.248 ε^{-1} ln δ^{-1}.","lead":"This paper gives two experimental recipes for checking that three qubits really live in the subspace spanned by GHZ and W states, using only local measurements and classical communication. The better recipe needs about 2.25 times as many copies as the ideal global measurement, so it offers a practical way to certify noisy three-qubit devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2.248 sample-complexity claim is proven only for independently prepared copies; a correlated joint state can violate the bound, so the abstract's unqualified claim needs the i.i.d. qualifier.","rationale":"The central spectral-gap calculation in Proposition 3 appears correct: the matrix form, the eigenstates v1–v6, and the crossing 47/80·x = 1 − 11/15·x at x = 240/317 all check out, and the numerical comparison in Fig. 1(b) supports the analytical expression. The reader's weakest assumption correctly identifies the most load-bearing concern: the 2.248 sample complexity is only proven under independent preparation, while the paper's abstract states it without that qualifier. A concrete correlated-state counterexample shows the bound can fail in the per-copy Bad-case formalism. The other issues in the reader's rationale — the incompletely proven unverifiability claim and the ν(Ωp) inconsistency — are real but secondary: they do not invalidate the rotation strategy's validity or its spectral gap, and the ν(Ωp) statement is internally inconsistent with the sample-complexity formula that follows it, indicating a typo rather than a substantive error. The appropriate disposition remains conditional: the paper should explicitly qualify the independence assumption in the abstract and conclusions, and tighten the unverifiability discussion. Since the reader already reached a conditional verdict, no further change is recommended.","tokens_in":14321,"tokens_out":38634,"duration_ms":374972,"concrete_test":"Compute the two-copy acceptance probability of the rotation strategy on the correlated state ρ = (ρ_g⊗ρ_g + ρ_b⊗ρ_b)/2 with ε = 1/4, using ν = 141/317 and ρ_b as the worst-case fidelity-1/2 state. If the acceptance probability exceeds (1 − ν/4)^2 ≈ 0.7899, the correlated-state violation is confirmed and the abstract must restrict the claim to independent copies.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section II states the bound in Eq. (3) under the condition 'If the states are independently prepared,' but the Good/Bad definitions are per-copy reduced-state conditions (Tr[Πσ_i] ≤ 1−ε). This mismatch lets a correlated joint state evade the bound. Concretely, let ρ_g be any state in V3 and let ρ_b = (|g><g| + |ψ_b><ψ_b|)/2, where |ψ_b> is the eigenvector of the projected operator Ω̂ with maximal eigenvalue (so Tr[Πρ_b] = 1/2 and Tr[Ωρ_b] = 1 − ν/2). For the joint state ρ = (ρ_g⊗ρ_g + ρ_b⊗ρ_b)/2, each reduced copy has fidelity 3/4, i.e., the state is Bad for ε = 1/4. The protocol's two-copy acceptance probability is 1/2 + 1/2(1 − ν/2)^2. With ν = 141/317 this is ≈0.8023, while the independent-copy bound (1 − νε)^2 = (1 − 141/(4·317))^2 ≈0.7899. Thus the stated sample complexity does not hold for correlated adversarial sources. This is load-bearing because the practical value of the protocol is precisely the promised copy count; the abstract and conclusions present '2.248 ε^{-1} ln δ^{-1}' without the independence qualifier.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies verification of the three-qubit GHZ-W subspace using local measurements and one-way classical communication. It first classifies two-dimensional two-qubit subspaces into verifiable, perfectly verifiable, and unverifiable types, then constructs two adaptive verification strategies for the GHZ-W subspace: the XZ strategy and the rotation strategy. The main quantitative claim is an analytical sample-complexity bound for the rotation strategy of approximately 2.248 ε^{-1} ln δ^{-1}, obtained by computing the spectral gap of the averaged verification operator in Proposition 3. The paper also reports a numerical optimization for the XZ strategy and discusses the gap to the globally optimal entangled-measurement strategy.","tokens_in":14640,"tokens_out":11824,"duration_ms":117269,"significance":"If the main claim holds, the paper provides an experimentally realistic LOCC protocol for verifying a genuinely entangled subspace with only a constant-factor overhead relative to the global optimum. A genuine strength is Proposition 3: the symmetrization of the test operators and the resulting eigenvalue calculation are explicit and checkable, and the reported constants (240/317 and 141/317) are consistent with the intersection of 47/80 μ(X) and 1−11/15 μ(X). The two-qubit classification section is also potentially useful, though its impossibility claim needs tightening. The paper does not rely on circular fitting; the only numerically optimized quantity is the mixing probability for the XZ strategy, which is a protocol parameter and not an inferred physical quantity.","major_comments":[{"comment":"The sample-complexity guarantee N ≈ 2.248 ε^{-1} ln δ^{-1} is proven only for independently prepared copies, as Eq. (3) explicitly states, but the abstract, Eq. (49), and the conclusions present it without that qualifier. This is not a stylistic issue: a correlated joint source can violate the bound. For example, take ρ_g ∈ V3 and ρ_b with Tr[Πρ_b] = 1/2 and Tr[Ωρ_b] = 1 − ν/2, and consider ρ = (ρ_g⊗ρ_g + ρ_b⊗ρ_b)/2. Each reduced copy has Tr[Πσ_i] = 3/4, so the source is Bad for ε = 1/4, yet the two-copy acceptance probability is 1/2 + (1/2)(1 − ν/2)^2 ≈ 0.802, which exceeds the independent-copy bound (1 − ν/4)^2 ≈ 0.790. The authors should either add the independence assumption to every statement of the claimed sample complexity or analyze the correlated/adversarial scenario.","section":"Section II, Eq. (3); Abstract; Section V"},{"comment":"The claimed impossibility result for unverifiable two-qubit subspaces is not proven as stated. The text analyzes only the single test operator Ω_u = I − |τ⟩⟨τ|, shows that an entangled state |τ′⟩ in V⊥ orthogonal to |τ⟩ passes it with certainty, and concludes that no LOCC projective verification strategy exists. A complete proof must rule out any mixed strategy built from other local projective test operators that each accept V. Concretely, the authors need to show that every local projective test M with M|v⟩ = |v⟩ for all |v⟩ ∈ V also satisfies Tr[M|τ′⟩⟨τ′|] = 1, or provide an equivalent argument. Without this step, the classification theorem is stronger than the evidence presented.","section":"Section III.A, Eqs. (9)-(10)"},{"comment":"There is an internal inconsistency in the perfectly verifiable case. Equation (11) defines Ω_p = |τ0⟩⟨τ0| + |τ1⟩⟨τ1|, but the text says ν(Ω_p) = 0, which would imply that no finite sample size suffices, while Eq. (12) immediately gives N = ε^{-1} ln δ^{-1}. The intended statement is presumably that Ω_p is the projector onto the target subspace (or the corresponding projector onto its complement) and that ν(Ω_p) = 1. The location of the product states τ_i (in V or in V⊥) should also be aligned with the conventions used in Eqs. (7)-(8).","section":"Section III.B, Eq. (11) and following paragraph"}],"minor_comments":[{"comment":"In the second case of the proof, \"|b⟩ = 0\" should read \"|b⟩ = |0⟩\".","section":"Section III.A, Lemma 1 proof"},{"comment":"The term \"local subspace\" is introduced without definition; either define it or remove it.","section":"Section III.B, end of subsection"},{"comment":"The statement that including Y measurements is unnecessary is supported only by numerical optimization; either provide an analytic argument or explicitly label this as a numerical observation.","section":"Section IV.C, Remark"},{"comment":"The formula for the required number of copies would be clearer if the denominator were written as ln((1 − ν ε)^{-1}), rather than with the reciprocal logarithm notation used here.","section":"Section II, Eq. (4)"},{"comment":"The reduction to min{47/80 μ(X), 1 − 11/15 μ(X)} should mention that 1 − 13/20 μ(X) ≤ 1 − 47/80 μ(X) and 131/240 μ(X) ≤ 11/15 μ(X) for all μ(X) ∈ [0,1].","section":"Eq. (46)"}],"recommendation":"major_revision","confidential_remarks":"The core spectral-gap calculation appears sound and the rotation strategy is a genuine advance, but the unqualified sample-complexity claim in the abstract and conclusions is a load-bearing overstatement, and the impossibility proof in Section III.A is incomplete. Both issues are fixable within the manuscript's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Key takeaway: the rotation strategy for verifying the three-qubit GHZ-W subspace is the real contribution, with sample complexity 2.248 ε^{-1} ln δ^{-1}, about 2.25 times the global optimum. I checked the spectral gap derivation in Proposition 3; the 141/317 constant comes out correctly, and the symmetry averaging is clean. The XZ strategy is a useful stepping stone but the rotation strategy is the result that matters.\n\nWhat's new: the two adaptive LOCC strategies, and the three-way classification of two-qubit subspaces into verifiable, perfectly verifiable, and unverifiable. The classification is a nice conceptual map, and the unverifiable case is a genuine boundary.\n\nSoft spots, in order of severity:\n\n1. The impossibility claim for unverifiable subspaces is not fully proven. The paper analyzes the single test operator 1-|τ><τ| and shows it is fooled, but then concludes no LOCC projective strategy can verify the subspace. That needs a general argument over all LOCC projective strategies, or an explicit reduction. As written it's a sketch. The abstract then drops the 'projective' qualifier, which overstates the result.\n\n2. Section III.B has a clear internal inconsistency: for a perfectly verifiable subspace, ν(Ωp) is printed as 0, but with the intended test operator (the projector onto the target subspace) the spectral gap is 1. This looks like a typo, but it will confuse readers.\n\n3. The sample complexity claims in the abstract and conclusions omit the independence qualifier that Eq. (3) requires. The bound is proven for independently prepared copies; a correlated joint adversary can violate it. This is standard in the verification literature, but the unqualified abstract claim is too strong.\n\nNone of these undermine the central rotation-strategy result. The analytical calculation is solid, the strategies are concrete and experimentally feasible, and the two-qubit classification is a useful addition.\n\nThis paper is for the subspace-verification community and anyone designing LOCC certification protocols. It deserves a serious referee, and I'd expect it to go through with moderate revision.","headline":"A solid LOCC subspace-verification paper with a correct analytical spectral-gap computation; needs a few caveats fixed before it is used as a recipe.","tokens_in":15128,"tokens_out":4374,"would_cite":false,"duration_ms":36551,"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":"Local measurements verify the GHZ-W subspace at about 2.25 times the optimal sample cost.","keywords":["genuinely entangled subspace","GHZ-W state","quantum state verification","subspace verification","local measurements","LOCC","adaptive measurement","sample complexity"],"falsifier":"Take the verification operator $\\Omega_{\\mu^\\star}$ and project it onto the six-dimensional orthogonal complement of $V_3$; compute the largest eigenvalue of the projected operator numerically. If it is not $176/317$, the claimed spectral gap and sample complexity are wrong; equivalently, search over states at infidelity $\\epsilon$ and check whether any passes the rotation strategy with probability exceeding $(1-(141/317)\\epsilon)^N$.","tokens_in":14116,"feed_emoji":"⚛️","tokens_out":11135,"duration_ms":92250,"temperature":0.7,"pith_summary":"This paper tries to show that the three-qubit GHZ-W genuinely entangled subspace can be verified efficiently using only local measurements and one-way classical communication, with sample complexity about $2.248\\,\\epsilon^{-1}\\ln\\delta^{-1}$ copies needed to reach infidelity $\\epsilon$ at confidence level $1-\\delta$. That is only about 2.25 times the sample complexity of the globally optimal strategy, which requires an entangled measurement and is therefore hard to implement. The authors also classify every two-dimensional two-qubit subspace into three types and prove that one type, containing exactly one product state in its complement, cannot be verified by any LOCC protocol built from projective measurements. If the claims hold, practical certification of a commonly used three-qubit entangled resource becomes feasible with current technology at a modest overhead in copy count.","feed_headline":"Local measurements verify GHZ-W states at 2.25x optimal cost","feed_subtitle":"Uses only one-way local measurements, within a factor of 2.25 of the entangled-measurement ideal.","key_machinery":"The load-bearing mechanism is the one-way adaptive test operator: measure one qubit in the Pauli $X$ or $Z$ basis and, conditioned on the outcome, apply a tailored two-qubit verification test to the remaining two qubits. Measuring $Z$ leads to the operator $M_Z$, measuring $X$ leads to $M_X$, and local symmetries of $V_3$ (qubit permutations together with the local phase rotations $U_1=R_{2\\pi/3}^{\\otimes 3}$ and $U_2=R_{4\\pi/3}^{\\otimes 3}$) generate ten valid test operators. Averaging these operators produces a verification operator whose projected effective operator on the complement of $V_3$ has a spectrum computable in closed form, giving the spectral gap $141/317$. The two-qubit classification is carried by a determinant criterion: representing each two-qubit state by a $2\\times 2$ matrix and counting product states through $\\det(\\alpha+\\lambda\\beta)=0$ shows that the number of product states in a subspace equals that in its complement.","core_discovery":"On the paper's own terms, the central discovery is that the subspace $V_3=\\mathrm{span}\\{|GHZ\\rangle,|W\\rangle\\}$ admits verification strategies based on one-way adaptive local measurements. The rotation strategy, built from ten test operators generated by the local symmetries of $V_3$, achieves a spectral gap of $141/317\\approx 0.445$ with optimal measurement probability $\\mu^\\star(X)=240/317$, giving the sample complexity $N=\\lceil \\frac{317}{141}\\epsilon^{-1}\\ln\\delta^{-1}\\rceil\\approx 2.248\\,\\epsilon^{-1}\\ln\\delta^{-1}$. The simpler XZ strategy uses four test operators and achieves a spectral gap of about $0.262$, with sample complexity $N\\approx 3.817\\,\\epsilon^{-1}\\ln\\delta^{-1}$. Along the way, the paper proves that every two-dimensional two-qubit subspace contains either 1, 2, or infinitely many product states in its complement, which yields three classes: unverifiable, verifiable, and perfectly verifiable; the unverifiable class cannot be certified by any LOCC strategy using projective measurements.","pith_inferences":["This is an inference beyond the paper: the factor $2.248$ is likely near the best possible for LOCC verification of $V_3$, but the paper does not prove a lower bound, so that optimality question remains open.","This is an inference beyond the paper: because the copy-count guarantee assumes independent state preparation, an adversary supplying correlated copies could in principle defeat the bound; a formal analysis of that adversarial model would be a direct stress test of the claim.","This is an inference beyond the paper: the two-qubit classification suggests a general diagnostic that a subspace is locally verifiable exactly when its complement is spanned by product states, which could be tested in larger dimensions.","This is an inference beyond the paper: the special single-qubit states in the $X$ measurement are fixed by a parameter involving $\\sqrt{5}$, and understanding why that parameter is optimal may yield a systematic construction of analytic strategies for other symmetric subspaces."],"forward_implications":["The rotation strategy certifies the GHZ-W subspace to infidelity $\\epsilon$ and confidence $1-\\delta$ with about $2.248\\,\\epsilon^{-1}\\ln\\delta^{-1}$ copies, only a factor of roughly 2.25 above the entangled-measurement ideal.","Because the protocol uses only local projective measurements and one-way classical communication, it is implementable on current qubit platforms without entangled measurements.","The XZ strategy provides a four-setting alternative whose sample complexity is about $3.817\\,\\epsilon^{-1}\\ln\\delta^{-1}$, trading some efficiency for fewer measurement settings.","A two-dimensional two-qubit subspace whose complement contains exactly one product state cannot be verified by any LOCC strategy using projective measurements, so some entangled subspaces are fundamentally beyond local verification."],"supporting_citations":[{"why":"Introduced the subspace verification framework, including the verification operator and the pass-probability bound used in Eqs. (2)-(4).","marker":"[57]"},{"why":"Proposed the one-way adaptive measurement subroutine that each adaptive test operator in this paper is built from.","marker":"[37]"},{"why":"Demonstrated that local symmetries of a target state can generate additional test operators, the method used to reach ten operators for $V_3$.","marker":"[35]"},{"why":"Supplied the $2\\times 2$ matrix representation and determinant concurrence criterion for counting product states in two-qubit subspaces.","marker":"[63]"},{"why":"Clarified the basic geometric properties of two-dimensional two-qubit subspaces on which the three-type classification rests.","marker":"[62]"},{"why":"Provided the X-Z measurement idea that the four-setting XZ strategy directly adapts.","marker":"[67]"},{"why":"Established the GHZ-W subspace as a genuinely entangled subspace, the resource whose verification is the paper's target.","marker":"[14]"}],"fun_headline_variants":["Adaptive local measurements verify GHZ-W subspace at 2.25x optimal","GHZ-W subspace verified with one-way adaptive local tests","Local strategies near-optimal for GHZ-W subspace verification","Adaptive local strategy checks GHZ-W subspace almost optimally"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof's copy-count guarantee assumes the $N$ state copies are prepared independently; if an adversary can supply correlated copies, the stated $2.248\\,\\epsilon^{-1}\\ln\\delta^{-1}$ bound is not established.","fun_headline_variants_meta":{"raw":{"variants":["Adaptive local measurements verify GHZ-W subspace at 2.25x optimal","GHZ-W subspace verified with one-way adaptive local tests","Local strategies near-optimal for GHZ-W subspace verification","Adaptive local strategy checks GHZ-W subspace almost optimally"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000363,"raw_usage":{"total_tokens":1969,"prompt_tokens":970,"completion_tokens":999,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":586,"completion_tokens_details":{"reasoning_tokens":927}},"tokens_in":586,"tokens_out":999,"duration_ms":9759,"temperature":1.0,"reasoning_tokens":927,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T00:15:24.465698+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the verification operator $\\Omega_{\\mu^\\star}$ and project it onto the six-dimensional orthogonal complement of $V_3$; compute the largest eigenvalue of the projected operator numerically. If it is not $176/317$, the claimed spectral gap and sample complexity are wrong; equivalently, search over states at infidelity $\\epsilon$ and check whether any passes the rotation strategy with probability exceeding $(1-(141/317)\\epsilon)^N$.","supporting_citations":[{"cited_title":"Experimental Optimal Verification of Entangled States Using Local Measurements","cited_arxiv_id":null,"evidence_quote":"Introduced the subspace verification framework, including the verification operator and the pass-probability bound used in Eqs. (2)-(4)."},{"cited_title":"Theory of Quantum System Certification","cited_arxiv_id":null,"evidence_quote":"Proposed the one-way adaptive measurement subroutine that each adaptive test operator in this paper is built from."},{"cited_title":"Variational Entanglement-Assisted Quantum Process Tomography with Arbitrary Ancillary Qubits.Physical Review Letters, 129:133601, 2022","cited_arxiv_id":null,"evidence_quote":"Demonstrated that local symmetries of a target state can generate additional test operators, the method used to reach ten operators for $V_3$."},{"cited_title":"Efficient Verification of Ground States of Frustration-Free Hamiltonians.Quantum, 8:1221, 2024","cited_arxiv_id":null,"evidence_quote":"Supplied the $2\\times 2$ matrix representation and determinant concurrence criterion for counting product states in two-qubit subspaces."},{"cited_title":"Efficient Verification of Affleck-Kennedy-Lieb-Tasaki States","cited_arxiv_id":null,"evidence_quote":"Clarified the basic geometric properties of two-dimensional two-qubit subspaces on which the three-type classification rests."},{"cited_title":"Efficient Verification of Dicke States","cited_arxiv_id":null,"evidence_quote":"Provided the X-Z measurement idea that the four-setting XZ strategy directly adapts."},{"cited_title":"Absolutely Maximally Entangled States, Quantum-Maximum-Distance-Separable Codes, and Quan- tum Repeaters","cited_arxiv_id":null,"evidence_quote":"Established the GHZ-W subspace as a genuinely entangled subspace, the resource whose verification is the paper's target."}],"review_version":1}