{"id":"29fc49f0-3ad2-47d5-9393-ea6134d7a791","arxiv_id":"2507.11180","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Quantum state verification is demonstrated as a feedback tool that guides preparation of a three-qubit nonstabilizer W state, achieving high fidelity with 100x fewer samples than tomography.","lead":"This paper shows experimentally that quantum state verification, a cheap way to check a quantum state, can also guide the preparation of the state itself. The authors use it to tune a three-qubit W state with far fewer measurements than full tomography, pointing toward real-time quantum device control.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"QSV fidelity estimate is inconsistent with QST at ~6σ; the unbiased-estimation assumption behind Eq. (10) is not validated.","rationale":"The reader's weakest assumption is the unbiasedness of the fidelity estimator, and I agree that this is the main soft spot. The paper has genuine strengths: a nonstabilizer three-qubit W state, a 9-setting QSV protocol, and an order-of-magnitude resource reduction in measurement settings; the two-qubit benchmark in Appendices B/C is a useful check. However, the headline comparison between QSV (97.07±0.26%) and QST (98.58±0.12%) is not just slightly off; it is a ~6σ discrepancy when both are treated as estimates of the same ensemble. The error bars in Eq. (11) are statistical only and do not include calibration bias. Since Eq. (10) inverts the passing frequency through a fixed ν, any miscalibration of the 9 settings directly shifts the reported fidelity. The paper does not provide the calibration data needed to rule this out, nor does it explain the 1.5 percentage-point gap. A concrete calibration or interleaved comparison would settle whether the discrepancy is due to operator miscalibration or state drift. Either way, the quantitative claim in the abstract is currently not fully supported, so the conditional verdict stands until the raw-data check is performed.","tokens_in":12335,"tokens_out":7431,"duration_ms":98227,"concrete_test":"Publish or re-analyze the raw coincidence counts for the nine QSV settings and the QST data taken on the same prepared W3 state. From the QST-reconstructed density matrix ρ_QST, compute the ideal expected pass rate p_exp=Tr(Ω_Hom ρ_QST) with Ω_Hom from Eq. (5), and compare to the observed f. If |p_exp-f| > 3√(f(1-f)/N) ≈ 0.36%, the unbiased-estimation assumption fails. To distinguish calibration error from drift, repeat with QSV and QST settings interleaved on the same copy stream, and also calibrate each implemented POVM element (e.g., via QST on prepared eigenstates) to compute the effective Ω; if its second eigenvalue is not 1-ν=1/2, Eq. (10) must be replaced by a calibrated inversion.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central prescriptive claim depends on Eq. (10), F=(f-(1-ν))/ν, which estimates average fidelity without tomography. This estimator is unbiased only if the implemented measurement settings exactly realize the homogeneous operator Ω_Hom of Eq. (5) with spectral gap ν=1/2. The reported numbers contradict that assumption. With F_QSV=0.9707 and N=10^4, the passing frequency is f=0.98535. If the same ensemble had the QST fidelity F_QST=0.9858, the expected frequency under ideal Ω_Hom would be f_exp=νF_QST+(1-ν)=0.9929. The difference f_exp-f_obs=0.00755 is about 6.2 standard deviations of the binomial frequency. This is far outside the claimed 'consistent' agreement. Either the implemented POVM elements deviate from Ω_Hom (biasing Eq. (10)) or the state changed between the QSV and QST runs; both possibilities break the quantitative reading of the fidelity estimate. Because the abstract explicitly promises 'quantitative fidelity indicators' that replace full tomography, this unvalidated unbiasedness is the load-bearing weakness.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports an experimental implementation of a modified homogeneous quantum state verification (QSV) protocol for a three-qubit nonstabilizer W state, with the central claim that QSV can be used not only to certify but also to actively guide state preparation. Using nine measurement settings and 10^4 samples, the authors report a QSV-estimated fidelity of 97.07(±0.26)%, which they describe as consistent with a quantum state tomography (QST) fidelity of 98.58(±0.12)% obtained with roughly 10^6 measurements. The paper also presents two-qubit benchmarking, a scaling analysis of the required number of tests versus infidelity, and a comparison of QSV-guided and QST-guided preparation.","tokens_in":12583,"tokens_out":5079,"duration_ms":64299,"significance":"If the central claim is sound, the paper makes a useful conceptual advance: QSV is elevated from a post-hoc diagnostic to a prescriptive tool that can inform state preparation with far fewer resources than full tomography. The experimental demonstration is substantive, including an independent QST cross-check, a nonstabilizer target state, and a comparison of QSV-guided and QST-guided preparation in the two-qubit case. However, the quantitative agreement between the QSV fidelity estimate and the QST result is not statistically consistent, and the scaling exponent is presented without uncertainty. These issues directly affect the paper's main claims and require careful revision.","major_comments":[{"comment":"The reported QSV estimate of 97.07(±0.26)% and the QST value of 98.58(±0.12)% differ by 1.51 percentage points, which is about 5.3 combined standard deviations. This is not 'consistent' as claimed. More concretely, if the same ensemble had the QST fidelity 0.9858 and the implemented operator were exactly the ideal homogeneous operator of Eq. (5) with spectral gap ν = 1/2, the expected passing frequency would be f_exp = νF + (1-ν) = 0.9929. The observed frequency f = 0.98535 is about 6.3 binomial standard deviations below this value. The unbiasedness of the estimator in Eq. (10) is the load-bearing assumption behind the 'quantitative fidelity indicator' claim, so the discrepancy indicates either a deviation of the implemented POVM from Ω_Hom or a change of state between the QSV and QST runs. The manuscript must provide a systematic-error analysis, a same-ensemble comparison, or a revised claim that does not assert quantitative agreement.","section":"Results, Eq. (10)"},{"comment":"The claimed favorable scaling, stated as O(ε^-1.39), is obtained from a fit but no confidence interval, fit range, or goodness-of-fit is reported. The same is true for the two-qubit scaling exponents t = 1.45, 1.37, and 1.38 in Fig. S5. Since the claim of outperforming the standard quantum limit O(ε^-2) rests on this exponent, the authors should either provide a proper least-squares fit with parameter uncertainties and a defined fitting window, or explicitly label the exponent as an illustrative value obtained from a limited range rather than a quantitative result.","section":"Fig. 4(a) and scaling text"}],"minor_comments":[{"comment":"The hypothesis test uses n in the sums but N for the number of tests; the notation should be made consistent.","section":"Eq. (7)"},{"comment":"Several reference numbers are duplicated or inconsistent (for example, [2] is used for both Dicke and Pallister et al., and [3] for both Paris and Řeháček and Zhu and Hayashi), which makes it difficult to identify the cited works.","section":"References"},{"comment":"The sentence 'These features posits QSV' should be 'These features posit QSV'.","section":"Summary"},{"comment":"The legend entries 'verification' and 'estimation' are not fully explained; please clarify which symbols correspond to the certified upper bound and which to the direct estimate.","section":"Fig. 4(a) caption"},{"comment":"The term 'real-time' is used to describe the feedback loop, but the described procedure appears to involve manual adjustment of wave-plate angles guided by QSV results rather than automated closed-loop control; consider qualifying the wording to avoid overstating the degree of automation.","section":"Introduction and Fig. 1"}],"recommendation":"major_revision","confidential_remarks":"The skeptical concern about the QSV-QST discrepancy lands. The central idea is promising, but the quantitative agreement claim must be either supported with a systematic-error analysis or revised downward to a lower-bound statement. The self-citation of Ref. [42] for the modified homogeneous protocol is not by itself problematic because the experimental implementation and the fidelity measurement are independent of that theoretical derivation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick take: this is an honest experimental advance, and the core claim—that QSV measurements can actively guide state preparation rather than merely verify after the fact—holds up. The paper demonstrates exactly that with a three-qubit W state: nine settings and 10^4 samples buy a fidelity estimate of 97.07(±0.26)%, while QST needs 64 settings and ~10^6 measurements to report 98.58(±0.12)%. That resource reduction is real, and the two-qubit benchmark in the supplemental material strengthens the case that QSV-assisted tuning matches QST-assisted tuning in final state quality.\n\nWhat's actually new: the first experimental use of QSV as a preparation loop, not just post-hoc verification. The protocol itself comes from the authors' own earlier work (Ref. 42), so the novelty is in the application and the W-state demonstration, not in the verification formalism. They are upfront about the protocol's origin, and the experimental demonstration is genuinely independent of the theory.\n\nNow the soft spots, in proportion. The advertised \"consistent\" agreement between QSV and QST is not what the numbers show. With F_QSV = 97.07% and N = 10^4, the passing frequency is f = 0.98535. If the true fidelity were the QST value 98.58%, under the ideal homogeneous operator in Eq. (5) the expected frequency is 0.9929. That is a ~6 sigma gap. Either the implemented measurements deviate from the ideal operator (biasing Eq. (10)) or the state drifted between the QSV and QST runs. Either way, the word \"consistent\" is doing load-bearing work it cannot support, and the abstract's promise of \"quantitative fidelity indicators\" needs a caveat: the estimate is only as good as the exact implementation of the verification operator. This is a fixable issue—report the raw passing statistics per setting, check the realized operator against the ideal, and either correct the number or weaken the consistency claim.\n\nMinor points: the scaling exponent t = 1.39 is a fit over a small range and is not a parameter-free prediction, so treat it as suggestive. The real-time closed-loop language is aspirational—this is off-line tuning with QSV feedback on the timescale of human intervention, not low-latency feedback. That is fine, but the text should say so.\n\nBottom line: the central demonstration is solid, the quantitative agreement claim is not. The paper deserves a serious referee, but only after the authors address the 6-sigma discrepancy and the unbiased-estimation assumption behind Eq. (10). I would not cite the headline fidelity number as a validated independent estimate until that is reconciled.\n\nRecommendation: send it to review, with a request that the fidelity-consistency issue be resolved before acceptance.","headline":"QSV-guided preparation is a real experimental step forward, but the trustworthiness of the headline fidelity number needs a correction or a caveat before it can be taken at face value.","tokens_in":13134,"tokens_out":697,"would_cite":true,"duration_ms":10492,"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":"Quantum state verification can steer state preparation, not just check it, cutting measurement cost by orders of magnitude.","keywords":["quantum state verification","nonstabilizer states","W state","quantum state preparation","fidelity estimation","homogeneous QSV operator","quantum state tomography","entangled photon pairs"],"falsifier":"Compare the empirical passing frequency on states independently characterized by quantum state tomography and check whether $f = 1 - \\nu(1 - F_{\\mathrm{QST}})$ holds. With the reported numbers, the QSV estimate (97.07%) lies about 1.5 points below the QST value (98.58%), roughly 5 standard deviations, so a direct measurement of the realized operator's spectral gap under the actual wave-plate settings would settle whether the estimate is biased.","tokens_in":12150,"feed_emoji":"⚛️","tokens_out":5975,"duration_ms":70438,"temperature":0.7,"pith_summary":"Quantum state verification is usually a post-hoc pass/fail test applied only after a state has already been carefully prepared. This paper aims to turn QSV into the steering wheel of preparation itself: the same measurement operator that certifies a target state also provides a direct fidelity estimate, so experimental settings can be tuned against that estimate with no full tomography. Using nine measurement settings and $10^{4}$ samples on a three-qubit nonstabilizer W state, the authors report a QSV-estimated fidelity of 97.07(±0.26)%, while full tomography on roughly $10^{6}$ measurements reports 98.58(±0.12)%. The claim is that QSV is a resource-efficient, prescriptive alternative to tomography for real-time state engineering, not merely a diagnostic.","feed_headline":"Verification steers state prep with 100x fewer samples","feed_subtitle":"Nine measurement settings and 10^4 samples yield a 97.07% fidelity estimate that matches tomography's 98.58%.","key_machinery":"The load-bearing object is the modified homogeneous QSV operator $\\Omega_{\\mathrm{Hom}}(W_3)$ of Eq. (5), a convex combination over permutations of a one-way adaptive measurement $\\Omega^{\\rightarrow}$. The adaptive measurement first projects all qubits except a preselected pair in the $Z$ basis; if no excitation is seen it tests the pair with $(XX)^+$, and if one excitation is seen it tests with $Z^+ Z^+$. Homogeneity means the same global operator is applied at every round, so the passing frequency can be inverted through the spectral gap $\\nu = 1/2$ to give a direct fidelity estimate without reconstructing the state.","core_discovery":"On the paper's own terms, the central discovery is that a modified homogeneous QSV operator can carry the full loop of preparation and verification. For the three-qubit W state, the operator $\\Omega_{\\mathrm{Hom}}(W_3) = \\frac{1}{3}\\sum_{k} P_k \\Omega^{\\rightarrow}$ built from one-way adaptive measurements has spectral gap $\\nu = 1/2$, giving sample complexity $N = 2\\epsilon^{-1}\\ln\\delta^{-1}$. Because the operator is homogeneous, the passing frequency $f$ directly estimates the average fidelity via $F = (f - (1-\\nu))/\\nu$. The experiment uses 9 measurement settings and $10^{4}$ samples to report 97.07(±0.26)% from QSV, independently matched by QST at 98.58(±0.12)% with 64 settings and about $10^{6}$ samples; a fitted scaling of $O(\\epsilon^{-1.39})$ in the low-sample region is closer to QSV's Heisenberg-limited behavior than to tomography's $O(\\epsilon^{-2})$.","pith_inferences":["Editorial: if the unbiasedness assumption survives calibration, the same passing-frequency feedback loop could be attached to low-latency platforms to re-tune sources between operations, replacing QST in iterative alignment.","Editorial: the homogeneous-operator construction may transfer to other Dicke and nonstabilizer states whenever one can enforce a constant spectral gap, converting verification operators into online fidelity estimators for a broader class of targets.","Editorial: the 1.5-point gap between QSV and QST estimates, though within the paper's claim of consistency, suggests a useful stress test: repeated comparison under deliberately misaligned wave plates would map how sensitive the estimator is to spectral-gap mismatch."],"forward_implications":["With only a few dozen tests, the protocol can certify a fidelity lower bound; with thousands of tests, it estimates the average fidelity directly, and both modes avoid full state reconstruction.","The same 9 measurement settings and 10^4 samples produce a QSV fidelity of 97.07(±0.26)%, while QST uses 64 settings and roughly 10^6 samples to give 98.58(±0.12)%, so the method offers a drastic resource reduction for the same task.","The fitted scaling $O(\\epsilon^{-1.39})$ in the low-sample regime outperforms the standard-quantum-limit scaling $O(\\epsilon^{-2})$ of tomography, so the verification advantage is visible even before the asymptotic regime.","Because the QSV result guides tuning during preparation rather than only checking afterward, the protocol is a step toward real-time feedback control of quantum sources."],"supporting_citations":[{"why":"Supplies the original QSV protocol for W and Dicke states with local operations and classical communication, which the present experiment extends.","marker":"[20]"},{"why":"Provides the modified homogeneous QSV operator with local measurements used to realize direct fidelity estimation in the experiment.","marker":"[42]"},{"why":"Builds the practical statistical model that converts verification outcomes into certified fidelity lower bounds via hypothesis testing.","marker":"[33]"},{"why":"Gives the Chernoff-Hoeffding bound used to compute significance levels and confidence statements in the verification analysis.","marker":"[46]"}],"fun_headline_variants":["Prescriptive QSV steers 3-qubit W state with 9 settings","Tomography-free prep: 10^4 samples match 10^6 tomography","Heisenberg-limited state verification without full tomography","Nine settings and 10^4 samples yield 97% fidelity estimate","Prescriptive verification cuts sample cost 100x without tomography"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire quantitative claim rests on the assumption that the nine experimentally implemented measurement settings realize the ideal homogeneous verification operator with spectral gap exactly 1/2; if the realized operator deviates, the reported fidelity estimate is biased rather than merely noisy.","fun_headline_variants_meta":{"raw":{"variants":["Prescriptive QSV steers 3-qubit W state with 9 settings","Tomography-free prep: 10^4 samples match 10^6 tomography","Heisenberg-limited state verification without full tomography","Nine settings and 10^4 samples yield 97% fidelity estimate","Prescriptive verification cuts sample cost 100x without tomography"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001061,"raw_usage":{"total_tokens":4462,"prompt_tokens":967,"completion_tokens":3495,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":583,"completion_tokens_details":{"reasoning_tokens":3404}},"tokens_in":583,"tokens_out":3495,"duration_ms":28075,"temperature":1.0,"reasoning_tokens":3404,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:14:19.497037+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the empirical passing frequency on states independently characterized by quantum state tomography and check whether $f = 1 - \\nu(1 - F_{\\mathrm{QST}})$ holds. With the reported numbers, the QSV estimate (97.07%) lies about 1.5 points below the QST value (98.58%), roughly 5 standard deviations, so a direct measurement of the realized operator's spectral gap under the actual wave-plate settings would settle whether the estimate is biased.","supporting_citations":[{"cited_title":"Liu, X.-D","cited_arxiv_id":null,"evidence_quote":"Supplies the original QSV protocol for W and Dicke states with local operations and classical communication, which the present experiment extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the modified homogeneous QSV operator with local measurements used to realize direct fidelity estimation in the experiment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Builds the practical statistical model that converts verification outcomes into certified fidelity lower bounds via hypothesis testing."},{"cited_title":"Hoeffding, Probability inequalities for sums of bounded random variables, in The Collected Works of Wassily Hoeffd- ing (Springer, 1994) pp","cited_arxiv_id":null,"evidence_quote":"Gives the Chernoff-Hoeffding bound used to compute significance levels and confidence statements in the verification analysis."}],"review_version":1}