{"id":"7e951ce5-a64a-4957-9fbe-89b069ad20e9","arxiv_id":"2502.06986","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every entangled measurement can be witnessed through swap steering, and every rank-one projective entangled measurement can be detected in a device-independent star network.","lead":"This paper shows that entanglement inside a joint quantum measurement can be caught without trusting the measurement device, using quantum networks. The result covers every entangled measurement with one trusted side, and every rank-one projective entangled measurement with no trusted devices at all.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof gap: Eq. (9) evaluates the wrong post-swap state; the reduced state after entanglement swapping is E_b^T, so the witness must be applied to the transposed element, requiring an explicit repair in both the 1SDI and DI sections.","rationale":"The central claim is that any entangled measurement can be witnessed in a 1SDI way and any rank-one projective entangled measurement in a DI way. The construction of the swap-steering witness genuinely proves soundness: if Bob's measurement is separable then every conditional state on Alice is separable for arbitrary sources, so -Tr(W σ_b)≤0. The missing step is not a false positive. The load-bearing problem is rather completeness: the explicit quantum violation in Eq. (9) uses the wrong reduced state. Entanglement swapping with the standard maximally entangled state produces the transpose of the POVM element on Alice's side. Since the transpose of an entangled operator is entangled, the theorem is repairable by choosing the witness for E_b^T (or transposing the witness). But as written, the derivation does not establish S>0 for every entangled measurement; there are concrete complex entangled projectors for which the displayed W fails. The same oversight affects the DI argument, though Gisin's theorem makes the repair immediate. I agree with the reader's weakest_assumption; this is exactly the transpose gap. The verdict CONDITIONAL is appropriate; no change is needed, but the authors must correct the post-swap state and the witness choice in both sections.","tokens_in":8916,"tokens_out":39988,"duration_ms":361045,"concrete_test":"For N=2, choose the entangled rank-one measurement element E_b=|ψ⟩⟨ψ| with |ψ⟩=(|00⟩+i|11⟩)/√2 and a valid witness W=αI-|ψ⟩⟨ψ| with α chosen so that Tr(Wσ)≥0 for all separable σ. Compute the quantum value of S in Eq. (8) with maximally entangled sources using the correct post-swap state E_b^T. Show that Tr(W E_b^T)≥0, so the manuscript's Eq. (9) would predict no violation, while Tr(W^T E_b^T)=Tr(W E_b)<0; then replace W by W^T and verify S>0. This confirms the gap is in the stated proof, not in the underlying claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing defect is in the proof of the 1SDI claim. With maximally entangled sources, after Bob obtains outcome b the unnormalized state on Alice's N systems is Tr_B[(I_A⊗E_b)|Φ+⟩⟨Φ+|^⊗N] = E_b^T/d^N, not E_b/d^N as the manuscript states before Eq. (9). Therefore Eq. (9) computes max_b (Tr E_b/d^N) Tr(-W E_b), whereas the actually measured functional is max_b (Tr E_b/d^N) Tr(-W E_b^T). For a general witness W constructed from E_b via Fact 1, W need not detect E_b^T; e.g., for E_b=|ψ⟩⟨ψ| with |ψ⟩=(|00⟩+i|11⟩)/√2, E_b^T is orthogonal to E_b and a witness for E_b can have Tr(W E_b^T)≥0. The claim survives because E_b^T is entangled iff E_b is, so Fact 1 applied to E_b^T supplies a valid witness (equivalently use W^T), but this step is absent. The same transposition issue appears in the DI argument before Fact 3, where the post-swap state is |ψ*⟩⟨ψ*|; the conclusion still holds by Gisin's theorem applied to the complex-conjugated state, but the text must say so.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript classifies composite quantum measurements into separable and entangled measurements, introduces a notion of entanglement witness for measurements via the Hahn-Banach theorem, and gives explicit witnesses for the Bell-basis measurement. It then claims that every entangled measurement can be witnessed in a one-sided device-independent way using the swap-steering scenario: for any entangled measurement element, a suitable witness yields a positive value of the functional S, while SOHS models give at most zero. Finally, it argues that every rank-one projective entangled measurement can be detected device-independently in a star network, using Gisin's theorem for bipartite measurements and generic nonlocality for multipartite ones. The main technical proofs are deferred to an appendix.","tokens_in":9198,"tokens_out":11153,"duration_ms":96767,"significance":"If the proof gaps are repaired, the paper's results are significant: they provide a unified reduction from witness-based detection of entangled measurements to state witnesses, and they show that swap steering detects any entangled measurement rather than only special families. The 1SDI and DI constructions are conceptually clean, and the explicit local decomposition of the Bell-basis witness is practically useful. The proofs of the SOHS and local bounds (Fact 2 and Fact 3) are standard but correctly identify the relevant bounds. The main caveat, which is in line with the state-witness literature, is that the witness inequalities are tailored to the target measurement, so the results establish existence of a detection scheme rather than a single universal inequality.","major_comments":[{"comment":"The unnormalized operator on Alice's systems after Bob's outcome b is E_b^T/d^N, not E_b/d^N: for maximally entangled sources, Tr_B[(I_A ⊗ E_b)|φ_d^+><φ_d^+|^⊗N] = E_b^T/d^N. Therefore the quantity computed from the correlations is max_b (Tr E_b/d^N) Tr(-W E_b^T), whereas Eq. (9) evaluates Tr(-W E_b). A witness W constructed from E_b via Fact 1 need not satisfy Tr(-W E_b^T)>0. The proof must be repaired by applying Fact 1 to E_b^T, or equivalently by using W^T, which is valid because E_b^T is entangled iff E_b is; without this step the conclusion S>0 does not follow from the stated assumptions.","section":"One-sided device-independent witness, Eq. (9)"},{"comment":"For a rank-one projective element E_b=|ψ_b><ψ_b|, the state on the external parties after the swap is |ψ_b^*><ψ_b^*|, the complex conjugate in the standard basis, not |ψ_b><ψ_b|. Thus Gisin's theorem and the generic-nonlocality results of [30,31] must be applied to |ψ_b^*>; the conclusion still holds because |ψ_b^*> is entangled iff |ψ_b> is, but the text should say this explicitly.","section":"Device-independent witness, paragraph before Fact 3"}],"minor_comments":[{"comment":"The phrase 'we do not have a prove' should be 'we do not have a proof'.","section":"Witnesses, Eq. (4)"},{"comment":"The term 'unextentible product bases' should be 'unextendible product bases'.","section":"Introduction"},{"comment":"The phrase 'informational complete' should be 'informationally complete'.","section":"Quantum measurement tomography"},{"comment":"In the definition of Γ, the second argument is written as ρ_λn but should be ρ_λN.","section":"Appendix A, Eq. (A3)"},{"comment":"The decomposition of W should state explicitly that the coefficients c_ij... are real and that the local projectors are drawn from a tomographically complete set, so that the decomposition is without loss of generality.","section":"Witnesses, Eq. (3)"},{"comment":"The sentence citing [30,31] for any rank-one projective entangled measurement should spell out that for N>2 the relevant inequality is a multipartite Bell inequality from generic nonlocality, not CHSH.","section":"Device-independent witness"}],"recommendation":"major_revision","confidential_remarks":"The transposition gap in Eq. (9) and in the DI section is the only substantive technical issue I found; it is local and easily repaired by applying the witness construction to the transposed element. The paper fits the journal's scope and the central claims are defensible once this repair is made explicit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core is a clean idea: export state-entanglement machinery to measurements via swap steering. That reduction is new as far as I know, and it is presented so an experimentalist can use it. The explicit local-state witness for the Bell measurement in Eq. (5) is a nice touch, and the paper is honest about the optimality conjecture there. I also like the star-network construction: it turns any Bell inequality into a measurement witness, and the claim that every rank-one projective entangled measurement is detectable is the kind of universal statement that makes the paper worth reading.\n\nThe main problem is a real but fixable transposition error. After Bob obtains outcome b, the unnormalized state on Alice's side is E_b^T / d^N, not E_b / d^N. So Eq. (9) evaluates Tr(-W E_b), while the experiment gives Tr(-W E_b^T). The fix is straightforward: E_b is entangled iff E_b^T is, so apply Fact 1 to the transposed element (or use W^T). The same slip appears in the DI section, where the post-swap state is the complex conjugate of the measurement element. Since complex conjugation preserves entanglement, Gisin's theorem still applies, but the text needs to say so.\n\nThere is a second, subtler gap in the DI proof. Fact 3 bounds the functional E only for correlations admitting the local model (11). To certify measurement entanglement, you need the stronger statement that any separable measurement, even with entangled independent sources, gives E <= 0. This is true: a separable POVM element is a convex mixture of product operators, so for each b the normalized conditional distribution is a mixture of product distributions across the Alices, hence Bell-local. But that argument is missing. A referee should ask for it.\n\nMinor issues: Eq. (7) could be clearer about the SOHS convention, and the proof of Fact 2 quietly assumes the source hidden states are product across A and B, which is fine but should be stated.\n\nOverall the construction is conceptually sound and the gaps are repairable. This deserves a serious referee rather than a desk reject. I would send it back for major revision with clear requests: fix the transpose in both places and add the separable-measurement bound in the DI section.","headline":"A genuinely useful transfer of state witnesses to measurement witnesses, with two specific proof gaps that are repairable; worth a serious referee.","tokens_in":736,"tokens_out":1351,"would_cite":true,"duration_ms":253693,"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":"Any entangled measurement can be detected without trusting the measurement device, and every rank-one projective entangled measurement with no trust at all.","keywords":["entangled measurements","measurement entanglement witnesses","swap steering","one-sided device-independent","device-independent detection","star network","network nonlocality","entanglement swapping"],"falsifier":"For a concrete check, take the Bell-basis measurement and the witness $W'_{\\mathrm{BM}}$ of Eq. (5), compute the swap-steering value with the actual post-measurement state $E_b^T$ rather than $E_b$, and see whether it is positive. If there is any entangled $E_b$ for which this value is non-positive while $\\operatorname{Tr}(W'_{\\mathrm{BM}}E_b) < 0$, the proof's Eq. (9) does not certify that measurement as written; more generally, testing every entangled measurement element against the witness used in the construction would settle whether the claimed one-sided device-independent detection holds.","tokens_in":8694,"feed_emoji":"⚛️","tokens_out":10772,"duration_ms":88130,"temperature":0.7,"pith_summary":"The paper seeks to put entanglement in composite quantum measurements on the same footing as entanglement in quantum states: a measurement is entangled if at least one of its elements is non-separable, and a witness is an operator that returns negative on the entangled element and non-negative on every separable element. The paper claims that such witnesses always exist and can be evaluated using only trusted states and untrusted measurement statistics. It then shows that every entangled measurement produces an advantage in swap steering, a network scenario with no inputs, which makes the witnessing one-sided device-independent: the measurement itself need not be trusted. For rank-one projective measurements, it shows that every entangled example produces network nonlocality in a star network, so the detection can be fully device-independent. If these claims hold, entanglement in measurements is not just a theoretical classification but a certified resource in quantum networks.","feed_headline":"Any entangled measurement can be certified without trust","feed_subtitle":"Swap steering catches every entangled measurement; rank-one projective ones need no trusted devices.","key_machinery":"The load-bearing objects are two network functionals built from a measurement witness. A measurement entanglement witness is an operator $W$ satisfying $\\min_i \\operatorname{Tr}(W E_i) < 0$ for an entangled measurement and $\\min_i \\operatorname{Tr}(W M_i) \\ge 0$ for every separable measurement; by Hahn-Banach it exists for every entangled measurement. In swap steering, this witness is rewritten as $W = -\\sum \\beta_{i_1\\ldots i_N} \\tau_{i_1} \\otimes \\cdots \\otimes \\tau_{i_N}$ and the functional $S = \\max_b \\sum \\beta_{i_1\\ldots i_N} p(0,b|i_1\\ldots i_N)$ is shown to be at most zero under separable-outcome hidden-state models; using maximally entangled sources, Bob's measurement element is swapped to Alice, making the quantum value positive when the element is entangled. In the star network, any Bell inequality $B \\le \\beta_{\\mathrm{LHV}}$ is converted into $\\mathcal{E} = \\max_b [\\sum c_{a,x} p(a,b|x) - \\beta_{\\mathrm{LHV}} p(b)] \\le 0$ for local correlations; since the post-measurement state at the outer parties is the rank-one projector $E_b$, every entangled such projector violates some Bell inequality, yielding the device-independent detection.","core_discovery":"The central claim is that entanglement of a joint measurement can be detected without trusting the measurement device. Using the Hahn-Banach theorem, the paper defines a measurement witness $W$ with $\\min_i \\operatorname{Tr}(W E_i) < 0$ for an entangled measurement and $\\min_i \\operatorname{Tr}(W M_i) \\ge 0$ for every separable measurement, and shows how to decompose $W$ into local trusted-state preparations. In the swap-steering scenario, with $N$ independent maximally entangled sources and tomographically complete measurements on the trusted side, the functional $S = \\max_b \\sum \\beta_{i_1\\ldots i_N} p(0,b|i_1\\ldots i_N)$ has separable-outcome hidden-state bound zero, while the quantum value is positive exactly when the measurement element $E_b$ is entangled, giving one-sided device-independent detection. The star-network argument takes any Bell inequality $B \\le \\beta_{\\mathrm{LHV}}$ and builds $\\mathcal{E} = \\max_b [\\sum c_{a,x} p(a,b|x) - \\beta_{\\mathrm{LHV}} p(b)] \\le 0$ for local correlations; when the swapped state $E_b$ is entangled, the same state violates the original Bell inequality, so $\\mathcal{E} > 0$, giving device-independent detection for every rank-one projective entangled measurement.","pith_inferences":["A rigorous rewrite of the swap-steering proof would replace $E_b$ by $E_b^T$ in Eq. (9); since transposition preserves separability, the same conclusion should follow from the transposed witness, so the stated theorem likely survives this correction.","Because the star-network construction starts from any Bell inequality, it suggests a practical certification pipeline: pick a Bell inequality violated by the transposed or original measurement element, build the network functional, and certify the measurement from observed correlations alone.","An immediate experimental target would be the four-correlation witness for the Bell-basis measurement, which needs only product states in the computational and conjugate bases; a photonic swap-steering experiment could demonstrate the one-sided device-independent detection."],"forward_implications":["Any entangled measurement can be certified one-sided device-independently: the measurement is untrusted, while the sources and one party's measurements are trusted.","Every rank-one projective entangled measurement can be certified fully device-independently in the star network, where no device is trusted.","The construction turns every measurement witness and every Bell inequality into a network functional, so improved witnesses or Bell inequalities directly improve the certification.","The Bell-basis witness in the paper requires only four local product-state preparations, which makes the witness practical with simple state rotations.","The result identifies entangled measurements as the resource behind input-free network steering, connecting two previously separate notions of nonclassicality."],"supporting_citations":[{"why":"Supplies the theory of entanglement witnesses for states, which Fact 1 extends directly to measurements.","marker":"[2]"},{"why":"Provides the extensive witness-construction toolkit the paper draws on for measurement witnesses.","marker":"[3]"},{"why":"Introduces the swap-steering scenario and the separable-outcome hidden-state model that gives the bound $S \\le 0$ in Fact 2.","marker":"[16]"},{"why":"Shows every bipartite entangled state is network-steerable without inputs, supporting the swap-steering detection method.","marker":"[25]"},{"why":"Defines the bilocality/star-network local model used in Fact 3 for the device-independent witness.","marker":"[26]"},{"why":"States that every pure bipartite entangled state violates a Bell inequality, the key step from entangled measurement elements to network nonlocality.","marker":"[28]"},{"why":"Provides the CHSH inequality, the concrete Bell inequality that every rank-one bipartite projective entangled measurement is claimed to violate.","marker":"[29]"},{"why":"Extends the conclusion to generic entangled states, used to cover arbitrary rank-one projective entangled measurements in the multipartite star network.","marker":"[30]"},{"why":"Completes the generic-nonlocality proof, supporting the extension in the previous step.","marker":"[31]"}],"fun_headline_variants":["Entangled measurements exposed without trust","Quantum networks certify any entangled measurement","Detect measurement entanglement with no trusted devices","Swap steering reveals every entangled measurement"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the state Alice ends up with after entanglement swapping is the measurement element $E_b$ itself, when the calculation gives its transpose $E_b^T$; the proof needs the true but unstated fact that transposing a measurement element does not change whether it is entangled.","fun_headline_variants_meta":{"raw":{"variants":["Entangled measurements exposed without trust","Quantum networks certify any entangled measurement","Detect measurement entanglement with no trusted devices","Swap steering reveals every entangled measurement"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000197,"raw_usage":{"total_tokens":1391,"prompt_tokens":997,"completion_tokens":394,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":613,"completion_tokens_details":{"reasoning_tokens":345}},"tokens_in":613,"tokens_out":394,"duration_ms":3686,"temperature":1.0,"reasoning_tokens":345,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T14:09:05.911717+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a concrete check, take the Bell-basis measurement and the witness $W'_{\\mathrm{BM}}$ of Eq. (5), compute the swap-steering value with the actual post-measurement state $E_b^T$ rather than $E_b$, and see whether it is positive. If there is any entangled $E_b$ for which this value is non-positive while $\\operatorname{Tr}(W'_{\\mathrm{BM}}E_b) < 0$, the proof's Eq. (9) does not certify that measurement as written; more generally, testing every entangled measurement element against the witness used in the construction would settle whether the claimed one-sided device-independent detection holds.","supporting_citations":[{"cited_title":"Gühne and G","cited_arxiv_id":null,"evidence_quote":"Provides the extensive witness-construction toolkit the paper draws on for measurement witnesses."},{"cited_title":"Sarkar, Network quantum steering enables random- ness certification without seed randomness, Quantum 8, 1419 (2024)","cited_arxiv_id":null,"evidence_quote":"Introduces the swap-steering scenario and the separable-outcome hidden-state model that gives the bound $S \\le 0$ in Fact 2."},{"cited_title":"Sarkar, Witnessing network steerability of ev- ery bipartite entangled state without inputs (2024), arXiv:2406.11994 [quant-ph]","cited_arxiv_id":null,"evidence_quote":"Shows every bipartite entangled state is network-steerable without inputs, supporting the swap-steering detection method."},{"cited_title":"Branciard, N","cited_arxiv_id":null,"evidence_quote":"Defines the bilocality/star-network local model used in Fact 3 for the device-independent witness."},{"cited_title":"Popescu and D","cited_arxiv_id":null,"evidence_quote":"Extends the conclusion to generic entangled states, used to cover arbitrary rank-one projective entangled measurements in the multipartite star network."},{"cited_title":"generic quantum nonlocality","cited_arxiv_id":null,"evidence_quote":"Completes the generic-nonlocality proof, supporting the extension in the previous step."}],"review_version":1}