REVIEW 1 major objections 8 minor 1 cited by
Search for lepton flavor-violating decay modes $B^0\to K_S^0\tau^\pm\ell^\mp~(\ell=\mu, e)$ with hadronic $B$-tagging at Belle and Belle II
T0 review · 1 major / 8 minor · reviewed 2026-08-11 · deepseek-v4-flash
Pith's one-line read The first search for $B^0\to K_S^0\tau^\pm\ell^\mp$ decays finds no signal and sets 90% confidence-level upper limits on the branching fractions in the range $[0.8,3.6]\times10^{-5}$.
desk verdict First search for B0 -> KS0 tau l is cleanly executed and the null-result limits are credible as stated, with a disclosed phase-space model dependence that the efficiency maps help mitigate. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the reconstructed recoil mass $M_\tau$, obtained from the beam energy and the momentum of the fully reconstructed other $B$ meson: since the signal $B$ is not fully visible (the $\tau$ decays with neutrinos), its momentum is taken as $\vec p_\tau = -\vec p_{B_{\rm tag}} - \vec p_{K_S^0} - \vec p_\ell$ and its energy as $E_\tau = E_{\rm beam} - E_{K_S^0} - E_\ell$. Signal events accumulate at the known $\tau$ mass while backgrounds are smooth. The tagging itself uses the full-event-interpretation algorithm, a machine-learning $B$-tagging tool with an average efficiency of 0.59% and 44% purity, calibrated on control samples ($B^0\to D^-\pi^+$ for the tag efficiency and $B^0\to D_s^+D^-$ for the signal PDF and the boosted-decision-tree selection). Signal decays are modeled with a uniform three-body phase-space distribution.
What would settle it
A future dataset with substantially more integrated luminosity would settle the claim: if the reconstructed $\tau$ mass distribution shows a peak at the $\tau$ mass in any of the four channels with a yield incompatible with background, the no-signal claim is wrong; alternatively, recomputing the limits with a specific non-uniform signal model using the published efficiency maps would show how much the limits shift.
Extended reading notes
Core claim
On its own terms, the paper establishes that $B^0\to K_S^0\tau^+\mu^-$, $B^0\to K_S^0\tau^-\mu^+$, $B^0\to K_S^0\tau^+e^-$, and $B^0\to K_S^0\tau^-e^+$ do not appear in the combined Belle and Belle II data at rates above a few times $10^{-5}$. The signal is searched for in the mass recoiling against a hadronically tagged $B$ meson; the reconstructed quantity $M_\tau$ peaks at the known $\tau$ mass for signal and is smooth for background. With no significant signal in any channel, the authors set observed upper limits at 90% CL of $1.1\times10^{-5}$, $3.6\times10^{-5}$, $1.5\times10^{-5}$, and $0.8\times10^{-5}$, respectively, and report expected limits of $[2.1,2.2]\times10^{-5}$. They also supply efficiency maps in ($M_{\tau\ell}^2$, $M_{K_S^0\ell}^2$) so the limits can be reinterpreted for signal models that do not follow a uniform three-body phase space.
Load-bearing premise
The quoted upper limits assume that the signal decays follow a uniform three-body phase-space distribution; if a real new-physics signal were distributed differently in phase space, the efficiency and therefore the limits would change.
Editorial extensions
If this is right
- These are the first upper limits on $B^0\to K_S^0\tau^\pm\ell^\mp$; no other experiment has searched for these modes.
- The limits on $B^0\to K_S^0\tau^\pm e^\mp$ are the most stringent on $b\to s\tau e$ transitions to date.
- The $B^0\to K_S^0\tau^\pm\mu^\mp$ limits are among the best on $b\to s\tau\mu$ transitions achieved so far.
- The results approach the $\mathcal{O}(10^{-6})$ branching-fraction level predicted by some models motivated by the $B^+\to K^+\nu\bar\nu$ excess, so the search is beginning to test those scenarios.
- The efficiency maps provided in the supplemental material let the limits be recast for arbitrary signal kinematics, not just the uniform phase-space model.
Reading between the lines
- If the Belle II $B^+\to K^+\nu\bar\nu$ excess is due to new physics with off-diagonal lepton-flavor couplings, predictions in that framework place $B(B\to K\tau^\pm\mu^\mp)$ near $[2,3]\times10^{-6}$; the limits here are roughly a factor of 5-10 above that, so with several times more Belle II data the same technique would either observe the decay or exclude the simplest version of that explanation
- The uniform phase-space assumption is likely conservative for signals that populate the high-efficiency region of phase space, but could understate limits for signals concentrated elsewhere; the published efficiency maps make this testable, and a recast for vector, scalar, or tensor operators would quantify the shift.
- The inclusion of the $\tau\to\rho\nu$ decay mode, which carries more than 20% of the $\tau$ branching fraction and had not been used in earlier $B\to K\tau\ell$ searches, suggests that similar searches for related modes such as $B^0\to K^{*0}\tau^\pm\ell^\mp$ or $B_s$ decays could gain sensitivity by the same route.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper reports the first search for the lepton flavor-violating decays B0→KS0τ±ℓ∓ (ℓ=μ,e), using 711 fb−1 of Belle and 365 fb−1 of Belle II data. Events are selected with a hadronic FEI B-tag; the signal is identified via the recoil mass Mτ reconstructed from the Btag, the KS0, and a primary lepton, with τ candidates reconstructed in the e, μ, π, and ρ decay modes. An unbinned maximum-likelihood fit to Mτ is performed on the combined data sample; no significant signal is found, and 90% CL upper limits on the branching fractions in [0.8, 3.6]×10−5 are set. The analysis is calibrated on control channels: B0→D−π+ for the tag-efficiency scale factors R_FEI, B0→Ds−D+ for the signal PDF and BDT efficiency ratios R_BDT, with a closure test of B(B0→Ds−D+) consistent with the world average. Limit-setting uses pseudo-experiments with systematic smearing.
Significance. If it holds, this result provides the first constraints on B0→KS0τ±ℓ∓; the electron modes are the most stringent limits on b→sτe transitions, and the muon modes are competitive with the best existing b→sτμ bounds. The limits approach the O(10−6) enhancements suggested by the BSM models cited in the paper. The strengths of the analysis are its data-driven calibration chain with an independent closure test, pseudo-experiment validation of the fitting procedure, and the Supplemental efficiency maps as functions of (M2τℓ, M2KS0ℓ), which provide a genuine route toward model reinterpretation. The principal caveat is the uniform-phase-space signal model underlying the headline limits, assessed in the major comments.
major comments (1)
- [Simulation section; Table I; Supplemental Fig. 1] The quoted 90% CL upper limits in Table I and in the abstract and conclusion are derived from a signal model with a uniform three-body phase-space distribution, as stated in the Simulation section ('The B0→KS0τ±ℓ∓ signal channels are modeled using an uniform three-body phase space model'). This assumption is load-bearing: the Supplemental efficiency maps show that the per-event acceptance varies from near zero to about 8×10−4 across the (M2τℓ, M2KS0ℓ) plane, whereas the phase-space-averaged efficiencies in Table I are only 1.7–2.1×10−4. A BSM signal with a different Dalitz distribution (such as the models of Refs. [3–5]) could therefore change the signal efficiency, and hence the limits, by roughly a factor of two to four, which is larger than the quoted 22–24% total systematic uncertainty. The efficiency maps are a useful mitigation, but they are binned in reconstructed variables (the τ momentum is inferred from the Btag), so a reinterpretation requires detector-level simulation, and the Mτ signal-PDF shape used in the profile fit is fixed to the phase-space sample, an effect the maps do not encode. I request: (i) an explicit qualification in the abstract and conclusion that the limits assume a uniform three-body phase-space model, and (ii) preferably, a quantitative robustness study in which the signal MC is reweighted to a few representative non-uniform Dalitz models and the efficiencies (or, if feasible, the full fit and ULs) are recomputed, so that the reader can see the range of possible limit shifts.
minor comments (8)
- [Simulation] The phrase 'an uniform three-body phase space model' should read 'a uniform three-body phase-space model'; please also use 'phase-space' consistently throughout.
- [Systematic uncertainties] The text quotes the R_FEI uncertainty as '4%', but the quoted values 0.74±0.04 and 0.81±0.04 correspond to relative uncertainties of 5.4% and 4.9%; the sentence should be corrected.
- [Fig. 2] The horizontal-axis label of the lower-right panel in Figure 2 appears garbled ('M2 −0' ...); please verify the typesetting.
- [Table I] Since the best-fit branching-fraction central values are negative in three of the four channels, the column label 'central value' is confusing; 'best-fit value' would be more accurate.
- [Introduction, Ref. [10]] Reference [10] combines two distinct publications (Kurokawa and Kikutani; Abe et al.); they should be cited separately.
- [Supplemental material] The caption of Supplemental Fig. 1 should state explicitly that the τ four-momentum used for M2τℓ is the value inferred from the Btag reconstruction, as this information currently appears only in the body text.
- [Signal categories] The definitions of 'same-sign' (SSℓ) and 'opposite-sign' (OSℓ) should be clarified in one sentence, since τ±ℓ∓ pairs have opposite charges in both cases.
- [Results] The expected ULs are quoted only as the range [2.1, 2.2]×10−5; listing the expected limit for each channel in Table I would help the reader assess the observed SSe limit (0.8×10−5), which is a downward fluctuation relative to expectation.
Circularity Check
No circularity: the upper limits come from a data fit with control-sample calibrations, and the uniform-phase-space signal model is a disclosed model dependence, not an input recycled as the result.
full rationale
The central result, the 90% CL upper limits on B(B0→KS0τ±ℓ∓), is obtained by an unbinned maximum-likelihood fit to the Mτ distribution in data, with the signal yield Nsig floated and the branching fraction computed from Eq. 2 using an efficiency calibrated on independent control channels. The calibration factors RFEI, RBDT, and f are measured from B0→D−π+ and B0→Ds+D− control samples, not from the signal region, and the Ds+D− measurement is checked against the world average as a closure test. No fitted parameter is renamed as a prediction, and no equation reduces to its own input by construction. The stated uniform three-body phase-space model for the signal is an assumption about the unknown BSM kinematics, and the paper explicitly provides efficiency maps in the Supplemental Material so results can be reinterpreted for other models; this is model dependence, not circularity. Self-citations to Belle and Belle II software, detector, and PID references are supporting infrastructure rather than load-bearing premises that assume the target result. Therefore no circular step is present.
Assumptions & free parameters
free parameters (7)
- R_FEI (Belle) =
0.74 ± 0.04
- R_FEI (Belle II) =
0.81 ± 0.04
- R_BDT (OS mu) =
0.93 ± 0.17
- R_BDT (SS mu) =
0.96 ± 0.16
- R_BDT (OS e) =
0.92 ± 0.16
- R_BDT (SS e) =
0.96 ± 0.18
- Signal width correction factor f =
1.04 ± 0.15
assumptions (4)
- domain assumption Signal decays are distributed according to a uniform three-body phase-space model.
- domain assumption The M_tau signal shape is described by a Johnson function with parameters fixed to simulation, except mean and width (with correction f).
- domain assumption The M_tau background is smooth and adequately modeled by a second-order polynomial.
- domain assumption B0-anti-B0 mixing is simulated correctly and accounted for in the efficiency.
Cite this review
Pith. "Pith review of Search for lepton flavor-violating decay modes $B^0\to K_S^0\tau^\pm\ell^\mp~(\ell=\mu, e)$ with hadronic $B$-tagging at Belle and Belle II." pith.science (2026). https://pith.science/paper/Y7RUSYC5
@misc{pith2026241216470,
author = {Pith},
title = {Pith review of: Search for lepton flavor-violating decay modes $B^0\to K_S^0\tau^\pm\ell^\mp~(\ell=\mu, e)$ with hadronic $B$-tagging at Belle and Belle II},
year = {2026},
howpublished = {\url{https://pith.science/paper/Y7RUSYC5}},
note = {Machine review of arXiv:2412.16470}
}
abstract
We present the first search for the lepton flavor-violating decay modes $B^0 \rightarrow K_S^0 \tau^\pm \ell^\mp~(\ell=\mu, e)$ using the 711 fb$^{-1}$ and 365 fb$^{-1}$ data samples recorded by the Belle and Belle II detectors, respectively. We use a hadronic $B$-tagging technique, and search for the signal decay in the system recoiling against the fully reconstructed $B$ meson. We find no evidence for $B^0 \rightarrow K_S^0 \tau^\pm \ell^\mp$ decays and set 90\% confidence level upper limits on the branching fractions in the range of $[0.8,\,3.6]\times10^{-5}$.
Figures
Forward citations
Cited by 1 Pith paper
-
Measurement of the CP asymmetry in $D^+ \to \pi^+ \pi^0$ decays at Belle II
The new world-best measurement of the CP asymmetry in D+ -> pi+ pi0 decays is (-1.8 +/- 0.9 +/- 0.1)%, consistent with zero and with the Standard Model.
Reference graph
Works this paper leans on
- [1]
-
[2]
I. Adachi et al . (Belle II Collaboration), Phys. Rev. D 109, 112006 (2024)
work page 2024
-
[3]
D. Becirevi´ c, O. Sumensari, and R. Zukanovich Funchal, Eur. Phys. J. C 76, 134 (2016)
work page 2016
- [4]
-
[5]
S. L. Glashow, D. Guadagnoli, and K. Lane, Phys. Rev. Lett. 114, 091801 (2015)
work page 2015
-
[6]
J. P. Lees et al . (BABAR Collaboration), Phys. Rev. D 86, 012004 (2012)
work page 2012
-
[7]
S. Watanuki et al. (Belle Collaboration), Phys. Rev. Lett. 130, 261802 (2023)
work page 2023
- [8]
Show all 35 references
-
[9]
Aaij et al
R. Aaij et al . (LHCb Collaboration), J. High Energy Phys. 06 2023 143
2023
-
[10]
Kurokawa and E
S. Kurokawa and E. Kikutani, Nucl. Instrum. Methods Phys. Res., Sect. A 499, 1 (2003). T. Abe et al , Prog. Theor. Exp. Phys. (2013) 03A001-03A011
2003
-
[11]
K. Akai, K. Furukawa, and H. Koiso, Nucl. Instrum. Methods Phys. Res., Sect. A 907, 188 (2018)
2018
-
[12]
Abashian et al
A. Abashian et al . (Belle Collaboration), Nucl. Instrum. Methods Phys. Res., Sect. A 479, 117 (2002)
2002
- [13]
-
[14]
D. J. Lange, Nucl. Instrum. Methods Phys. Res., Sect. A 462, 152 (2001)
2001
-
[15]
Barberio and Z
E. Barberio and Z. Was, Comp. Phys. Commun. 79, 291 (1994)
1994
-
[16]
B. Ward, S. Jadach and Z. Was, Nucl. Phys. B Proc. Suppl. 116, 73 (2003)
2003
-
[17]
Sj¨ ostrandet al ., Comput
T. Sj¨ ostrandet al ., Comput. Phys. Commun. 178, 852 (2008)
2008
-
[18]
Brun et al ., CERN-DD-EE-84-01 (1987)
R. Brun et al ., CERN-DD-EE-84-01 (1987)
1987
-
[19]
Agostinelli et al ., Nucl
S. Agostinelli et al ., Nucl. Instrum. Methods Phys. Res., Sect. A 506, 250 (2003)
2003
-
[20]
Kuhr et al
T. Kuhr et al. (Belle II Collaboration), Comput. Software Big Sci. 3, 1 (2019)
2019
-
[21]
Gelb et al ., Comput
M. Gelb et al ., Comput. Software Big Sci. 2, 9 (2018)
2018
-
[22]
Keck et al ., Comput
T. Keck et al ., Comput. Software Big Sci. 1, 6 (2019)
2019
-
[23]
Navas et al
S. Navas et al. (Particle Data Group), Phys. Rev. D 110, 030001 (2024)
2024
-
[24]
Won et al
E. Won et al . (Belle Collaboration), Phys. Rev. D 80, 111101(R) (2009)
2009
-
[25]
Nakano, Nucl
E. Nakano, Nucl. Instrum. Methods Phys. Res., Sect. A 494, 402 (2002). 8
2002
-
[26]
Hanagaki, H
K. Hanagaki, H. Kakuno, H. Ikeda, T. Iijima, and T. Tsukamoto, Nucl. Instrum. Methods Phys. Res., Sect. A 485, 490 (2002)
2002
-
[27]
D. N. Brown, J. Ilic, and G. B. Mohanty, Nucl. Instrum. Methods Phys. Res., Sect. A 592, 254 (2008)
2008
-
[28]
Khotanzad and Y
A. Khotanzad and Y. Hong, IEEE Trans. Pattern Anal. Mach. Intell. 12, 489 (1990)
1990
-
[29]
Longo et al., Nucl
S. Longo et al., Nucl. Instrum. Methods Phys. Res., Sect. A 982, 164562 (2020)
2020
-
[30]
J. D. Bjorken and S. J. Brodsky, Phys. Rev. D 1, 1416 (1970)
1970
-
[31]
S. H. Lee et al . (Belle Collaboration), Phys. Rev. Lett. 91, 261801 (2003)
2003
-
[32]
Punzi, eConf C030908, MODT002 (2003)
G. Punzi, eConf C030908, MODT002 (2003)
2003
-
[33]
N. L. Johnson, Biometrika 36, 149 (1949)
1949
- [34]
-
[35]
Aggarwal et al
L. Aggarwal et al . (Belle II Collaboration), Phys. Rev. Lett. 131, 051804 (2023). Supplemental material SUPPLEMENT AL FIGURES Figure 1 presents the selection efficiencies for the four signal modes: B0 → K 0 Sτ +µ−, B0 → K 0 Sτ −µ+, B0 → K 0 Sτ +e−, and B0 →K 0 Sτ −e+. The eff...
2023 arXiv
Reviewed August 11, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.