{"id":"40e42205-4c29-4c45-b026-563fea7be94d","arxiv_id":"1908.04982","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"The claimed violation of complementarity rests on mixing visibility from scattered photons with distinguishability from reflected photons, so the quoted V^2+D^2 exceeds 1 without invalidating wave-particle duality.","lead":"This experiment reports single-photon interference visibility of 0.97 and a which-path distinguishability of 0.83 in a milk-coated prism interferometer, and claims V^2+D^2=1.63 violates wave-particle duality. The two numbers come from different photons: visibility is read from scattered light, while path information is read from reflected light.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed violation mixes disjoint ensembles: V is measured on the scattered-photon subensemble at APD3, while D is taken as the reflectivity R of the reflected subensemble; no single ensemble of photons yields V=0.97 and D=0.83 simultaneously.","rationale":"The paper's central claim is precisely the abstract's V²+D²≈1.63. The weakest link is not the quality of the interference data but the logical identification of the quantities entering the inequality. The reader's objection is correct: the conditions of the Englert inequality include that V and D characterize the same ensemble and the same which-path measurement. Here, the scattered photons that produce the fringes are exactly the ones for which the reflected path tags are absent; the reflected photons that give R have no interference. In fact D=R can be recovered as the optimal distinguishability of a three-outcome detector in which APD1/APD2 are conclusive and APD3 is inconclusive, so the identification itself is defensible under that model; the fatal step is using the conditional APD3 visibility rather than the visibility of the full detection record. The paper's Eqs. (1)-(6) derive a conditional scattered intensity and never compute the joint phase-dependent statistics of all three detectors. No machine-checked proof or independent parameter-free derivation supports the step; the data plots show APD1/APD2 traces flat under the same phase scans, which is direct evidence that the global fringe contrast is diluted. The recommended verdict is unchanged from the reader's rejection, because the foundational claim does not follow from the data even though the experimental technique may be of some interest.","tokens_in":9293,"tokens_out":8358,"duration_ms":85132,"concrete_test":"Reanalyze the wedge-plate scan in Fig. 4 using the full detection record: form N_total(φ)=N_APD1(φ)+N_APD2(φ)+N_APD3(φ)/A_rel, with A_rel the independently measured relative collection/detection efficiency of the APD3 channel (NA 0.65 objective, 150-µm slit, APD quantum efficiency). Compute V_total=(N_max−N_min)/(N_max+N_min) over the same phase range. Under the paper's model, V_total should be near S·V_APD3≈0.17 (or smaller once A_rel is included), not 0.97. If V_total²+0.83²≤1, the reported violation is an artifact of conditioning on APD3.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 stipulates 'the path distinguishability is taken simply as D≈R' and Section 4 uses R=83% from APD1/APD2 reflected intensities with V=97% from APD3 scattered photons to obtain V²+D²≈1.63. The load-bearing flaw is the averaging over different subensembles. In the Englert inequality, V and D must refer to the same ensemble of photons under the same measurement record. APD1/APD2 clicks are essentially perfect which-arm tags, and Fig. 4(a,c) show those channels are phase-flat; the high-visibility fringes are seen only in APD3, i.e., in the scattered subensemble that has no which-arm tag from the reflected detectors. R=83% is a partition probability (reflected vs scattered), not a per-outcome distinguishability; it can be made into a global D only by averaging over the conclusive APD1/APD2 outcomes and the inconclusive APD3 outcome. But then the companion quantity must be the global fringe visibility over all three outputs, not the conditional APD3 visibility. Since the reflected output is phase-flat, the global visibility is diluted by the reflected fraction R; in an ideal equal-efficiency model it would be near S·0.97≈0.17, and with the actual small APD3 collection efficiency it is smaller. This gives V²+D²≈0.72≤1. Equations (1)-(6) compute only the conditional scattered intensity, so they provide no support for combining V_APD3 with D=R. The violation is therefore an artifact of conditioning.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a single-photon experiment in a modified Mach-Zehnder interferometer in which a prism surface coated with a weakly scattering milk film acts as an 'interference screen.' The authors measure an interference visibility of V=0.97 (longitudinal) or 0.84 (lateral) on the weakly scattered channel (APD3) and infer a path distinguishability D=0.83 from the measured reflectivity R of the prism surface. They combine these into V²+D²≈1.63 and claim this violates the orthodox wave-particle duality bound V²+D²≤1. The theoretical analysis in Section 3 derives the scattered-light interference pattern from the evanescent field, and Section 4 presents the phase-dependent and position-dependent APD3 count rates together with flat response of the reflected channels APD1 and APD2.","tokens_in":9606,"tokens_out":2236,"duration_ms":23856,"significance":"If the claim were correct, it would constitute a major empirical challenge to the Englert–Jaeger–Shimony–Vaidman complementarity bound and to standard quantum theory. The experimental implementation of a weak-scattering interferometer using single photons from a quantum dot source is in principle interesting, and the paper includes useful single-photon characterization (g(2) anti-bunching, dark-count subtraction, bright-state selection). However, the central quantitative claim is not supported because the quantities V and D are not defined for the same ensemble of photons, and the identification of path distinguishability with the reflectivity R is not grounded in a quantum measurement model. The reported violation is therefore an artifact of the analysis rather than a property of the measured photons.","major_comments":[{"comment":"The definition 'the path distinguishability is taken simply as D≈R' is not a legitimate operational definition of which-path distinguishability. In the Englert inequality, D is the maximum probability with which an observer can correctly guess which path a photon took based on the outcome of a path-detection measurement performed on that same photon. Here R is the average probability that a photon is reflected rather than scattered; it is a partition probability over the two subensembles, not a per-outcome distinguishability for the scattered photons whose interference is being observed. A photon that reaches APD3 has not been reflected and carries no which-arm information from the reflected detectors; a photon that reaches APD1 or APD2 is essentially perfectly tagged but does not contribute to the APD3 interference pattern.","section":"Section 3"},{"comment":"The values V=0.97 and D=0.83 are not measured on the same subensemble. The visibility is extracted from the scattered channel APD3, while D is inferred from the time-averaged reflectivity calibration of Figure 3(a) using the reflected signals that appear phase-flat in Figures 4(a) and 4(c). For a valid complementarity test, V and D must correspond to the same ensemble of photons under the same measurement record. Had the authors evaluated a global visibility over all three outputs, the phase-flat reflected channels would dilute the contrast: in an ideal equal-efficiency model with reflection probability R=0.83, the global visibility would be at most S·V_APD3 ≈ 0.17·0.97, giving V²+D² ≈ 0.72 ≤ 1. The conditional APD3 visibility cannot be paired with a path-distinguishability derived from the complementary reflected subensemble.","section":"Section 4 / Figure 4"},{"comment":"The theoretical derivation supports only the conditional intensity of the scattered subensemble. Equation (6) gives the intensity distribution at APD3 after scattering, proportional to S·A, and its contrast is unity only within that scattered subensemble. The calculation does not address the full positive-operator-valued measure (POVM) that includes the reflected outputs, and it provides no justification for combining the conditional APD3 visibility with a global reflectivity-based D. The complementarity inequality is a statement about a single interferometer output distribution, not about separately conditioned subsets; the paper does not supply the full multi-channel quantum description needed for such a claim.","section":"Equations (1)–(6)"},{"comment":"The claim that the experiment 'agrees well with theoretical prediction as made in Eq. 6 and in Ref. [25]' is weakened by the fact that Ref. [25] is by the same research group and was developed for this specific design. More importantly, the agreement is with Eq. (6), which only predicts the scattered-subensemble fringe contrast, not with a derivation of V²+D²>1. No independent quantum-mechanical calculation of the combined inequality is provided; the result V²+D²≈1.63 is forced by the identification D≡R rather than by a measurement of the complementarity quantity.","section":"Section 4, 'The above experimental data clearly indicates'"}],"minor_comments":[{"comment":"The caption of Figure 3(a) repeats 'reflection from a silver mirror' for both traces; one of the two should refer to the milk-coated prism surface.","section":"Figure 3"},{"comment":"The phrase 'convincingly show the possibility of breaking the limit' overstates the evidence, given the conditional-visibility issue described above; a more cautious wording would be appropriate.","section":"Abstract / Introduction"},{"comment":"There are several typographical errors, e.g., 're fection' in the text before Figure 3 and inconsistent use of 'positon' in Section 4; a careful proofread is needed.","section":"Section 2"},{"comment":"The visibility of the lateral interference pattern is extracted from a Gaussian-overlap fit that includes additional free parameters (beam waist, overlap, inclination); the uncertainty on the reported V=0.84 is not quoted, and the fit model is not described in a reproducibility level of detail.","section":"Figure 4"},{"comment":"The reflectivity calibration accounts for the silver mirror reflectivity and prism surface transmission, but the paper does not state the uncertainty on the 83.3% value or propagate it into the final V²+D² estimate; a quantitative uncertainty budget is needed.","section":"Section 4"}],"recommendation":"reject","confidential_remarks":"The central claim depends on a circular definition of D as the reflectivity R, and the measured V and D are taken from different photon subensembles. These are not presentation issues that can be fixed in revision. I also note that the theoretical prediction cited as independent support (Ref. [25]) is authored by a co-author of this manuscript, so the 'agreement' is partly self-confirmatory. Given the paper's scope as an experimental test of complementarity, the core result is unsound and rejection is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe headline: the paper claims a single-photon experiment with a milk-coated prism in a Mach-Zehnder interferometer that gives V=0.97 and D=0.83, so V^2+D^2≈1.63, allegedly violating the Englert bound. That claim is not supported. The result is an artifact of mixing two different subensembles: V is measured on photons that scatter into APD3, while D is just the reflectivity R of the prism, inferred from APD1/APD2 counts. No single photon or set of photons gives both values. In the Englert inequality, V and D must refer to the same measurement record. If you use the global fringe visibility across all three detectors, the scattered-only visibility is diluted by the large reflected fraction and the inequality is restored. The paper itself notes S+R≈1, which makes the inconsistency internal to its own model.\n\nWhat is genuinely new: the weakly-scattering milk-coated prism as an interference screen is a clever trick to get some scattered light out of a total-internal-reflection surface, and the single-photon source work is solid. The interference fringes in Fig. 4 are real and the visibility fits are fine. The experiment is probably a reasonable demonstration of weak scattering from an interface, and the technical detail of collecting scattered photons with a high-NA objective is worth something.\n\nThe soft spots are not minor. Defining D≡R is purely ad hoc; reflectivity is a partition probability, not a which-path distinguishability per outcome. There are no error bars on V or D, and the abstract overstates the conclusion. The prior theoretical proposal, Ref [25], is by the same corresponding author, so the citation pattern reinforces the echo. The data themselves are not the problem; the interpretation is.\n\nWho this is for: someone working on weak measurements or interferometry might get a useful experimental idea from the setup, but the foundational claim should be ignored.\n\nI would not desk-reject it out of hand—an expert referee could quickly identify the ensemble error and the paper could be revised into a modest technical note. But as it stands, it fails as a wave-particle duality test. If it comes to you, send it to a quantum foundations referee; it deserves serious review even though the conclusion will not survive.","headline":"Claimed violation of wave-particle duality is an artifact of comparing visibility on scattered photons with reflectivity inferred from reflected photons; the experiment is real but the conclusion is not.","tokens_in":10172,"tokens_out":3667,"would_cite":false,"duration_ms":35716,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A modified Mach-Zehnder interferometer with a milk-coated, weakly scattering prism reports single-photon interference visibility of 0.97 and path distinguishability of 0.83, giving V²+D²≈1.63, beyond the orthodox wave–particle duality…","keywords":["wave-particle duality","single photon","Mach-Zehnder interferometer","weak measurement","interference visibility","path distinguishability","quantum dot single-photon source","complementarity"],"falsifier":"A coincidence-resolved version of the experiment would settle the claim: for each detected photon, record which reflected output (APD1 or APD2) it comes from and whether it also contributes to the scattered fringe signal at APD3. If D and V are computed from the same photon stream, the inequality V²+D²≤1 should be restored, whereas the paper's claim predicts it stays near 1.6.","tokens_in":9023,"feed_emoji":"","tokens_out":5366,"duration_ms":53052,"temperature":0.7,"pith_summary":"This paper claims that a single photon can simultaneously display strong wave-like interference and strong particle-like path information in one interferometer, a combination the usual wave–particle duality rule forbids. In the experiment, a modified Mach-Zehnder interferometer replaces the second beam splitter with a prism surface coated with a thin milk film. The film weakly scatters a small fraction of the two beams into a detector that records an interference pattern with visibility as high as 0.97, while the reflected beams reach two path detectors with estimated distinguishability 0.83. The paper reports V²+D²≈1.63 for longitudinal fringes, far above the orthodox limit V²+D²≤1. If the two quantities describe the same photon ensemble, this would mean complementarity is not an absolute obstacle in weak-measurement interferometry.","feed_headline":"Single photons beat wave-particle duality limit in a prism test","feed_subtitle":"Weak scattering gives 0.97 visibility and 0.83 path distinguishability, pushing V²+D² to 1.63.","key_machinery":"The key element is the milk-coated prism surface, which acts as a weakly scattering interference screen. Two single-photon beams arrive at the prism hypotenuse with a small angle difference, interfere there, and mostly reflect into two path detectors, while a small scattered fraction is collected by a microscope objective and sent to a movable slit detector. The theoretical description uses the scattered-field intensity expression, proportional to 8SAP₀[1+cos(...)], to predict the interference pattern, and equates the path distinguishability with the measured reflectivity R≈0.83. This combination of high reflectivity and weak scattering is what lets the same surface provide both path information and interference fringes.","core_discovery":"The central claim is that a weakly scattering total-internal-reflection prism surface can serve as an interference screen inside a Mach-Zehnder interferometer, allowing a single photon to show both wave and particle properties at once. The measured longitudinal fringe visibility is V=0.97, the reflectivity-based path distinguishability is D=0.83, and their squares sum to 1.63, which exceeds the orthodox bound of V²+D²≤1. The lateral interference pattern gives V=0.84 and V²+D²≈1.39. The authors state that these observations are consistent with standard quantum mechanical calculations for this weak-measurement setup, yet they go beyond the familiar principle of wave–particle duality as usually formulated.","pith_inferences":["A stricter test would measure D with a genuine which-path observable, such as a weak path marker in each arm, applied to the same photons whose scattering contributes to V; until then, equating D with the reflectivity R remains an assumption.","The scattered subensemble used for V may be a biased sample: photons that scatter are not necessarily representative of the reflected ensemble used for D, so comparing V from APD3 with D from APD1/APD2 might involve different populations.","If the result survives a same-ensemble test, it would not invalidate quantum mechanics but would show that the standard V²+D²≤1 inequality encodes a tradeoff between strong, mutually exclusive measurements rather than a fundamental limit for weak measurements."],"forward_implications":["Single-photon interference and which-path information can be extracted from the same interferometer without inserting or removing the second beam splitter.","Values of V²+D² greater than 1 become experimentally accessible, so the orthodox limit is not a universal constraint on every interferometric arrangement.","The measured longitudinal visibility of 0.97 and reflectivity-based distinguishability of 0.83 imply V²+D²≈1.63, while the lateral fringes still give 1.39.","Improving the single-photon source, the scattering film, and the collection efficiency should push the combination toward the predicted V=1 and D→1 regime, i.e. V²+D²→2."],"supporting_citations":[{"why":"Supplies the orthodox inequality V²+D²≤1 that the experiment claims to exceed.","marker":"[22]"},{"why":"Provides the theoretical weak-measurement Mach-Zehnder proposal and the prediction that weak scattering allows V=1 and D→1 simultaneously.","marker":"[25]"},{"why":"Establishes the single-photon character of quantum-dot emission used as the source.","marker":"[27]"},{"why":"Supplies the single-mode-fiber coupling method that gives the roughly 100 kcps single-photon beam.","marker":"[29]"},{"why":"Characterizes the quantum-dot blinking and charge states used to select bright-state data for the reflectivity and interference measurements.","marker":"[31]"}],"fun_headline_variants":["V²=0.97, D²=0.83: photon breaks duality limit","Single photon outshines wave-particle rule in prism test","Weak-scattering prism reveals photon's dual nature at once","Photons exceed V²+D²≤1: wave and path info together","Simultaneous observation of photon path and interference achieved"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result depends on identifying path distinguishability D with the measured reflectivity R=0.83 of the milk-coated prism and on evaluating V from a scattered subensemble that may differ from the reflected subensemble used for D; if D and V do not describe the same photons, the claimed violation does not follow.","fun_headline_variants_meta":{"raw":{"variants":["V²=0.97, D²=0.83: photon breaks duality limit","Single photon outshines wave-particle rule in prism test","Weak-scattering prism reveals photon's dual nature at once","Photons exceed V²+D²≤1: wave and path info together","Simultaneous observation of photon path and interference achieved"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001001,"raw_usage":{"total_tokens":4200,"prompt_tokens":876,"completion_tokens":3324,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":492,"completion_tokens_details":{"reasoning_tokens":3233}},"tokens_in":492,"tokens_out":3324,"duration_ms":24338,"temperature":1.0,"reasoning_tokens":3233,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:26:39.922537+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A coincidence-resolved version of the experiment would settle the claim: for each detected photon, record which reflected output (APD1 or APD2) it comes from and whether it also contributes to the scattered fringe signal at APD3. If D and V are computed from the same photon stream, the inequality V²+D²≤1 should be restored, whereas the paper's claim predicts it stays near 1.6.","supporting_citations":[],"review_version":1}