{"id":"58c7baad-b1d2-4bf4-868a-608ab7a7673f","arxiv_id":"hep-ex/0404034","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A dip in the atmospheric muon-neutrino L/E distribution near 500 km/GeV gives the first direct evidence that neutrino survival follows the sinusoidal oscillation prediction.","lead":"Super-Kamiokande reports the first direct observation of the sinusoidal L/E dependence predicted by neutrino oscillation, using 1489 days of atmospheric neutrino data. A dip in the muon-neutrino L/E distribution near 500 km/GeV separates oscillation from decay and decoherence models.","discovery_kind":"new_method","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"The dip is visible in the data, but its significance and the quoted Δm² interval rest on a no-oscillation prediction and 24-parameter systematic model available only in an unpublished draft; the paper never quotes the no-oscillation χ², and its own 'factor of 3 to 4' sentence about the dip position,","rationale":"The reader's weakest_assumption — that the no-oscillation prediction from Ref. [16] ('draft in preparation') is accurate enough that the dip is not a modeling artifact — is the right place to look, and I agree it is load-bearing. I would sharpen it in two ways. First, the concern is not only that [16] is unpublished but that the paper never quotes the no-oscillation χ², so the reader cannot see how much of the dip the 24-parameter systematic model could absorb; the decay/decoherence comparison brackets the no-oscillation case only indirectly. Second, the paper's own sentence about the dip position being 'a factor of 3 to 4 away from that of the predicted event number minimum as seen in Fig. 3' is unexplained and, taken at face value, would undercut the identification of the dip with the first oscillation minimum — the very identification the central claim rests on. I would not treat this as a correctness failure: the dip is visible by eye in Fig. 4, the e-like control and direction-flip tests are real support for the flux/cross-section shape, and the extracted parameters agree with the published zenith-angle analysis [1]. The qualitative conclusion survives; what is not independently checkable is the precision of the Δm² interval and the statistical weight of the dip against smooth alternatives. That is a reproducibility and presentational gap, not a flaw in the physics argument, so the reader's CONDITIONAL verdict is appropriate and I would not move it. The concrete test I propose (freeze the nuisance parameters and quote all four χ² values) is the single check that would settle whether the systematic freedom is doing load-bearing work; the factor-3-to-4 sentence should be resolved in the same pass by overlaying the oscillated prediction on Fig. 3 and reporting the L/E of its event-number minimum.","tokens_in":9914,"tokens_out":7027,"duration_ms":119919,"concrete_test":"Re-run the fit with the 24 constrained nuisance parameters frozen at their central values (no profiling) and report χ² for no-oscillation, oscillation, decay, and decoherence. If the oscillation-vs-no-oscillation gap and the oscillation-vs-decay gap (currently 11.3, quoted as 3.4σ) shrink substantially when the systematics are frozen, the dip's significance is carried by the model's reshaping freedom rather than by the data; if they hold, the concern does not land.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the L/E dip is the first oscillation minimum and therefore direct evidence for the sinusoidal survival probability. Two conditions must hold: (i) the dip sits at the L/E where the survival probability is minimal, and (ii) the dip is significant against smooth alternatives given the model's reshaping freedom. On (i): the paper states 'the position of the dip is about a factor of 3 to 4 away from that of the predicted event number minimum as seen in Fig. 3.' For Δm² = 2.4×10⁻³ eV² the survival minimum is at L/E ≈ 515 km/GeV, which matches the observed dip at ~500 km/GeV. A factor 3–4 mismatch would put the predicted minimum near 130–170 km/GeV, which is not what oscillation predicts, and Fig. 3 shows only the non-oscillated MC, which has no minimum at all. The sentence is either garbled or points to a real mismatch; either way it is unexplained and bears directly on whether the dip is the first oscillation minimum. On (ii): the fit (Eqs. 2–4) profiles 24 Gaussian-constrained nuisance parameters at every grid point, and the structure of that model — flux, cross section, detector — is delegated entirely to Ref. [16], listed as 'draft in preparation.' The paper reports χ² = 37.9/40 for oscillation and 49.1, 52.4 for decay/decoherence, but never quotes χ² for no-oscillation, so a reader cannot bound how much of the dip the nuisance parameters could absorb or create. The e-like control and direction-flip tests constrain the flux/cross-section shape and are genuine support, but they do not substitute for the missing baseline χ².","agreement_with_reader":"partial"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The paper reports a search for the sinusoidal L/E dependence of muon-neutrino disappearance using 1489 live-days of Super-Kamiokande fully-contained mu-like and partially-contained atmospheric neutrino events (2726 events after the L/E resolution cut of 70%). A dip is observed in the ratio of data to non-oscillated Monte Carlo at L/E ≈ 500 km/GeV. A fit with 25 systematic parameters (Eqs. 2–4) gives a best fit of (sin²2θ, Δm²) = (1.00, 2.4×10⁻³ eV²) with χ² = 37.9/40 DOF, and a 90% C.L. allowed region of 1.9×10⁻³ < Δm² < 3.0×10⁻³ eV² and sin²2θ > 0.90. Neutrino decay and neutrino decoherence, which reproduce the zenith-angle deficit without a dip, are disfavored at 3.4σ and 3.8σ respectively, calibrated with 10,000 pseudo-experiments. The authors conclude that the L/E distribution provides the first direct evidence for the sinusoidal survival probability predicted by neutrino oscillation.","tokens_in":10270,"tokens_out":7333,"duration_ms":127793,"significance":"If it holds, this result converts the atmospheric muon-neutrino deficit from a zenith-angle anomaly into a measured oscillation pattern, and it is the first observation of the oscillatory L/E dependence itself rather than of its flux-averaged consequence. The paper deserves credit for several things that are easy to understate: a genuine null test in which the event direction vector is artificially reversed, a control sample of e-like events shown to be flat in L/E, cross-checks at 60%, 80% and 90% resolution cuts and at several bin sizes, a full 25-parameter systematic treatment with Gaussian pull terms rather than a single normalization uncertainty, and an explicit, MC-calibrated comparison against two physically motivated smooth alternatives. The quoted Δm² interval is a real constraint on the oscillation parameter, and the falsifiable prediction that the second oscillation maximum should be unobservable at this resolution is stated plainly.","major_comments":[{"comment":"The sentence 'We note that the position of the dip is about a factor of 3 to 4 away from that of the predicted event number minimum as seen in Fig. 3' cannot be reconciled with the figures as presented. Fig. 3 shows only the non-oscillated Monte Carlo, which falls monotonically in L/E and has no minimum at all, and for the quoted best fit Δm² = 2.4×10⁻³ eV² the first survival minimum lies at L/E ≈ 515 km/GeV, coincident with the observed dip at roughly 500 km/GeV. As written, the sentence implies that the dip is not at the first oscillation minimum, which would undercut the central claim of the paper. The authors should either correct the sentence or explain precisely what 'predicted event number minimum' refers to and why a factor of 3 to 4 displacement is expected.","section":"§3, paragraph following Fig. 3 (dip position)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The no-oscillation prediction and the full 25-parameter systematic model — the σ_j and f_ij of Eqs. (2)–(3) — are delegated entirely to Ref. [16], which is listed as 'draft in preparation.' Because the dip is a ratio to that prediction, a reader cannot judge how much of the structure the nuisance parameters could absorb or create. At minimum the Letter should include a table of the dominant systematic uncertainties with their fractional effect on the normalization and on the shape of the L/E distribution, and should quote the absolute energy-scale uncertainty separately, since Δm² scales inversely with the L/E scale and the quoted 1.9–3.0×10⁻³ eV² interval is directly sensitive to it.","section":"§4, Eqs. (2)–(4) and Ref. [16]","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper reports χ²_min = 37.9/40 for oscillation and 49.1 and 52.4 for decay and decoherence, but never quotes χ² for the no-oscillation hypothesis, nor for a smooth dip-free alternative fitted with the same 25 nuisance parameters. The quoted 3.4σ and 3.8σ are statements about two specific alternatives, not about the general smooth hypothesis that the dip is a modeling artifact. Since the central claim is that the dip is significant against smooth disappearance, the authors should quote the no-oscillation χ² with systematics minimized, and ideally the χ² for a smooth exponential in L/E, so that the reader can see the Δχ² that the dip actually buys.","section":"§4, significance discussion","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The Gaussian product in Eq. (2) runs over 24 terms while the sum in Eq. (3) runs over 25 parameters. The text explains that the absolute normalization is free, but the notation should state explicitly which index is unconstrained so that the counting is unambiguous.","section":"Eqs. (2)–(3)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The caption refers to 'three additional lines in (a)' but then describes only the solid and dashed lines. The third line should be identified, and its meaning relative to the 70% resolution contour stated.","section":"Fig. 2 caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The caption should state the number of events plotted and confirm that it equals the 2726 events of Table I after the L/E resolution cut, so that the reader can connect the two.","section":"Fig. 3 caption","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The phrases 'first direct evidence' and 'direct' carry weight. Since the dip position is obtained from a fit to the same L/E data, the authors should either justify the word 'direct' explicitly or soften the claim to something like 'the first observation of the oscillatory L/E dependence.'","section":"Abstract and final paragraph","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The 1489 live-day exposure should be accompanied by the calendar period over which it was accumulated, since the detector configuration and calibration changed over the run.","section":"§2, data set","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"For χ² = 37.9 with 40 DOF, the goodness-of-fit p-value should be quoted so that the reader can see that the oscillation hypothesis describes the data well, not merely better than the alternatives.","section":"§4, fit result","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The parenthetical percentages are the νµ + ν̄µ CC purity of each sample; this should be stated in the caption. Adding the data/MC ratio per row would make the deficit pattern easier to read.","section":"Table I","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The rise of the data/prediction ratio at large L/E is attributed to energy-dependent systematics. The authors should name the dominant term and state whether the fit's preferred value of that nuisance parameter lies within one standard deviation of its nominal value.","section":"§3, large-L/E rise","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision","confidential_remarks":"The physics appears sound and the analysis is careful; my recommendation reflects two verifiability gaps rather than any doubt about the result. First, the systematic model that shapes the entire comparison is available only in a manuscript listed as 'draft in preparation,' which is a poor fit for a Letter making a 'first direct evidence' claim. Second, the 'factor of 3 to 4' sentence about the dip position reads as an internal inconsistency and should be checked against the version the authors intend to publish. If both are resolved, I would expect to support publication."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"This is the Super-K paper that turned the atmospheric deficit into a shape measurement, and it holds up. What's new isn't the parameters — those were already pinned by the zenith-angle analyses — it's the event selection that keeps only events with good L/E resolution and then shows a dip in data/prediction near 500 km/GeV, right where the first survival minimum sits for Δm² ≈ 2.4×10⁻³ eV². That's the first direct look at the sinusoidal form rather than a two-bin up/down deficit, and the claim in the conclusion is earned at that level of precision. It is not oversold as a parameter-free prediction, because it isn't one, and the paper doesn't pretend otherwise — the fit is to the same L/E data, and the discrimination comes from shape against smooth alternatives, not from predicting the dip position blind. That's the honest framing and they use it. The model comparison is fair: decay and decoherence each get two free parameters, same as oscillation, and they convert Δχ² into a probability with 10,000 MC pseudo-experiments instead of quoting a naive Gaussian tail. The direction-flip test is a nice trick — flip the reconstructed direction and the dip should vanish, and it does. The e-like control sample staying flat is the right cross-check. Resolution cuts at 60/80/90% and alternate binning all reproduce the dip. Soft spots, in order of how much they matter, which is not much. The 25-parameter systematic model lives in Ref. [16], listed as 'draft in preparation.' The paper does break it down by category (7 flux, 3 detector, 2 reduction, 5 reconstruction, 8 interaction), so it isn't opaque, but the precision on the Δm² interval does depend on it. At submission this was the collaboration's own forthcoming paper; it has since appeared, so this is a citation-timing issue, not a hole. No-oscillation χ² is never quoted — 'very strongly disfavored' with no number. Minor, since the zenith analysis already killed no-oscillation, but it costs nothing to state it. The 'factor of 3 to 4' sentence about the dip position versus the predicted event-number minimum is genuinely confusing and should be rewritten. I'd push back on reading it as a real mismatch: 500 km/GeV is exactly where the first minimum belongs. It reads like a garbled reference to resolution smearing, not a discrepancy. Who should read it: anyone working on long-baseline or atmospheric oscillations, and anyone who wants a template for arguing shape against smooth alternatives. Send it to referees.","headline":"The L/E dip is real and the model comparison is done properly; the soft spots are presentational — an unpublished systematic reference and one garbled sentence — not load-bearing.","tokens_in":10921,"tokens_out":2398,"would_cite":true,"duration_ms":39186,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["14.60.Pq","96.40.Tv"],"model":"selfhosted/deepseek-v4.1-flash","headline":"Super-Kamiokande reports a dip in the muon-neutrino survival probability plotted against flight length over energy, the shape a sine-squared oscillation predicts.","keywords":["atmospheric neutrinos","neutrino oscillation","Super-Kamiokande","L/E distribution","muon neutrino disappearance","mass splitting measurement","water Cherenkov detector","neutrino decoherence"],"falsifier":"An independent atmospheric or long-baseline experiment with comparable $L/E$ resolution should see the same dip near 500 km/GeV and fit the same mass splitting; alternatively, a revised flux or cross-section model that reproduces the dip without oscillation would undercut the claim.","tokens_in":11241,"feed_emoji":"🌊","tokens_out":2282,"duration_ms":42272,"temperature":0.7,"pith_summary":"If neutrinos really oscillate, the probability that a muon neutrino survives should fall and rise as a sine-squared function of $L/E$, the distance travelled divided by the energy. That sinusoidal shape had never been seen directly: solar neutrinos lose coherence, and reactor and accelerator experiments lacked the statistics. This paper selects the 2,706 atmospheric neutrino events in Super-Kamiokande with the sharpest $L/E$ resolution and plots their rate against $L/E$. A dip appears near 500 km/GeV, at the place where the first oscillation maximum should sit, and the fit to that dip pins the oscillation parameters to $1.9-3.0$ times $10^{-3}$ eV$^2$ in mass splitting with near-maximal mixing. The point is not a new deficit but the shape of the deficit: a smooth disappearance, as neutrino decay or decoherence would produce, cannot make a dip.","feed_headline":"First dip spotted in the neutrino L/E curve","feed_subtitle":"A trough near 500 km/GeV in 1,489 days of Super-K data matches the sine-squared shape oscillation predicts.","key_machinery":"The load-bearing variable is $L/E$ itself: flight length inferred from the reconstructed neutrino direction, divided by the neutrino energy inferred from the total observed Cherenkov light (plus outer-detector deposit for partially contained events). Because the oscillation argument is $1.27 Δm^2 L/E$, a fixed detector resolution in $L/E$ translates into a fixed phase resolution, so events can be cut on their own $L/E$ uncertainty — the paper keeps events with $Δ(L/E) < 70%$ — rather than on energy or angle separately. That cut is what concentrates the sample where the first oscillation maximum is visible, and it is why the analysis can look for the shape of the survival probability rather than only its average.","core_discovery":"The collaboration claims the first direct evidence that the muon-neutrino survival probability follows the sinusoidal $L/E$ dependence of flavour oscillation. Using 1,489 live days of fully contained muon-like and partially contained atmospheric neutrino events, they bin the data in $L/E$ over 43 bins and fit a two-flavour $P(ν_μ → ν_μ) = 1 - $sin$^2 2θ$ sin$^2(1.27 Δm^2 L/E)$ with 25 systematic parameters floating within their estimated uncertainties. The fit gives a minimum $χ^2$ of 37.9 for 40 degrees of freedom at $(sin^2 2θ, Δm^2) = (1.00, 2.4 × 10^{-3}$ eV$^2)$, and the 90% confidence region is $1.9 × 10^{-3} < Δm^2 < 3.0 × 10^{-3}$ eV$^2$ with $sin^2 2θ > 0.90$. Against two alternatives that also reproduce the zenith-angle deficit but predict no dip — neutrino decay and neutrino decoherence — the oscillation fit is better by $Δχ^2$ of 11.3 and 14.5, which the authors translate into roughly 3.4 and 3.8 standard deviations, with a Monte Carlo study putting the decay-mimics-oscillation probability near 0.1%.","pith_inferences":["The authors note that the second and higher oscillation maxima should be washed out by detector resolution, so this is a one-dip measurement; a detector with better $L/E$ resolution would be able to see the next maximum and turn the shape test into a multi-node fit.","Because the dip position scales as $1/Δm^2$, the same plot made with a different $L/E$ binning or resolution cut is a consistency test the paper already performs; a future analysis could push this into a shape-only fit that does not rely on the overall normalisation at all.","The paper's control sample of electron-like events is flat in $L/E$, which is what one expects if the effect is specific to muon neutrinos; extending that control to a quantitative $ν_e$ oscillation fit would test whether the atmospheric anomaly is purely a two-channel story.","If the dip is real, the same $L/E$ structure should appear in any sufficiently large atmospheric data set taken with an independent detector, which makes the plot a portable check rather than a Super-K-specific result."],"forward_implications":["The mass splitting is pinned to a narrow window, $1.9 × 10^{-3} < Δm^2 < 3.0 × 10^{-3}$ eV$^2$ at 90% confidence, so the location of the first oscillation maximum is now predicted rather than merely bounded.","Near-maximal mixing is confirmed by the depth of the dip, with $sin^2 2θ > 0.90$ at 90% confidence.","Models that explain the atmospheric deficit by neutrinos decaying or losing coherence are disfavoured, because a smooth disappearance cannot produce a dip in $L/E$.","The same $L/E$ technique becomes a template for any detector with enough events and adequate $L/E$ resolution to resolve the oscillation shape rather than just the rate.","The 500 km/GeV dip position becomes a cross-check target for accelerator and atmospheric measurements made with different systematics."],"supporting_citations":[{"why":"Supplies the earlier Super-Kamiokande zenith-angle evidence for oscillation that this analysis is refining.","marker":"[1]"},{"why":"Supplies the atmospheric neutrino Monte Carlo prediction and the systematic error terms the fit relies on; the paper describes it as a draft in preparation.","marker":"[16]"},{"why":"Defines the neutrino-decay survival probability used as the main alternative hypothesis against which the dip is tested.","marker":"[18]"},{"why":"Provides the decay-model framework and parameters compared with the oscillation fit.","marker":"[19]"},{"why":"Defines the neutrino-decoherence survival probability used as the second alternative hypothesis.","marker":"[20]"},{"why":"Supplies the decoherence treatment and parameters against which the oscillation fit is compared.","marker":"[21]"},{"why":"Provides the two-dimensional confidence-interval method used to draw the allowed parameter contours.","marker":"[17]"},{"why":"Explains why solar neutrinos lose coherence and therefore cannot show a sinusoidal survival probability, motivating the atmospheric search.","marker":"[15]"}],"fun_headline_variants":["The neutrino L/E curve dips where oscillation says it should","Super-K sees the L/E dip that oscillation predicts","Oscillation beats decay and decoherence by 3.4σ and 3.8σ","L/E dip fits oscillation at only 3.4–3.8σ over alternatives","Muon-neutrino survival follows the predicted sine-squared L/E shape"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The no-oscillation prediction — built from the atmospheric neutrino flux, interaction cross sections and detector model described in a reference listed as a draft in preparation — is accurate enough that the dip near 500 km/GeV is a real feature of the data rather than a residual of the model.","fun_headline_variants_meta":{"raw":{"variants":["The neutrino L/E curve dips where oscillation says it should","Super-K sees the L/E dip that oscillation predicts","Oscillation beats decay and decoherence by 3.4σ and 3.8σ","L/E dip fits oscillation at only 3.4–3.8σ over alternatives","Muon-neutrino survival follows the predicted sine-squared L/E shape"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":1776,"prompt_tokens":922,"completion_tokens":854,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":767}},"tokens_in":922,"tokens_out":854,"duration_ms":13797,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":767,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T16:49:05.944954+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"An independent atmospheric or long-baseline experiment with comparable $L/E$ resolution should see the same dip near 500 km/GeV and fit the same mass splitting; alternatively, a revised flux or cross-section model that reproduces the dip without oscillation would undercut the claim.","supporting_citations":[{"cited_title":"Fukuda et al., Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier Super-Kamiokande zenith-angle evidence for oscillation that this analysis is refining."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the atmospheric neutrino Monte Carlo prediction and the systematic error terms the fit relies on; the paper describes it as a draft in preparation."},{"cited_title":"Phys. Rev. Lett. 82, 2640","cited_arxiv_id":null,"evidence_quote":"Defines the neutrino-decay survival probability used as the main alternative hypothesis against which the dip is tested."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the decay-model framework and parameters compared with the oscillation fit."},{"cited_title":"Atmospheric \\nu_\\mu Deficit from Decoherence","cited_arxiv_id":"hep-ph/9807511","evidence_quote":"Defines the neutrino-decoherence survival probability used as the second alternative hypothesis."},{"cited_title":"Phys. Rev. Lett. 85, 1166","cited_arxiv_id":null,"evidence_quote":"Supplies the decoherence treatment and parameters against which the oscillation fit is compared."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the two-dimensional confidence-interval method used to draw the allowed parameter contours."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Explains why solar neutrinos lose coherence and therefore cannot show a sinusoidal survival probability, motivating the atmospheric search."}],"review_version":1}