{"id":"4f9302e5-c64f-45c9-b7b3-e0fea7abb6db","arxiv_id":"2607.15874","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A quantum measurement's statistics alone determine the minimum disturbance any implementation of it must cause, and this minimum is computable and experimentally estimable.","lead":"Quantum measurements disturb the systems they probe; this paper shows that the bare statistics of a measurement—its POVM—place a hard upper bound on how gently any implementation of it can be. The bound is computable, tight for uniform input ensembles, and estimable from partial data, which matters for quantum cryptography and randomness generation.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Finite-sample statistics can break the weighted state exclusion upper bound; Section 4.3 only addresses preparation errors, so the experimental certification claim is not yet rigorous.","rationale":"I read the paper in good faith and focused on the central claim: for any POVM and ensemble, the maximum average fidelity over compatible instruments is given by the SDP in Definition 1, with the Haar closed form of Theorem 1. The SDP formulation and strong duality are sound: the primal is feasible via the Lüders instrument, and the dual is strictly feasible by taking Y_a = y_a 1 with y_a > d. The proof of Theorem 1 via Lemma 2 appears correct; I checked the key algebraic steps, including the trace identity for Φ_+ and the construction of Y_{a,r}. The typesetting of Eq. (12)/(41) omits the factor 1/(d+1) on the first term, but the proof and the examples use the correct expression, so this is a presentational defect rather than a flaw in the argument. The reader's weakest assumption correctly identifies the finite-sample gap in the weighted state exclusion technique: Theorem 2 assumes exact conditional probabilities, and the robustness treatment in Section 4.3 addresses preparation uncertainties only. This is the most load-bearing concern because the paper advertises the technique as an experimental method to determine the bound without detector tomography, and a finite-sample downward fluctuation in the objective can make the computed value fail to be an upper bound on F_E(A). The theoretical result is unaffected, but the experimental claim needs a finite-statistics analysis or confidence-interval construction. This matches the reader's CONDITIONAL verdict, and no change is needed.","tokens_in":17665,"tokens_out":22099,"duration_ms":181460,"concrete_test":"Simulate the weighted state exclusion procedure for the noisy SIC POVM S_μ of Section 5.2 at μ=0.8, using the probe set ρ_Z={|0⟩⟨0|,|1⟩⟨1|}. First compute the exact bound F_E(S_μ,ρ_Z) by solving (23) with true probabilities. Then perform 10^4 simulated experimental runs, each with N=10^4 shots per probe state, estimating p̂(a|i) as empirical frequencies. Re-solve (23) with p̂(a|i). Record the fraction of runs in which the sample-based value falls below F_E(S_μ,ρ_Z) by more than 1%. If this fraction is non-negligible, the claimed upper bound is not valid at this sample size; the same test should be repeated with an objective modified to use one-sided upper confidence bounds on p(a|i), which should restore validity across all runs.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weighted state exclusion technique (Theorem 2, Section 4.2) defines F_E(A,ρ) as the solution of an optimization problem that uses the exact conditional probabilities p(a|i)=Tr(A_a ρ_i). In any real experiment, only finite-sample estimates p̂(a|i) are available. Since the objective in (23) is linear in these probabilities, substituting empirical frequencies can yield a computed value strictly below the true F_E(A,ρ), and hence below F_E(A). The paper's certification criterion — F_E(A,ρ,ϵ)<1 implies the measurement necessarily disturbs the states — is therefore not justified at finite sample size. Section 4.3 adds robustness only against state-preparation errors (trace distance ϵ), not against counting statistics. This is load-bearing because the paper's main practical claim is that the bound can be determined experimentally without detector tomography; without a finite-statistics guarantee, the reported procedure can overstate the tightness of the bound and falsely certify disturbance. The central mathematical result, Theorem 1, is not affected by this issue, but the experimental contribution is incomplete.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the statistical disturbance bound F_E(A), defined as the largest E-average input-output fidelity achievable by any instrument compatible with a POVM A. The bound is formulated as an SDP (Definition 1, Eq. (4) and Eq. (8)), and a closed-form expression is claimed for Haar-distributed input states (Theorem 1). The paper also proposes the weighted state exclusion technique (Theorem 2) to estimate the bound experimentally without detector tomography, including a robustness extension to state-preparation errors (Section 4.3). Examples demonstrate that the bound distinguishes noise models with identical informativeness, that disturbance can be certified with non-tomographically complete probe sets, and that the Lüders instrument is not optimal for non-Haar ensembles. A simple quantum randomness generation protocol links the certified fidelity to an eavesdropper's guessing probability.","tokens_in":17854,"tokens_out":9780,"duration_ms":81082,"significance":"If fully correct, the framework provides a tight, efficiently computable characterization of how the statistical description of a POVM constrains the disturbance of any compatible instrument, going beyond informativeness-based information-disturbance relations. Strengths include the explicit dual-feasible construction in Appendix A, the rank-based Proposition 1, the clear examples, and the public code repository. However, the experimental claim of determining the bound without detector tomography is not yet supported at finite sample size, and the closed-form theorem statement contains a factor error. The theoretical core is largely sound, but the paper needs revision before publication.","major_comments":[{"comment":"The closed-form expression is missing a factor 1/(d+1) on the summed term. From Eq. (11), \\sum_a Tr(R(H)I_a^L) = (d+S)/(d(d+1)) with S = \\sum_a [Tr(\\sqrt{A_a})]^2, so the correct value is 1/(d+1) + S/[d(d+1)], not S/d + 1/(d+1). The same error appears in Theorem 1 (Eq. (12)) and its restatement in Appendix A (Eq. (41)). The proof in Appendix A and the examples in Section 5.1 actually use the correct expression. The theorem statement and all derived formulas must be corrected.","section":"Eq. (11), (12), (41)"},{"comment":"The weighted state exclusion bound (Eq. (23)) is defined using exact probabilities p(a|i)=Tr(A_a\\rho_i). In any real experiment only finite-sample estimates \\hat p(a|i) are available. Since the objective in (23) is linear in these probabilities, substituting empirical frequencies can produce a value below F_E(A,\\rho), and consequently below F_E(A). The certification test F_E(A,\\rho,\\epsilon)<1 is therefore not justified at finite sample size. Section 4.3 only addresses preparation errors (trace distance \\epsilon), not counting statistics. To support the claimed experimental determination without detector tomography, the authors should provide finite-sample confidence intervals or a conservative estimator that remains an upper bound with high confidence.","section":"Sec. 4.2 / Theorem 2"}],"minor_comments":[{"comment":"The statement that 'throughout the northern hemisphere, the optimal value is attained by the bit-flip instrument' is not proven analytically. The inequality F_Eθ(B_flip)>F_Eθ(B_L) demonstrates suboptimality of the Lüders instrument but not optimality of the flip instrument. If the 'optimal' curves in Fig. 5 come from numerical SDP, please state this explicitly; otherwise provide a proof.","section":"Sec. 5.3"},{"comment":"The SDP for the guessing probability would benefit from more explanation. In particular, the constraint \\sum_a I_a^\\dagger(1_d)=1_d/d and the relation between the instrument elements I_a, the channel C, and the certified fidelity f should be spelled out; as written, the instrument appears not to be trace-preserving in the usual sense.","section":"Sec. 6, Eq. (36)"},{"comment":"The preparation-uncertainty parameter is given as \\epsilon=0.005 in the Figure 3 caption but as \\epsilon=0.05 in the main text. Please make these consistent.","section":"Fig. 3 caption vs Sec. 5.2"},{"comment":"Minor typographical issues: 'statsitical' in Definition 1, 'orthornormal' in Section 4.2, and the phrase in Theorem 2 'the upper bound given by F_E(A,\\rho)\\ge F_E(A)' is awkward—F_E(A,\\rho) is an upper bound on F_E(A), so the inequality should be presented as such.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The reader's conditional assessment aligns with my reading. The factor error in the main theorem is a simple but load-bearing typo. The more substantial gap is the absence of finite-sample statistics in the weighted state exclusion technique; without it, the experimental certification claim is incomplete. Both issues are fixable within the scope of the paper, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The statistical disturbance bound is a genuinely new result: for any POVM and any pure-state ensemble, the maximum average input–output fidelity over all compatible instruments is characterized by an SDP, and for the Haar ensemble it has a closed form. The proof in Appendix A is careful — the dual construction with explicit feasible variables and Lemma 2 does the job — and the examples are consistent with the intended formula. The paper also earns credit for showing that informativeness alone does not determine disturbance: loss and depolarization with the same informativeness are cleanly separated. The Lüders instrument not being optimal for latitude ensembles is a nice counterexample.\n\nThe main soft spot is the experimental claim. The weighted state exclusion technique (Theorem 2) uses exact conditional probabilities p(a|i). In an experiment you only have finite-sample estimates, and substituting them can produce a value below the true bound. That means the certification criterion — computed value < 1 implies disturbance — is not justified without a finite-statistics analysis. Section 4.3 handles preparation errors, not counting statistics. This is not a flaw in the math, but it is a gap in the advertised experimental method. Either add confidence intervals or soften the claim.\n\nAlso, Eq. (12)/(41) as typeset is missing the factor 1/(d+1) on the first term; the proof and examples use the correct formula. That is a fixable typo in a theorem statement, but it needs fixing.\n\nSection 5.3's optimality claim for the bit-flip instrument is numerical only, with no solver details or analytic certificate. That is a minor point; the qualitative conclusion does not rest on it.\n\nOverall, the core theoretical contribution is solid and deserves a serious referee. I'd send it to review. The main revision request should be the finite-statistics treatment, and fixing the formula typo.","headline":"Core Theorem 1 is new and correct; the experimental estimation technique has an unaddressed finite-statistics gap that should be fixed before publication.","tokens_in":18376,"tokens_out":4205,"would_cite":true,"duration_ms":36639,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P15","81P45"],"pacs":["03.65.Ta","03.67.-a"],"model":"deepseek-v4-flash","headline":"The bare statistics of a quantum measurement determine the least disturbance any implementation of that measurement must cause.","keywords":["statistical disturbance bound","POVM","average fidelity","Lüders instrument","weighted state exclusion","information-disturbance relation","quantum randomness generation","semidefinite programming"],"falsifier":"For a fully characterised POVM and input ensemble, solve the SDP for FE(A) and then numerically search over Choi operators of compatible instruments aiming at a higher average fidelity; the theorem says no such instrument exists. For the experimental method, apply weighted state exclusion to a characterised measurement using finite counts and test whether the reported bound ever falls below the directly computed FE(A).","tokens_in":17505,"feed_emoji":"⚛️","tokens_out":7519,"duration_ms":65400,"temperature":0.7,"pith_summary":"Quantum measurements inevitably disturb the systems they probe, but how much has seemed to depend on how the measurement is implemented. This paper defines a quantity—the statistical disturbance bound—that depends only on the outcome probabilities encoded in a positive operator-valued measure (POVM), together with the chosen ensemble of input states. It shows this bound is the largest average input–output fidelity any measurement channel compatible with those probabilities can achieve, computable as a semidefinite program; for uniformly random (Haar) input states it has the closed form (1 + (1/d)Σ Tr(√A_a)^2)/(d+1), attained by the Lüders instrument. A new technique, weighted state exclusion, lets an experimenter estimate this bound directly from observed conditional probabilities without knowing the POVM effects, and can certify measurement-induced disturbance using only partial state preparations. The result matters because it turns the bare statistics of a measurement into a quantitative guarantee about a dynamical property—state disturbance—that previously required detailed knowledge of the measurement channel.","feed_headline":"Quantum outcome statistics fix the minimum disturbance they cause","feed_subtitle":"Even without knowing the measurement's inner workings, its outcome statistics set a hard floor on disturbance.","key_machinery":"Key machinery: the ensemble operator R(E)=d∫dµ(ψ)(|ψ⟩⟨ψ|)^T⊗|ψ⟩⟨ψ|, which turns average input–output fidelity into an inner product Tr(R(E)I) with the Choi operator of an instrument. The central object is the semidefinite program of Definition 1: maximise Σ_a Tr(R(E)I_a) subject to complete positivity and the POVM constraint (Tr₂I_a)^T=A_a/d; strong duality holds. For Haar inputs, Lemma 2 characterises Y⊗1≽|φ⁺⟩⟨φ⁺| as Y≻0 with Tr(Y⁻¹)≤d, giving explicit dual solutions that attain the closed form and prove Lüders optimality. Weighted state exclusion restricts dual variables to Y_a=Σ_i q_ai ρ_i built from trusted probe states, making Tr(Y_a A_a) a linear function of measured conditional probab","core_discovery":"Central claim: for any POVM A on a d-dimensional system and any pure-state ensemble E, the largest average input–output fidelity over all compatible instruments is a value FE(A), computable as a semidefinite program. For uniformly random (Haar) inputs, FE(A) = (1 + (1/d)Σ_a Tr(√A_a)^2)/(d+1), attained by the Lüders instrument. Consequences claimed: Haar bound is 2/(d+1) iff all effects are rank one; weighted state exclusion estimates the bound from probe statistics without knowing the POVM (robust to preparation errors); loss and depolarisation with equal informativeness are distinguished; Lüders is suboptimal for non-Haar ensembles; fidelity bounds limit an eavesdropper's guessing probabili","pith_inferences":["An experimentalist applying weighted state exclusion with finite counts should treat the reported value as an estimate, not a guaranteed bound, because the paper's robustness result covers preparation errors only and does not provide finite-sample confidence intervals.","Because the bound separates loss from depolarisation, it could provide an operational test for distinguishing detector inefficiency from decoherence in a single device, without tomographic reconstruction.","The non-optimality of Lüders for latitude ensembles suggests one can tailor the post-measurement update to the prior ensemble, potentially leading to disturbance-optimal state discrimination or state-preserving measurements in applications with known input distributions.","The state-exclusion formulation may carry over to continuous-variable systems, where disturbance bounds could be expressed through analogous dual semidefinite programs, though the paper only gestures at this direction."],"forward_implications":["Any instrument that implements a POVM must have input–output fidelity no larger than FE(A) on the chosen ensemble, so bare outcome statistics lower-bound the state disturbance every implementation causes.","For Haar inputs, the closed form ties disturbance to the operator square roots of the effects: rank-one effects saturate the minimum possible disturbance 2/(d+1), and effects of rank at most k give bound at most (k+1)/(d+1).","The weighted state exclusion bound can certify nonzero disturbance using state preparations that are informationally incomplete for detector tomography, so disturbance detection requires less characterisation than full measurement tomography.","Because loss and depolarisation have equal informativeness but different statistical disturbance bounds, informativeness alone is insufficient to predict disturbance; the full POVM statistics carry disturbance information.","In the proposed randomness protocol, a certified lower bound on average fidelity directly yields an upper bound on an eavesdropper's guessing probability, linking disturbance certification to randomness generation."],"fun_headline_variants":["Outcome statistics alone set quantum disturbance floor","Measure disturbance without knowing the measurement","Quantum disturbance bound from measurement statistics","Statistical bound links measurement outcomes to disturbance"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the conditional probabilities p(a|i)=Tr(Aaρi) used in the weighted state exclusion optimisation are known exactly; with only finite measurement samples, the calculated value need not be a valid upper bound on FE(A).","fun_headline_variants_meta":{"raw":{"variants":["Outcome statistics alone set quantum disturbance floor","Measure disturbance without knowing the measurement","Quantum disturbance bound from measurement statistics","Statistical bound links measurement outcomes to disturbance"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000146,"raw_usage":{"total_tokens":1008,"prompt_tokens":725,"completion_tokens":283,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":469,"completion_tokens_details":{"reasoning_tokens":233}},"tokens_in":469,"tokens_out":283,"duration_ms":3322,"temperature":1.0,"reasoning_tokens":233,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T22:07:24.138625+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a fully characterised POVM and input ensemble, solve the SDP for FE(A) and then numerically search over Choi operators of compatible instruments aiming at a higher average fidelity; the theorem says no such instrument exists. For the experimental method, apply weighted state exclusion to a characterised measurement using finite counts and test whether the reported bound ever falls below the directly computed FE(A).","supporting_citations":[],"review_version":1}