{"id":"516f4e1f-4702-4cfc-9a0e-ebc7d0f25707","arxiv_id":"2607.25699","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Interaction-free interferometry carries half the Fisher information of direct probing for transmissivity, equal information per absorbed photon, and only beats direct schemes when distinguishing an object from empty space, at a rate growing like the number of Zeno cycles.","lead":"Interaction-free measurement claims to sense fragile objects with photons that never touch them. This paper shows in estimation-theoretic terms that it buys nothing for measuring transmissivity, but offers a real, quantifiable advantage for simple presence detection.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Sec. V bound is stated for irreversible loss into untouched environment modes, but the abstract's 'arbitrary fixed optics' omits this qualifier, making the headline claim broader than the theorem as proven.","rationale":"I re-derived the core mathematical steps and found them internally consistent: the beam-splitter dilation T=cos²φ, the Lemma's induction (the B factors cancel because B commutes with J, so the cross term reduces to ⟨J_Kψ_{K-1}|∂φψ_{K-1}⟩ and vanishes by freshness), and the monotonicity step under partial trace. The numerical checks reported in the paper, such as R=40.03 at N=100 and the N n_abs values at N=10^8, match the closed-form asymptotics. The reader's weakest_assumption identifies the same substantive fragility: the proof depends on the object being memoryless, non-dispersive, and on the loss modes being untouched, so that each pass meets a fresh vacuum mode and the generator expectation vanishes. I agree this is the only soft point in the central argument, but the paper itself states it as a scope condition and explicitly says that beating the exchange rate requires breaking one of these assumptions. The real issue is presentational: the abstract's 'arbitrary fixed optics' is broader than the theorem's environment-untouched, irreversible-loss setting. The three mechanical errors listed by the reader (the T=cos²φ typo, the Fig. 1 caption reference, and the classical multi-pass Fisher formula) are genuine but do not touch Eq. (25) or Eq. (13). I therefore do not change the CONDITIONAL verdict; the paper should correct the mechanical errors and add the environment qualifier to the abstract, but the core claims are sound within their stated scope.","tokens_in":14692,"tokens_out":39901,"duration_ms":365082,"concrete_test":"Construct a minimal two-pass network: W0=I, |in⟩=|o⟩, B1(φ), W1 a tunable beam splitter/phase that mixes |o⟩ and the first loss mode |e1⟩, then B2(φ). Numerically compute the quantum Fisher information of the reduced photon state (trace over e1,e2) for several T=cos²φ in (0,1), normalized by n_abs=(1−T)(|v1|²+|v2|²), and compare with the bound 1/[T(1−T)²]. Repeat with an optimized K=3 network allowing recycling of earlier loss modes. If the ratio exceeds the bound anywhere, the environment-untouched condition is genuinely load-bearing and must appear in the abstract; if it never exceeds, the theorem may be robust to loss-mode recycling and only the wording needs tightening.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The theorem of Sec. V, Eq. (25), is proved for a single photon whose kth encounter with the object couples to a fresh vacuum mode e_k, and both the Lemma and the purification identity use the freshness condition ⟨e_k|ψ_{k-1}⟩=0 (Eqs. (22)-(23)). The intervening optics W_j are required not to act on those environment modes, and this is what makes n_abs=(1−T)Σ|v_k|² a genuine count of irreversibly absorbed photons. The abstract and the conclusion, however, advertise the bound for 'arbitrary fixed optics' without this qualifier. The gap is not purely rhetorical: if W_j may redirect an earlier loss mode e_1 back into the object mode, the photon that was 'absorbed' on pass 1 can be re-injected, so the loss is no longer irreversible and the per-absorbed-photon reading of the bound is not the same physical quantity. The paper does flag the assumption in its 'Three points about scope' and states that beating the exchange rate requires breaking it, so the body is honest; the concern is that the unqualified abstract and concluding sentences overstate the theorem's domain.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper analyses interaction-free measurement (IFM) as a parameter-estimation and hypothesis-discrimination problem. It shows that the Elitzur–Vaidman interferometer carries exactly half the Fisher information about the intensity transmissivity T carried by direct transmission probing, and the same Fisher information per absorbed photon. It then studies the N-cycle quantum Zeno interrogator: for discriminating a semitransparent object from empty space, the Chernoff information per absorbed photon grows as (8/pi^2)(1-√T)/(1+√T) N ln N, while discrimination between two partial transparencies does not grow; parasitic loss per cycle caps the conclusive-interrogation ratio at R_max ≈ 0.2625/epsilon at N_opt ≈ 1.5936/epsilon. The final section proves a general bound for a single photon meeting a memoryless, non-dispersive object any number of times through fixed optics that do not touch the environment modes: F_Q(T)/n_abs ≤ 1/[T(1-T)^2], recovering the Massar–Mitchison–Pironio bound for this class and identifying when it is tight.","tokens_in":14837,"tokens_out":55548,"duration_ms":459571,"significance":"If correct, the paper cleanly separates the two senses of 'supersensitivity' in the IFM literature and gives closed-form, experiment-designable statements: no estimation advantage per damage, a genuine N ln N discrimination advantage only against the 'absent' hypothesis, and a loss-per-cycle figure of merit. The central derivations (Eqs. (4)–(7), (9)–(13), (16)–(17), (24)–(25)) are analytic and benchmarked against exact numerics; the paper is unusually careful in stating assumptions, including the three conditions on which the Sec. V theorem rests, and in honestly reporting that it cannot propose a calibration-free test of Eq. (17). These strengths make the core contribution valuable despite the local errors noted below.","major_comments":[{"comment":"Eq. (11) as printed gives the steady-state rotated-output probability as P_V → T pi^2/[4N^2](1-√T)^2 (or T pi^2/[4N^2(1-√T)^2], depending on how the denominator is read), but an exact asymptotic evaluation of the transfer matrix A = diag(1,√T) R(theta) gives P_V → pi^2/[4N^2(1-√T)^2]. For example, at T=0.3, N=100, exact diagonalization gives P_V ≈ 1.11e-3, while the printed formula gives 1.5e-5 or 3.6e-4 depending on the reading. Consequently the additive constant in C = -ln P_V should be ln[4(1-√T)^2/pi^2], not ln[4(1-√T)^2/(T pi^2)]. The leading N ln N coefficient in Eq. (13) is unaffected, but the statement that the additive constant agrees to three decimals cannot be correct as written; please correct Eq. (11), the constant in C, and the numerical verification.","section":"Section III, Eq. (11)"},{"comment":"The advertised domain of the negative result is broader than the theorem proved. Eq. (25) is derived under the three assumptions stated in Sec. V, in particular that the intervening optics W_j do not act on the environment modes e_k, so that each pass meets a fresh vacuum mode and n_abs counts irreversible absorption. The abstract's 'arbitrary fixed optics' and the conclusion's 'every fixed multi-pass arrangement' omit this qualifier. If W_j can redirect an earlier loss mode back into the object mode, the quantity (1-T) sum |v_k|^2 is no longer the number of photons irreversibly absorbed, and the per-absorbed-photon reading of the bound changes. The body is honest ('Three points about scope'), so this is a presentation fix, but it should be made in the abstract and conclusion.","section":"Abstract and Sec. V"}],"minor_comments":[{"comment":"The formula for the classical Fisher information per unit dose contains an extra factor (1-T); it should read m^2 T^{m-2}/(1-T^m). With the printed formula, the m=1 value is 1/T rather than 1/[T(1-T)], and the stated '1% of the bound at m=1' and '65% at m=159' are off by a factor (1-T).","section":"Section V, multi-pass microscopy"},{"comment":"The phrase 'steady V amplitude √T θ/(1−√T)' is confusing: in the normalized steady state of the chain, the quantity √T θ/(1−√T) is the H amplitude, not the V amplitude. This wording likely contributed to the Eq. (11) error.","section":"Section III, before Eq. (11)"},{"comment":"The parasitic factor in the text is garbled ('√1−ϵ⊮'); it should be √(1−ε) as an amplitude factor (or 1−ε as an intensity factor), used consistently in the exact expressions.","section":"Section IV, Eq. (18)"},{"comment":"The comparison with 'the same incident flux spent on independent single-pass probes' should be clarified: a K-pass scheme involves one incident photon, while matching the summed exposure requires nbar_tot photons; the comparison is clean only in the per-absorbed-photon currency of Eq. (25).","section":"Abstract and conclusion"}],"recommendation":"major_revision","confidential_remarks":"The paper is worth publishing after correction. The Eq. (11) error is local but should be fixed before acceptance; the main theorem and the N ln N scaling are sound. The abstract overstatement should be corrected. The author's disclosure of AI assistance for derivations is explicit; I did not find evidence of undisclosed problems."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this paper settles a dispute that mostly shouldn't have existed. It computes, analytically and correctly, what interaction-free measurement actually buys: nothing for estimating transmissivity (factor of two worse than direct probing, identical per absorbed photon), and a real but carefully bounded advantage for discriminating an object from empty space. The quantitative results — exact Fisher information factor 2, per-absorbed-photon identity, closed-form Chernoff growth (8/π²)(1−√T)/(1+√T) N ln N, and the loss-limited optimum R_max = 0.2625/ε at N_opt = 1.5936/ε — are new and check out numerically. Section V's bound is a nicely compact proof of the Massar-Mitchison-Pironio limit for fixed sequential single-photon strategies, with useful tightness conditions.\n\nThe paper is also honest about its limits: it states the three assumptions (memoryless, non-dispersive, optics don't touch environment modes), says plainly that breaking any of them is the only escape from the exchange rate, and admits the lack of a calibration-free test of Eq. (17). That is a real limitation, not hidden.\n\nThe soft spots are mostly mechanical. Three errors need correcting: 'T = cos 2ϕ' should be 'T = cos² ϕ' in Sec. V and the Fig. 1(c) caption; the Fig. 1(c) caption references Eq. (9) when it should reference the general arrangement (Eq. (20) presumably); and the classical multi-pass Fisher formula has an extra (1−T) factor — without it, the numbers match the claim of 65% of the bound at m=159, with it they don't. Also, the abstract says the Sec. V bound holds for 'arbitrary fixed optics,' but the proof requires the optics to leave the environment modes untouched; the body flags this in 'Three points about scope,' so it's fixable, but the headline overstates the theorem's domain.\n\nMy verdict: this deserves a serious referee. The core mathematics is sound, the contribution is a useful clarification rather than a breakthrough, and the errors are local. I'd want the corrected version before citing, but it's well worth engaging with.","headline":"Quantifies what interaction-free measurement can and cannot do: valid core, fixable errors, and an abstract that overstates the theorem's domain.","tokens_in":15438,"tokens_out":6782,"would_cite":true,"duration_ms":57844,"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":"Interaction-free measurement, the paper shows, gains nothing for estimating an object's transparency but grows a detection advantage with Zeno cycles.","keywords":["interaction-free measurement","quantum Zeno effect","Fisher information","Chernoff information","transmissivity estimation","quantum channel discrimination","absorption measurement","counterfactual sensing"],"falsifier":"An experiment that measures the Fisher information per absorbed photon of a fixed multi-pass single-photon transmissivity estimator at $T\\approx0.5$ would refute the theorem if it exceeded $1/[T(1-T)^2]$. For the positive claim, a Zeno chain against empty space with $T_0=0.7$ should show a Chernoff exponent scaling as $N\\ln N$ that freezes to a constant once crosstalk or detection inefficiency is introduced; observing no freeze, or a linear instead of $N\\ln N$ growth, would falsify the discrimination analysis.","tokens_in":14395,"feed_emoji":"⚛️","tokens_out":9688,"duration_ms":71534,"temperature":0.7,"pith_summary":"The paper asks what interaction-free measurement (IFM) actually buys when the goal is to estimate an object's transparency rather than merely detect it. For estimating the transmissivity $T$, the Elitzur–Vaidman interferometer carries exactly half the Fisher information of direct transmission probing, and the same Fisher information per absorbed photon: the interferometer buys nothing. The advantage lives in discrimination: against the hypothesis of empty space, the Chernoff information per absorbed photon grows as $N\\ln N$ with the number of Zeno cycles, even for a weakly absorbing object, while between two partial transparencies it does not grow. A general theorem extends the negative result to every fixed multi-pass arrangement of a single photon, bounding the quantum Fisher information per absorbed photon by $1/[T(1-T)^2]$. Why it matters: the paper splits 'supersensitivity' into a false claim about estimation and a true claim about detection, giving a practical rule for when IFM is the right tool for fragile samples.","feed_headline":"Interaction-free sensing gains nothing for estimating transparency","feed_subtitle":"Advantage exists only for detecting presence against empty space, and is capped by per-cycle loss.","key_machinery":"The load-bearing machinery is a purification identity and the freshness property that powers it. Dilating the object to a beam splitter of angle $\\phi$ with $T=\\cos^2\\phi$, each encounter leaks into a fresh vacuum mode $|e_k\\rangle$ the photon has never met, so $\\langle e_k|\\psi_{k-1}\\rangle=0$ at every pass. This makes the generator expectation vanish and yields the exact identity $F_Q^{\\mathrm{pur}}(T)=\\bar n_{\\mathrm{tot}}/[T(1-T)]$, where $\\bar n_{\\mathrm{tot}}=\\sum_k |v_k|^2$ is the sum of squared incident amplitudes. Monotonicity of quantum Fisher information under the partial trace then gives the theorem $F_Q(T)/n_{\\mathrm{abs}} \\le 1/[T(1-T)^2]$ for any fixed multi-pass arrangement. For the Zeno chain the key object is the exact null of the empty apparatus: with no object, the $N$ rotations sum to $\\pi/2$ and the forbidden port has probability zero, which turns a count into proof and drives the $N\\ln N$ Chernoff growth.","core_discovery":"Assessed on its own terms, the paper proves that interaction-free measurement is two different tasks with two different fates. As a channel-estimation problem, the Elitzur–Vaidman interferometer has Fisher information $F_{\\mathrm{MZI}}(T)=1/[2T(1-T)]$, exactly half of direct transmission probing's $F_{\\mathrm{dir}}(T)=1/[T(1-T)]$, and the information per mean absorbed photon is identical in both, $1/[T(1-T)^2]$. The Zeno chain's advantage is not opacity but a null experiment: with empty space as the alternative, the forbidden port has probability exactly zero, so the Chernoff information per absorbed photon grows as $(8/\\pi^2)(1-\\sqrt T)/(1+\\sqrt T)\\,N\\ln N$ for every $T<1$, while for two partial transparencies the rate does not grow at all. Parasitic per-cycle loss $\\epsilon$ caps the gain, with the maximum conclusive interrogations per absorbed photon equal to $(0.2625/\\epsilon)(1-\\sqrt T)/(1+\\sqrt T)$ at $N_{\\mathrm{opt}}=1.5936/\\epsilon$ for an opaque object. The central negative theorem, Eq. (25), states that for a single photon meeting a memoryless, non-dispersive object any number of times through arbitrary fixed optics, the quantum Fisher information of the reduced photon state obeys $F_Q(T)/n_{\\mathrm{abs}} \\le 1/[T(1-T)^2]$, recovering and making tight the previously known bound for this class.","pith_inferences":["The theorem's freshness assumption suggests a concrete testable boundary: an absorber with a coherence time longer than the interrogation interval, or a dispersive phase accumulating over passes, should show a Fisher information per absorbed photon exceeding $1/[T(1-T)^2]$; observing that would define the regime where the bound fails.","The paper leaves open whether multi-photon sequential or adaptive strategies can exceed the exchange rate; the Sec. V argument does not cover them, and generic channel-use arguments do not imply the single-photon bound.","The absence of a calibration-free observable exhibiting the predicted maximum of Eq. (17) is stated as the main obstacle to experiment; designing such an observable would be a natural next step.","The null-experiment logic identified here suggests that other sensing protocols that engineer an exact forbidden outcome, rather than merely a suppressed one, may exhibit similarly unbounded discrimination rates per unit damage."],"forward_implications":["For estimating an unknown transmissivity, no fixed multi-pass arrangement of a single photon is more informative per absorbed photon than the same incident flux spent on independent single-pass probes; arrangements that genuinely revisit the object are strictly worse.","Interaction-free detection works for any partially transmitting object against empty space, not just opaque ones, because the advantage tracks the exact zero of the null hypothesis rather than the object's opacity.","Per-cycle parasitic loss $\\epsilon$, not the physics of the interrogation, sets the practical limit: improving $\\epsilon$ by a factor of ten increases the maximum conclusive interrogations per absorbed photon tenfold, and cycling past $N_{\\mathrm{opt}}=1.5936/\\epsilon$ is actively harmful.","The reported multi-pass microscopy improvements are gains toward the single-photon bound, not past it: classical light starts a factor $(1-T)^{-1}$ below the bound, and multi-passing climbs inside that gap without crossing it.","Discriminating between two partial transparencies gives no growing rate, which is why estimating a continuous $T$ is the hard case and why the negative bound binds there."],"supporting_citations":[{"why":"Introduces the interaction-free measurement and the bomb-testing scenario that the paper formalizes as a channel-estimation problem.","marker":"[1]"},{"why":"Supplies the high-efficiency Zeno interrogation chain and the crosstalk imperfection model used in the discrimination analysis.","marker":"[3]"},{"why":"Provides the broader minimal-absorption bound that the Sec. V theorem recovers for single-photon multi-pass schemes.","marker":"[12]"},{"why":"Proves the absorption-free discrimination dichotomy (transparent vs partial, and partial vs partial) that the paper extends with closed-form rates.","marker":"[13]"},{"why":"Reached by particle counting the two main conclusions about transparency estimation and zero-loss discrimination, which the paper makes quantitative.","marker":"[16]"},{"why":"Models a lossy Zeno interrogator; the paper extracts from it the optimal cycle number and maximum conclusive detection ratio.","marker":"[17]"},{"why":"Provides the purification route and the quantum-limited loss-sensing identity that Eq. (24) specializes to one object met K times.","marker":"[24]"},{"why":"Gives the underlying quantum-limited loss-sensing constant that single-photon Fock states saturate.","marker":"[25]"},{"why":"Reports the multi-pass microscopy variance reduction that the paper reinterprets as approaching, not exceeding, the single-photon bound.","marker":"[20]"},{"why":"Analyzes multi-pass phase interferometry, the contrast case showing why phase accumulates over passes while loss trials do not compound.","marker":"[22]"}],"fun_headline_variants":["Interaction-free sensing: no gain for transparency, only for presence","Counterfactual sensing wins only against empty space, capped by loss","Interaction-free measurement: nothing for estimation, something for detection","Zeno chain buys detection, not transparency; loss caps the gain"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The object must be memoryless and non-dispersive, with each pass meeting a fresh untouched vacuum mode so that no re-emission or accumulated phase couples the passes; if those conditions fail, the exact identity behind the bound no longer holds.","fun_headline_variants_meta":{"raw":{"variants":["Interaction-free sensing: no gain for transparency, only for presence","Counterfactual sensing wins only against empty space, capped by loss","Interaction-free measurement: nothing for estimation, something for detection","Zeno chain buys detection, not transparency; loss caps the gain"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000266,"raw_usage":{"total_tokens":1734,"prompt_tokens":1191,"completion_tokens":543,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":807,"completion_tokens_details":{"reasoning_tokens":471}},"tokens_in":807,"tokens_out":543,"duration_ms":4941,"temperature":1.0,"reasoning_tokens":471,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:29:09.548118+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An experiment that measures the Fisher information per absorbed photon of a fixed multi-pass single-photon transmissivity estimator at $T\\approx0.5$ would refute the theorem if it exceeded $1/[T(1-T)^2]$. For the positive claim, a Zeno chain against empty space with $T_0=0.7$ should show a Chernoff exponent scaling as $N\\ln N$ that freezes to a constant once crosstalk or detection inefficiency is introduced; observing no freeze, or a linear instead of $N\\ln N$ growth, would falsify the discrimination analysis.","supporting_citations":[{"cited_title":"High-efficiency quantum interrogation measurements via the quantum Zeno effect","cited_arxiv_id":"quant-ph/9909083","evidence_quote":"Supplies the high-efficiency Zeno interrogation chain and the crosstalk imperfection model used in the discrimination analysis."},{"cited_title":"Quantum Optical Metrology -- The Lowdown on High-N00N States","cited_arxiv_id":"0904.0163","evidence_quote":"Provides the broader minimal-absorption bound that the Sec. V theorem recovers for single-photon multi-pass schemes."},{"cited_title":"Versatile Super-Sensitive Metrology Using Induced Coherence","cited_arxiv_id":"1907.09004","evidence_quote":"Proves the absorption-free discrimination dichotomy (transparent vs partial, and partial vs partial) that the paper extends with closed-form rates."},{"cited_title":"Mini- mum number of photons needed to distinguish two transparencies,","cited_arxiv_id":null,"evidence_quote":"Reached by particle counting the two main conclusions about transparency estimation and zero-loss discrimination, which the paper makes quantitative."},{"cited_title":"Quantum Zeno tomography","cited_arxiv_id":"quant-ph/0104021","evidence_quote":"Models a lossy Zeno interrogator; the paper extracts from it the optimal cycle number and maximum conclusive detection ratio."},{"cited_title":"Dose-efficient quantum phase estimation in lossy optical interferometry,","cited_arxiv_id":null,"evidence_quote":"Provides the purification route and the quantum-limited loss-sensing identity that Eq. (24) specializes to one object met K times."},{"cited_title":"Optimization of quantum interferometric metrological sensors in the presence of photon loss","cited_arxiv_id":"0908.3008","evidence_quote":"Gives the underlying quantum-limited loss-sensing constant that single-photon Fock states saturate."},{"cited_title":"Interaction-free measurement study as a quantum channel discrimination problem","cited_arxiv_id":"1703.03976","evidence_quote":"Reports the multi-pass microscopy variance reduction that the paper reinterprets as approaching, not exceeding, the single-photon bound."},{"cited_title":"Multi-pass microscopy,","cited_arxiv_id":null,"evidence_quote":"Analyzes multi-pass phase interferometry, the contrast case showing why phase accumulates over passes while loss trials do not compound."}],"review_version":1}