{"id":"a31f5b4f-f4d3-448a-949b-793359718d86","arxiv_id":"1908.09719","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A displaced-vertex search in tau decays at BaBar, Belle, and Belle II could set limits on tau-mixing sterile neutrinos below the current DELPHI bound.","lead":"This paper proposes a way to search for a hypothetical heavy neutrino that mostly mixes with the tau neutrino at B-factory particle colliders, using the long lifetime of such a particle to find its decay away from the collision point. A smart generalist would read it because it shows how existing and near-future experiments could probe a mixing parameter that is currently almost unconstrained.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Projected limits rely on generator-level vertex information; realistic vertex resolution could degrade the claimed background rejection and shift the Belle II sensitivity curves.","rationale":"The reader correctly identified the hand-assigned efficiencies and generator-level background estimates as a weak point, and the verdict CONDITIONAL is appropriate. My stress-test sharpens this into a more specific, load-bearing issue: the paper's background rejection and mass reconstruction rely on the direction of the N extracted from the vector between the X1 production point and the X2 decay vertex, but for one-prong tau decays the X1 production vertex is not experimentally observable. Approximating it with the beam spot introduces an unquantified smearing that is not included in the simulated distributions. This affects the central quantitative claim more directly than an individual efficiency number, because it could degrade the background suppression on which the 'background-free' Belle II conclusion rests. I do not see an internal inconsistency in the constraint counting or the physics; the issue is the gap between generator-level truth and realistic detector performance. The qualitative conclusion that B-factory samples can surpass the DELPHI limit is plausible because the event samples are enormous and the displaced-vertex signature is powerful; even order-of-magnitude errors in the signal efficiency would not erase the projected improvement over DELPHI. However, if the vertex-direction smearing substantially broadens the signal peaks, the background rejection could fail, and the quantitative limits could move by a large factor. A full detector simulation or an actual analysis is needed, so the CONDITIONAL verdict should stand.","tokens_in":9576,"tokens_out":12653,"duration_ms":141635,"concrete_test":"Generate signal events (mN = 1 GeV, |V_tauN|^2 = 1e-5) and tau -> pi K_L nu, K_L -> pi mu nu background events with EvtGen. For each event, replace the true X1 production point with (i) the beam spot and (ii) the point of closest approach of the X1 track to the beam axis, then recompute the N direction unit vector, solve the 12-constraint system, and apply the same m1/m2 and E1/E2 cuts used in the paper. If the signal retention drops below 70% or the background retention rises above 6% (corresponding to more than about 50 background events in Belle II), the Fig. 3 Belle II curves are not reliable without a full vertex-resolution study.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the 95% CL limit curves in Fig. 3, which depend on the statement that bin-by-bin cuts on m1/m2 and E1/E2 retain 90% of signal and reject over 97% of background, leaving only 6 KL background events in the full Belle II sample. This background rejection is derived from EvtGen-level distributions (Fig. 2) with no detector resolution smearing; the paper explicitly says 'additional smearing is expected due to detector resolution, not included in this simulation.' The kinematic constraints themselves use the direction of p_N, obtained from the vector between the X1 production point and the X2 decay vertex. For all signal modes examined (X1 = pi, pi pi0, and the recommended lepton modes), the signal tau decays are one-prong, so the X1 production point (the tau decay vertex) is not directly measurable from a single charged track. If that point is approximated by the beam spot, the tau flight length (typically several hundred microns, up to ~1 mm in the exponential tail) introduces an angular uncertainty of up to ~1e-2 rad for an N decay at r = 10 cm. This unquantified smearing broadens the m1/m2 and E1/E2 peaks, so the 97% background rejection may degrade. Combined with the acknowledged factor-of-2 uncertainty in the pion mis-ID rate, the remaining background could rise from 6 to O(100) events, shifting the limits by a significant factor. The core idea remains plausible, but the quantitative projections are not yet robust without a detector-level study.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a search for a sterile neutrino N with mass below the tau mass that mixes predominantly with the tau neutrino, using tau-pair events at B-factories (BaBar, Belle, Belle II). The proposed signature is tau -> X1 N followed by N -> nu_tau l+l- with a displaced vertex, exploiting the long lifetime of N. The authors use EvtGen-generated events to estimate acceptance and kinematic variables, and they estimate detector-level efficiencies by hand. They find that a bin-by-bin cut on the two mass solutions (m1, m2) and two CM-energy solutions (E1, E2) retains 90% of signal and rejects over 97% of the KL background, leading to a nearly background-free Belle II analysis. Expected 95% CL limits on |V_tauN|^2 are shown in Fig. 3, surpassing the DELPHI bound by up to several orders of magnitude at Belle II.","tokens_in":9901,"tokens_out":5687,"duration_ms":61648,"significance":"The method addresses a genuine experimental gap: for mN < m_tau, the best existing limit on |V_tauN|^2 comes from the lepton-flavor-agnostic DELPHI search, and a dedicated tau-decay search at B-factories could improve it substantially. The paper is conceptually sound and transparent about its simplifications; it clearly labels the analysis as an estimate and identifies the need for full simulation of leptonic tau modes. If the projected sensitivities are validated by a detector-level study, the paper would provide a strong motivation for a concrete experimental search. However, the central quantitative claim depends on several hand-assigned efficiencies and generator-level background rejection rates, so the numerical curves in Fig. 3 are not yet robust.","major_comments":[{"comment":"The kinematic constraints use the unit vector from the X1 production point to the X2 decay vertex as the direction of p_N. For the X1 = pi mode, the tau decay vertex is not directly reconstructable from the single charged pion track; if the beam spot is used as an approximation, the tau flight length (hundreds of microns to ~1 mm) introduces an angular uncertainty on p_N of order 1e-2 rad for N decays at r ~ 10 cm. This smearing is not included in the EvtGen-level distributions of Fig. 2, as the paper itself notes that detector resolution is not simulated. The claimed 90% signal retention and 97% background rejection, and hence the estimated 6 background events in the full Belle II sample, therefore rest on an unvalidated idealization. Please provide a quantitative estimate of this smearing or a detector-level simulation, and show how the retained signal and rejected background change.","section":"Further background suppression paragraph, p. 3-4"},{"comment":"The background yield in the full Belle II sample is estimated as 850 events before the mass-energy cuts, based on a pion mis-identification efficiency of 0.5% that the authors acknowledge can vary by a factor of 2, and on a 1.4% acceptance obtained from EvtGen. The post-cut background of 6 events follows from applying the 97% rejection derived from generator-level distributions. If the mis-ID rate or the acceptance is a factor of 2 higher, the pre-cut background doubles; if the rejection is 95% instead of 97%, the post-cut background becomes roughly 40 events at the nominal pre-cut level. These variations would shift the 95% CL limits in Fig. 3 by a nontrivial factor and could invalidate the 'background-free' statement for Belle II. Please quantify the sensitivity under these plausible variations and show the resulting uncertainty band in Fig. 3.","section":"KL background estimate, p. 4"},{"comment":"The text states that a bin-by-bin cut 'retains 90% of the signal events' and then, after concluding that a sophisticated analysis 'can come close to being background-free,' the authors 'assume an additional efficiency loss of 75% to account for these cuts.' It is unclear whether this means a retained fraction of 75% (0.9 x 0.75 = 0.675) or a loss of 75% (0.9 x 0.25 = 0.225). The wording 'efficiency loss' suggests the latter, but that is hard to reconcile with the 90% retention claim. The effective signal efficiency that goes into Eq. (3) for the limits in Fig. 3 should be stated explicitly, and the relation between the 90% retention, the 75% factor, and the previously estimated epsilon = 10% should be clarified.","section":"Efficiency accounting after cuts, p. 4"}],"minor_comments":[{"comment":"The phrase 'by initial-state radiaion (ISR)' contains a typo; it should read 'initial-state radiation.'","section":"Paragraph after Fig. 2"},{"comment":"The phrase 'a ντ-mixing' is grammatically awkward; consider 'mixing with the τ neutrino' for clarity.","section":"Abstract"},{"comment":"The 'Belle II(+other modes)' curve includes X1 = l nu and X2 = pi+ pi- modes assuming the same signal efficiency as X1 = pi, while the text states that the leptonic X1 mode requires full detector simulation and is not studied in detail. Please label this curve as an idealized extrapolation in the caption and in the summary, so that it is not read as a validated sensitivity projection.","section":"Fig. 3 caption and final sensitivity estimate"},{"comment":"The acceptance a is defined as essentially the probability for N to decay inside the detector, but the numerical acceptance values used for the Fig. 3 curves are not tabulated. A table listing a and epsilon for each mass point and experiment would improve reproducibility.","section":"Section 'Our aim here is to estimate...', Eq. (3)"}],"recommendation":"major_revision","confidential_remarks":"This is a projection paper whose central numeric claim depends on unvalidated detector-level assumptions. The authors are candid about this, and the core idea is plausible and well-motivated. A major revision that adds a treatment of vertex smearing and a sensitivity-band estimate would convert this into a solid proposal paper. I would not reject it on the current evidence, but the 'background-free' conclusion for Belle II must be substantiated before the limits in Fig. 3 can be taken at face value."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proposes a concrete search strategy for a tau-mixing sterile neutrino N with mN < m_tau at B-factories, using tau -> X1 N followed by N -> nu_tau l+l- with a displaced vertex. What's new is the combination: earlier tau-decay searches relied on lepton-number/flavor/CP violation, and the two existing B-factory displaced-vertex searches didn't use this kinematic reconstruction. The kinematic trick—12 unknowns vs 12 constraints, solving for the N mass with a two-fold ambiguity—is clever and clearly described.\n\nCredit where it's due: the authors are candid about their simplifications. They use EvtGen for acceptance, consider the dominant KL background explicitly, estimate ISR effects with KKMC, and quote efficiencies with stated assumptions. The sensitivity curves in Fig. 3 follow directly from stated inputs. The qualitative conclusion—that BaBar, Belle, and Belle II can surpass DELPHI bounds on |VtauN|^2 for mN around 1 GeV—is plausible and worth taking seriously.\n\nThe soft spots are real but not fatal. The projections rest on hand-assigned efficiencies and generator-level distributions, with no detector simulation and no systematic uncertainties. The stress-test note raises a legitimate issue: the N direction is taken from the vector between the X1 production point and the X2 decay vertex, but for one-prong tau decays the production point isn't directly measurable. Using the beam spot introduces an angular uncertainty from the tau flight length, which will broaden the m1/m2 and E1/E2 peaks and may degrade the claimed 97% background rejection. The paper acknowledges detector-resolution smearing is not included, but doesn't quantify the effect. Combined with the factor-of-2 uncertainty in pion mis-ID, the Belle II background could rise from 6 to O(100) events, shifting the limits by a significant factor. That said, the method's core idea survives; a full detector-level study or an actual search is needed to firm up the numbers.\n\nThe math and citation pattern look solid. The kinematic constraints are derived from conservation laws and mass shell conditions, and the references cover the relevant existing searches and theoretical frameworks. The self-citations to Refs. [63] and [18] are legitimate published calculations, not hidden dependencies.\n\nThis is a useful phenomenological proposal, clearly written and honest about its limitations. It deserves a serious referee. Anyone working on sterile neutrino searches or long-lived particle phenomenology should read it; I'd cite it as the relevant proposal for tau-mixing sterile neutrinos.","headline":"A clever, honest proposal for a tau-mixing sterile neutrino search at B-factories; the sensitivity projections are rough but the method is novel and worth refereeing.","tokens_in":10449,"tokens_out":1997,"would_cite":true,"duration_ms":21969,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["14.60.Pq","12.60.Jv","14.80.Cp"],"model":"deepseek-v4-flash","headline":"A displaced-vertex search in B-factory tau samples can probe tau-mixed sterile neutrinos down to |VτN|² ≈ 10⁻⁵.","keywords":["neutrino masses and mixing","tau decays","sterile neutrino","long-lived particle","displaced vertex","B-factory","Belle II","tau neutrino mixing"],"falsifier":"A full detector simulation of $e^+e^- \\to \\tau^+\\tau^-$ events containing $\\tau \\to \\pi K_L \\nu$, $K_L \\to \\pi \\mu \\nu$ in Belle II, with the proposed four-track, displaced-dilepton-vertex, and $m_1/m_2$/$E_1/E_2$ selection applied, would settle the projection: if the background yield exceeds about 17 events per ab$^{-1}$ or the displaced-track signal efficiency falls below roughly 10%, the claimed background-free sensitivity and the $\\sim 10^{-5}$ limit would not survive.","tokens_in":9388,"feed_emoji":"⚛️","tokens_out":15030,"duration_ms":133379,"temperature":0.7,"pith_summary":"The paper proposes a new search for a sterile neutrino $N$ with mass below the tau mass, $m_N < m_\\tau$, that mixes predominantly with the tau neutrino. The method uses the large $e^+e^- \\to \\tau^+\\tau^-$ samples collected at B-factory experiments (electron-positron colliders that copiously produce tau pairs): $N$ is produced in a tau decay such as $\\tau \\to \\pi N$, travels a macroscopic distance because its lifetime scales roughly as $m_N^{-5}$, and then decays to a neutrino plus an $e^+e^-$ or $\\mu^+\\mu^-$ pair. The displaced vertex suppresses most backgrounds, and kinematic constraints that close the decay chain up to a two-fold ambiguity allow the sterile neutrino mass to be reconstructed. The paper estimates 95% confidence-level reach on $|V_{\\tau N}|^2$ that beats the current best limits, reaching about $10^{-5}$ for $m_N \\approx 1$ GeV at Belle II.","feed_headline":"Tau-decay search could probe sterile neutrinos down to 10^-5","feed_subtitle":"Using displaced vertices from e+e−→τ+τ− events, Belle II could reach tau-neutrino mixing below 10^-5.","key_machinery":"The load-bearing mechanism is the displaced-vertex decay of the long-lived sterile neutrino combined with a kinematic-closure fit. The lifetime grows as $\\tau_N \\propto m_N^{-5}$, so for sub-GeV masses the $N$ flies tens of centimeters before decaying; the acceptance is defined by a decay-volume cut (10--80 cm in radius, $-40$ to $120$ cm in $z$) that rejects prompt tracks and most $K_S$ and material-interaction backgrounds. The fit uses 12 constraints — four-momentum conservation in the $\\tau$ and $N$ decays, the known $\\tau$ mass, the massless final neutrino, and the measured direction of flight of $N$ — to solve for the 12 unknown momentum components, leaving a two-fold ambiguity that yields two candidate masses $m_1,m_2$ and two candidate $\\tau$ energies $E_1,E_2$. For signal, one solution peaks at the true $m_N$ and $E_\\tau = \\sqrt{s}/2$; for the dominant $K_L$ background the distributions are broad. The paper also uses the seesaw-motivated effective Lagrangian of Eq. (2), with the branching fractions of the production and decay modes taken from Refs. [63] and [68].","core_discovery":"The paper's central claim is that a sterile neutrino $N$ with $m_N < m_\\tau = 1.777$ GeV and mixing $|V_{\\tau N}| \\gg |V_{eN}|, |V_{\\mu N}|$ can be probed at B-factories through the decay chain $e^+e^- \\to \\tau^+\\tau^-$, $\\tau \\to X_1 N$, $N \\to \\nu_\\tau X_2$, where $X_1 = \\pi^\\pm$ or $\\pi^\\pm\\pi^0$ and $X_2 = \\ell^+\\ell^-$. Because $N$ is long-lived, its decay vertex is displaced from the interaction point, and the paper shows that the unobserved neutrino still leaves enough constraints to determine the $N$ momentum and mass up to a two-fold ambiguity: four-momentum conservation in both decays, the known $\\tau$ mass, the massless $\\nu_\\tau$, and the measured flight direction of $N$. These constraints make the $m_1$--$m_2$ and $E_1$--$E_2$ distributions narrow for signal and broad for background, so a simple cut retains 90% of signal while rejecting over 97% of the background. With the assumed detector efficiencies, the projected limits on $|V_{\\tau N}|^2$ improve on the existing bound and reach about $10^{-5}$ at Belle II for $m_N$ near 1 GeV, with the full Belle II sample expected to be close to background-free.","pith_inferences":["Editorial inference: the same kinematic closure could in principle reconstruct $m_N$ from the leptonic tau mode $\\tau \\to \\ell\\nu N$, but the extra missing neutrino would make the fit underdetermined; the hadronic $X_1$ choice is what makes the mass extraction practical.","Editorial inference: because the $N$ lifetime scales as $m_N^{-5}$, sensitivity below $m_N \\sim 0.3$ GeV is limited mainly by the decay volume — most decays would happen beyond the 80 cm cylinder — so a detector with a larger instrumented radius could extend the search to lower masses.","Editorial inference: the displaced $e^+e^-$ or $\\mu^+\\mu^-$ signature is flavor-blind, so the same selection could be reinterpreted to constrain a sterile neutrino with a subdominant $V_{\\mu N}$ or $V_{eN}$ component, providing a cross-check on the tau-mixing interpretation.","Editorial inference: if the predicted background of roughly six events in the entire Belle II sample is realized, a small cluster of events in the same-mass window would already be a strong new-physics signal, making early Belle II data scientifically interesting."],"forward_implications":["BaBar and Belle data already in hand could improve on the current limit for $m_N < m_\\tau$ without any new accelerator running.","With the full Belle II sample the search becomes nearly background-free, and the projected limit around $10^{-5}$ at $m_N \\approx 1$ GeV would reach a corner of parameter space that lepton-flavor-based searches and dedicated long-lived-particle experiments are not designed to cover.","If a signal appears, the $(m_1,m_2)$ and $(E_1,E_2)$ distributions give a direct measurement of the sterile neutrino mass and confirm that it originated in a $\\tau$ decay.","Including the modes $X_1 = \\ell\\nu$ and $X_2 = \\pi^+\\pi^-$ would roughly double the exploited branching fraction and further improve sensitivity, provided full simulation validates the assumed efficiencies."],"supporting_citations":[{"why":"Supplies the current best limit on $|V_{\\tau N}|^2$ for $m_N < m_\\tau$; the paper's projected sensitivities are compared against it.","marker":"[37]"},{"why":"Provides the formulas used for the production branching fractions $B(\\tau \\to \\pi N)$ and $B(\\tau \\to \\pi\\pi N)$ after the replacement $N \\to \\tau$, $\\ell \\to N$.","marker":"[63]"},{"why":"Supplies the $N$ decay width, lifetime, and leptonic branching fractions $N \\to \\nu_\\tau \\ell^+\\ell^-$ that set the signal acceptance and yield.","marker":"[68]"},{"why":"The event generator used to simulate signal and $K_L$ background events and to compute the detector acceptance.","marker":"[64]"},{"why":"Used to generate $e^+e^- \\to \\tau^+\\tau^-$ events with initial-state radiation, giving the $E_\\tau$ smearing that shapes the mass-energy cut efficiencies.","marker":"[69]"},{"why":"Defines the Belle II detector geometry and beam energies that determine the acceptance volume and the projected event-sample size.","marker":"[40]"},{"why":"A previous B-factory displaced-vertex search; source of the pion-to-muon misidentification rate and displaced-vertex background rejection practice.","marker":"[45]"},{"why":"A previous B-factory displaced-vertex search; basis for the vertex-rejection requirements and the 90% efficiency assigned to them.","marker":"[46]"}],"fun_headline_variants":["B-factory tau decays: a new probe for sterile neutrinos","Belle II could reach 10^-5 mixing for tau-coupled sterile neutrinos","Displaced tau-decay vertices: a new key to sterile neutrinos","Tau decay search at B-factories: new sterile neutrino probe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projected sensitivity rests on hand-assigned detector efficiencies and background estimates — especially the 25% track-reconstruction efficiency for displaced tracks, the 90% vertex-rejection efficiency, and the $7.5 \\times 10^{-5}$ acceptance-efficiency product for the $K_L$ background — rather than on a full detector simulation, and the paper itself notes the pion misidentification rate could vary by a factor of two.","fun_headline_variants_meta":{"raw":{"variants":["B-factory tau decays: a new probe for sterile neutrinos","Belle II could reach 10^-5 mixing for tau-coupled sterile neutrinos","Displaced tau-decay vertices: a new key to sterile neutrinos","Tau decay search at B-factories: new sterile neutrino probe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001206,"raw_usage":{"total_tokens":5008,"prompt_tokens":1028,"completion_tokens":3980,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":644,"completion_tokens_details":{"reasoning_tokens":3900}},"tokens_in":644,"tokens_out":3980,"duration_ms":29529,"temperature":1.0,"reasoning_tokens":3900,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:03:45.844931+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A full detector simulation of $e^+e^- \\to \\tau^+\\tau^-$ events containing $\\tau \\to \\pi K_L \\nu$, $K_L \\to \\pi \\mu \\nu$ in Belle II, with the proposed four-track, displaced-dilepton-vertex, and $m_1/m_2$/$E_1/E_2$ selection applied, would settle the projection: if the background yield exceeds about 17 events per ab$^{-1}$ or the displaced-track signal efficiency falls below roughly 10%, the claimed background-free sensitivity and the $\\sim 10^{-5}$ limit would not survive.","supporting_citations":[{"cited_title":"Abreu et al","cited_arxiv_id":null,"evidence_quote":"Supplies the current best limit on $|V_{\\tau N}|^2$ for $m_N < m_\\tau$; the paper's projected sensitivities are compared against it."}],"review_version":1}