Pith. sign in

REVIEW 3 major objections 5 minor 50 references

Amplitude analysis for charmed meson decays at BESIII

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read This paper claims to give the complete BESIII amplitude-analysis methodology, from measured charm-decay events to intermediate-resonance magnitudes, phases, fit fractions, and branching fractions.

desk verdict A useful but flawed methods review: the background-normalization formula in Eq. (11) inverts an importance-sampling weight and must be corrected before anyone uses it as a reference. read the letter →

arxiv 2608.09338 v1 pith:2DUDY7V6 submitted 2026-08-10 hep-ex

classification hep-ex
keywords amplitudeanalysischarmedmesondecaysBESIIIunbinnedmaximumlikelihoodisobarmodelXGBoostbackgroundmodelingresonancepropagatorsfitfractions
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Amplitude analysis is the bridge between what a detector records — stable final-state particles — and the short-lived intermediate resonances whose couplings theory can predict. This paper lays out the complete machinery BESIII uses for charmed-meson decays: a signal density built from a coherent sum of intermediate amplitudes, normalized by Monte Carlo integration that folds in detection efficiency and resolution, and a likelihood extended with a background shape learned by an XGBoost classifier. If the account is correct, Eqs. (1)–(18) form the operational route from BESIII's $20.3\ \mathrm{fb}^{-1}$ dataset at $\sqrt{s}=3.773$ GeV to fitted resonance magnitudes, phases, fit fractions, and branching fractions. The paper's value is that it makes that route explicit enough to be checked, reproduced, or modified.

What carries the argument

The load-bearing object is the unbinned maximum-likelihood signal density $f_S(p)=\epsilon(p)|M(p)|^2R(p)/\int\epsilon(p)|M(p)|^2R(p)\,dp$, together with the extended log-likelihood of Eq. (6). Three mechanisms carry the argument: Monte Carlo integration over PHSP, reconstructed, and signal-MC samples turns normalization, efficiency, and resolution into sums over simulated events; the XGBoost classifier's odds ratio supplies the corrected background density $B_\epsilon(p)$; and the isobar amplitude $A_n=P_n S_n F_n^r F_n^D$ connects resonance lineshapes to data through spin factors, Blatt-Weisskopf barriers, and propagators such as Breit-Wigner, Gounaris-Sakurai, Flatt\'e, $K$-matrix, and LASS forms.

What would settle it

Generate pseudo-experiments from a known amplitude model plus a known background density, apply the XGBoost-based likelihood of Eq. (6), and check whether the fitted magnitudes and phases recover the injected values within statistical uncertainty. A systematic miss concentrated in any phase-space region would show that the density-ratio assumption behind $B_\epsilon$ fails.

Watch

Extended reading notes

Core claim

The central claim is that Eqs. (1)–(18) faithfully describe the amplitude-analysis methodology used by the BESIII Collaboration. The signal probability density is $f_S(p)=\epsilon(p)|M(p)|^2 R(p)/\int \epsilon(p)|M(p)|^2 R(p)\,dp$, and the total amplitude is $M(p)=\sum_n c_n A_n(p)$, a coherent sum over intermediate processes. The likelihood is extended with an efficiency- and phase-space-corrected background density $B_\epsilon(p)$ obtained from an XGBoost odds ratio $P_{\rm BKG}(p)/P_{\rm PHSP}(p)$. Monte Carlo integration over reconstructed samples folds in detector efficiency and resolution without analytic modeling; fit fractions are defined at generator level to isolate dynamics from acceptance; and the semileptonic decay rate is factorized into hadronic and leptonic currents in five kinematic variables. The authors present this formula chain as the map from measured charm-decay events to the intermediate-resonance parameters BESIII publishes.

Load-bearing premise

The whole background treatment depends on the XGBoost odds ratio being the true density ratio of background to phase-space events, and the paper does not show a calibration or closure test of that equality.

Editorial extensions

If this is right

  • Every published BESIII branching fraction or fit fraction from these channels can be traced to the likelihood, amplitude, and background definitions in Eqs. (1)–(18).
  • Generator-level PHSP Monte Carlo gives fit fractions that are independent of detector acceptance, so comparisons between experiments are meaningful only when the same propagator and production-vector conventions are used.
  • Reconstructed signal-MC normalization removes the need for multidimensional resolution convolution: detector smearing is folded in by evaluating $|M(p^{\rm rec})|^2/|M_{\rm gen}(p^{\rm rec})|^2$.
  • The XGBoost background model lets arbitrarily complex multidimensional background shapes enter the fit, provided the odds ratio is a faithful density ratio.
  • Simultaneous fits across multiple $D$ and $D_s$ channels can be performed on the same likelihood footing, which is the paper's route to probing $K^0_S$–$K^0_L$ asymmetries and $U$-spin breaking.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A validation the authors leave out: calibration curves or pseudo-experiment closure tests for the XGBoost odds ratio; a miscalibrated classifier would shift every fitted phase and magnitude in the same direction.
  • The same density-ratio normalization trick could be transplanted to unbinned amplitude fits at other charm or beauty facilities, where background compositions differ but the mathematics of Eq. (6) does not.
  • The fit-fraction uncertainty prescription (Gaussian width from covariance-matrix sampling) is explicitly flagged by the paper as fragile near boundaries; a natural extension is to report full posterior shapes instead of single Gaussian widths.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This paper is a methodological review of the amplitude-analysis framework used by the BESIII Collaboration for multibody charmed-meson decays. It develops the signal probability density function and the unbinned likelihood, describes Monte Carlo normalization with PHSP and signal MC samples, discusses efficiency and resolution treatment, introduces an XGBoost-based multidimensional background reweighting procedure, and explains projection and fit-fraction evaluation. It then summarizes the resonance parametrizations used in the collaboration's hadronic and semileptonic analyses, including relativistic Breit-Wigner, Gounaris-Sakurai, Flatté, coupled-channel, K-matrix, and LASS forms.

Significance. If the formulas are correct, the paper would serve as a useful practical reference for amplitude analyses at BESIII, consolidating the standard isobar-model and K-matrix machinery with concrete normalization and background-modeling recipes. It is honest about the limitations of Gaussian uncertainties for fit fractions near physical boundaries and clearly distinguishes reconstructed MC from generator-level PHSP MC, which is valuable for practitioners. The paper presents no new data or measurements, so its significance rests entirely on the accuracy and reproducibility of the methodological equations; one of the load-bearing equations, Eq. (11), is erroneous as written, and the key density-ratio assumption of Section II B is not validated.

major comments (3)
  1. [II B, Eqs. (11)–(12)] The importance-sampling estimator for the background normalization integral is written with the wrong power of |M_gen|^2. If the reconstructed signal-MC sample is distributed as ε(p)|M_gen(p)|^2 R(p), then the estimator for ∫ ε(p)B_ε(p)R(p)dp = ∫ B(p)dp should use the weight [P_BKG/P_PHSP]/|M_gen|^2, not [P_BKG/P_PHSP]·|M_gen|^2. As printed, the sum converges to a weighted integral proportional to ∫ B(p)|M_gen(p)|^2 dp, so the normalized background shape in Eq. (12) and hence the background term in Eq. (6) inherit a spurious dependence on the generator amplitude M_gen. This is an internal algebraic error that does not vanish unless |M_gen|=1, which is inconsistent with the text's instruction to use the same signal-MC footprint as the signal term in Eq. (10). Please correct Eqs. (11) and (12) and re-derive the normalization, or explicitly clarify the density of the MC sample being summed.
  2. [II B] The central identification P_BKG(p)/P_PHSP(p) = B_ε(p) = B(p)/[ε(p)R(p)] is presented as a direct consequence of density-ratio estimation, but the paper gives no validation that the trained XGBoost classifier reliably recovers this ratio over the full phase space. A miscalibrated classifier, an input feature set that misses discriminating correlations, or a PHSP MC sample that does not faithfully map ε(p)R(p) would bias the background shape and propagate into every fitted amplitude. The review should state explicitly that calibration curves, closure tests, or pseudo-experiments are required for this step and should point to the BESIII analyses (e.g., Refs. [7], [29], [34]) where such validation is performed.
  3. [II, Eq. (7)] Equation (7) is not unambiguously typeset: the fraction structure among N_data, wN_bkg, and N_data + w^2 N_bkg is unclear in the provided text, and no derivation or reference is given for the prefactor that is claimed to ensure correct statistical uncertainties. Since this equation is offered as an alternative background-subtracted likelihood, the authors should rewrite the formula unambiguously and either derive the prefactor from the statistical weights or cite the specific BESIII analysis where this form is used.
minor comments (5)
  1. [II A, Eq. (10)] Please clarify whether |M_gen| in the denominator is evaluated at the generator-level momentum or at the reconstructed momentum p_rec. As written, using p_rec in both numerator and denominator is not the standard importance-sampling weight unless resolution effects are deliberately included through a separate empirical procedure; this should be stated explicitly.
  2. [II B] The description of the XGBoost procedure would be more reproducible if the input feature set and training hyperparameters were listed, or if the authors stated that these are channel-dependent and referred to the original BESIII analyses for details.
  3. [III C 6, Eq. (27)] The text should make explicit that A_i, P_j, K, and ρ are matrices/vectors in the same five-channel basis (ππ, KKbar, 4π, ηη, ηη′), and that the indices i, j denote these channels throughout.
  4. [III C 5, Eq. (26)] The bare P used for the Cauchy principal value should be typeset as ℘ or “PV” to avoid confusion with a kinematic variable.
  5. [References] Refs. [38]–[40] are arXiv preprints; if they have appeared in journals by the time of publication, the published versions should be cited.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a methodology review whose equations are standard MC and likelihood identities; BESIII and self-citations are external anchors, not inputs to the derivation.

full rationale

This is a methodology review, not an original measurement, so no fitted parameter is renamed as a prediction and no result reduces to its own input. The signal PDF (Eq. (1)), likelihood (Eqs. (3)-(6)), and MC normalization (Eqs. (8)-(10)) are standard importance-sampling identities: the reconstructed signal-MC sample is distributed as epsilon(p)|M_gen(p)|^2 R(p), and dividing by |M_gen|^2 removes the generation model from the normalization integral. The background shape B_epsilon is learned by XGBoost from independent inclusive-MC or sideband samples, not from the fitted amplitude coefficients, so the background term does not encode the parameters it is used to extract. The BESIII experimental papers cited as Refs. [7], [28]-[40] supply measured branching fractions and fit results used for context or external validation; the self-citations such as Ref. [9] point to standard formalism reviews whose key equations, including Eq. (18), are stated in the present paper. No load-bearing argument reduces to a self-citation. A possible algebraic issue in Eq. (11) (the odds-ratio weight may require division by |M_gen|^2 rather than multiplication) is a correctness concern about MC integration, not a circularity, because it does not make an output equal to an input by construction. The paper is therefore self-contained as a review, and no circular step is present.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The paper introduces no new fitted parameters, axioms, or entities of its own; it is a review of an established methodology. The items listed are the elements of the described method on which a faithful reconstruction of the workflow depends: the isobar-model ansatz, CP conservation, lepton-hadron factorization, multiplicative efficiency, unbiased Monte Carlo importance sampling, the XGBoost density-ratio identification, and the asserted but unproven statistical prefactor of Eq. (7). A reader implementing the method must trust or supply each of these externally.

free parameters (4)
  • Resonance coefficients c_n = rho_n exp(i phi_n) = fitted in the described analyses, not in this paper
    Section II, Eq. (2): magnitudes and phases of intermediate amplitudes are the fit outputs of the analyses being reviewed; they are parameters of the described method, not quantities this paper determines.
  • Resonance masses and widths (m_0, Gamma_0) = usually fixed to PDG values, optionally free
    Section III C 1: the propagator formulas treat masses and widths as inputs, fixed to Ref. [1] or left free in fits; the paper provides no values.
  • K-matrix production parameters f_prod and beta_alpha = left free in fits
    Section III C 6, Eq. (28): process-dependent production parameters are explicitly left free; the paper does not give their values.
  • XGBoost hyperparameters and feature set = not specified
    Section II B: the background density estimate depends on training choices that the paper does not enumerate, limiting reproduction.
assumptions (7)
  • domain assumption The multibody decay amplitude is a coherent sum of quasi-two-body amplitudes (isobar model), with each amplitude factorized as spin factor times barrier factors times propagator (Eqs. 16-17).
    Section III A: the entire framework rests on this modeling assumption, which is valid only for well-separated resonances; the paper notes K-matrix alternatives for S-waves.
  • domain assumption CP conservation: the Dbar amplitude equals the D amplitude evaluated at the CP-conjugated phase-space point; for semileptonic Dbar decays, chi goes to -chi.
    Sections III A and III B state 'assuming strict CP conservation' and 'under the assumption of CP conservation'; this is an input symmetry assumption of the fitted formalism.
  • domain assumption Factorization of leptonic and hadronic currents in semileptonic decays (no final-state interactions between the two systems), enabling the five-variable parametrization of Eq. (18).
    Section III B, citing Ref. [9]; a standard but non-trivial assumption of the semileptonic formalism.
  • domain assumption Density-ratio estimation: the XGBoost odds ratio P_BKG/P_PHSP equals B_eps(p) = B(p)/[eps(p)R(p)].
    Section II B: valid if the classifier is well trained and the PHSP MC maps eps(p)R(p); mis-training or correlated features bias the background shape.
  • domain assumption Detection efficiency factorizes as a multiplicative function eps(p) applied to |M|^2 (Eq. 1); efficiency does not interfere with the amplitude.
    Section II A: standard, but only approximate when detector acceptance correlates with decay dynamics in a way not encoded in p.
  • standard math Monte Carlo importance-sampling estimates are unbiased (law of large numbers): Eqs. (9) and (10) approximate the normalization integral.
    Section II A: standard MC integration; the paper states the approximation explicitly.
  • ad hoc to paper The prefactor in Eq. (7) yields correct statistical uncertainties for the background-subtracted likelihood.
    Section II, Eq. (7): asserted without derivation; the form is non-standard and the background sum lacks the weight w present in the same equation, so the correctness of this assertion is unverified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Amplitude analysis for charmed meson decays at BESIII." pith.science (2026). https://pith.science/paper/2DUDY7V6

@misc{pith2026260809338,
  author       = {Pith},
  title        = {Pith review of: Amplitude analysis for charmed meson decays at BESIII},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2DUDY7V6}},
  note         = {Machine review of arXiv:2608.09338}
}
read the original abstract

Amplitude analysis bridges the gap between experimental measurements of multibody charmed-meson decays and theoretical predictions of intermediate two-body processes. This work presents a comprehensive overview of the amplitude-analysis methodology employed by the BESIII Collaboration, emphasizing practical implementation. We detail the construction of the probability density function and likelihood function for unbinned maximum-likelihood fits. This encompasses Monte Carlo integration techniques for normalization, the incorporation of detection efficiency and resolution effects, and multidimensional background modeling utilizing XGBoost classifiers. Furthermore, we describe the amplitude formalism for both hadronic and semileptonic decays, incorporating standard resonance-propagator parametrizations. Key analytical aspects, including the evaluation of fit fractions, the generation of kinematic projections, and the estimation of statistical uncertainties, are also discussed.

Figures

Figures reproduced from arXiv: 2608.09338 by the authors.

Figure 1
Figure 1. FIG. 1. Projections of a background MC sample for [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 41 canonical work pages

  1. [7]

    Agostinelliet al.(GEANT4 Collaboration),Nucl

    S. Agostinelliet al.(GEANT4 Collaboration),Nucl. Instrum. Meth. A506, 250 (2003)

  2. [29]

    Aaijet al.(LHCb Collaboration), Eur

    R. Aaijet al.(LHCb Collaboration), Eur. Phys. J. C78, 443 (2018)

  3. [34]

    Ablikimet al.(BESIII Collaboration), JHEP08(2022) 196

    M. Ablikimet al.(BESIII Collaboration), JHEP08(2022) 196

  4. [1]

    Relativistic Breit-Wigner For most isolated resonances that are narrow and far from decay thresholds, a relativis- tic Breit-Wigner (RBW) propagator provides an adequate description [10]. Resonances commonly parameterized in this way includeω,ϕ,b 1(1235),a 1(1260),f 2(1270),a 2(1320), f0(1370),η(1405),a 0(1450),f 0(1500),K ∗(892),K 1(1270),K 1(1400),K ∗ 2...

  5. [2]

    Gounaris-Sakurai For broad vector resonances, such asρ(770) andρ(1450), a simple RBW form fails to describe the lineshape accurately near threshold. In these cases, the Gounaris-Sakurai (GS) parametrization [12] is adopted, which imposes analyticity constraints on theππP-wave amplitude: PGS(s) = 1 +dΓ 0/m0 m2 0 −s+f(s)−im 0Γ(s) .(21) The functionf(s) and ...

  6. [3]

    Their detailed parametrizations and numerical values are taken from Ref

    Hereρ ππ(s) andρ 4π(s) are the Lorentz-invariant phase- space factors for the two-pion and four-pion channels, andg1,2 are the corresponding coupling constants. Their detailed parametrizations and numerical values are taken from Ref. [15]. 4.f 0(980) Thef 0(980) couples strongly toππandK ¯K, and lies just below theK ¯Kmass threshold. A Flatt´ e formula [1...

  7. [4]

    Navaset al.(Particle Data Group), Phys

    S. Navaset al.(Particle Data Group), Phys. Rev. D110, 030001 (2024)

  8. [5]

    N. N. Achasov and G. N. Shestakov, Phys. Rev. D96, 036013 (2017)

Show all 50 references
  1. [6]

    Ablikimet al.(BESIII Collaboration), Chin

    M. Ablikimet al.(BESIII Collaboration), Chin. Phys. C48, 123001 (2024)

  2. [8]

    B. Liu, X. Xiong, G. Hou, S. Song and L. Shen, EPJ Web Conf.214, 06033 (2019)

  3. [9]

    X. D. Team, Xgboost official website, Available at: https://xgboost.readthedocs.io/en/latest/

  4. [10]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. Lett.135, 161902 (2025)

  5. [11]

    B. S. Zou, and D. V. Bugg, Eur. Phys. J. A16, 537-547 (2003)

  6. [12]

    Zhang, B

    H. Zhang, B. C. Ke, Y. Yu and E. Wang, Chin. Phys. C47, 063101 (2023) 19

  7. [13]

    J. D. Jackson, Nuovo Cim.34, 1644-1666 (1964)

  8. [14]

    d’Argentet al.(CLEOc Collaboration), JHEP05, 143 (2017)

    P. d’Argentet al.(CLEOc Collaboration), JHEP05, 143 (2017)

  9. [15]

    G. J. Gounaris and J. J. Sakurai, Phys. Rev. Lett.21, 244 (1968)

  10. [16]

    R. R. Akhmetshinet al.(CMD-2 Collaboration), Phys. Lett. B527, 161 (2002)

  11. [17]

    D. V. Bugg, A. V. Sarantsev, and B. S. Zou, Nucl. Phys. B471, 59 (1996)

  12. [18]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Lett. B598, 149 (2004)

  13. [19]

    Ablikimet al.(BES Collaboration), Phys

    M. Ablikimet al.(BES Collaboration), Phys. Lett. B607, 243 (2005)

  14. [20]

    Y. K. Hsiao, S. Q. Yang, W. J. Wei, and B. C. Ke, JHEP12, 226 (2025)

  15. [21]

    Y. K. Hsiao, Y. Yu, and B. C. Ke, Eur. Phys. J. C80, 895 (2020)

  16. [22]

    Zhang, Y

    H. Zhang, Y. H. Lyu, L. J. Liu, and E. Wang, Chin. Phys. C47043101 (2023)

  17. [23]

    Zhang, H

    X.H. Zhang, H. Zhang, B. C. Ke,et al.Phys. Rev. D110, 114050 (2024)

  18. [24]

    G. S. Adamset al.(CLEO Collaboration), Phys. Rev. D84, 112009 (2011)

  19. [25]

    D. V. Bugg, Phys. Rev. D78, 074023 (2008)

  20. [26]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. D95, 032002 (2017)

  21. [27]

    V. V. Anisovich and A. V. Sarantsev, Eur. Phys. J. A16, 229 (2003)

  22. [28]

    Adachiet al.(BABAR and Belle Collaborations), Phys

    I. Adachiet al.(BABAR and Belle Collaborations), Phys. Rev. D98, 112012 (2018)

  23. [30]

    J. M. Linket al.(FOCUS Collaboration), Phys. Lett. B653, 1 (2007)

  24. [31]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. D99, 091101 (2019)

  25. [32]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. D110, 092006 (2024)

  26. [33]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. D107, 052010 (2023)

  27. [35]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. Lett.134, 011904 (2025)

  28. [36]

    Ablikimet al.(BESIII Collaboration), JHEP09, 077 (2023)

    M. Ablikimet al.(BESIII Collaboration), JHEP09, 077 (2023)

  29. [37]

    Ablikimet al.(BESIII Collaboration), JHEP03(2026) 060

    M. Ablikimet al.(BESIII Collaboration), JHEP03(2026) 060

  30. [38]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. D105, L051103 (2022)

  31. [39]

    M.Ablikimet al.(BESIII Collaboration), Phys. Rev. Lett.129, 182001 (2022)

  32. [40]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. Lett.123, 1112001 (2019)

  33. [41]

    Ablikimet al.(BESIII Collaboration), [arXiv:2604.10444 [hep-ex]]

    M. Ablikimet al.(BESIII Collaboration), [arXiv:2604.10444 [hep-ex]]

  34. [42]

    Ablikimet al.(BESIII Collaboration), [arXiv:2603.18521 [hep-ex]]

    M. Ablikimet al.(BESIII Collaboration), [arXiv:2603.18521 [hep-ex]]

  35. [43]

    Ablikimet al.(BESIII Collaboration), [arXiv:2512.23389 [hep-ex]]

    M. Ablikimet al.(BESIII Collaboration), [arXiv:2512.23389 [hep-ex]]. 20

  36. [44]

    Y. Yu, Y. K. Hsiao, and B. C. Ke, Eur. Phys. J. C81, 1093 (2021)

  37. [45]

    H. Y. Cheng and C. W. Chiang, Phys. Rev. D110, 094052 (2024)

  38. [46]

    Y. J. Shi, J. Zeng and Z. F. Deng, Phys. Rev. D109, no.1, 016027 (2024)

  39. [47]

    Ablikimet al.(BESIII Collaboration), JHEP 09 (2022) 242

    M. Ablikimet al.(BESIII Collaboration), JHEP 09 (2022) 242

  40. [48]

    Ablikimet al.(BESIII Collaboration), JHEP 05 (2025) 195

    M. Ablikimet al.(BESIII Collaboration), JHEP 05 (2025) 195

  41. [49]

    Ablikimet al.(BESIII Collaboration), Phys

    M. Ablikimet al.(BESIII Collaboration), Phys. Rev. Lett.134, 201902 (2025)

  42. [50]

    Ablikimet al.(BESIII Collaboration), arXiv:2605.11464 [hep-ex]

    M. Ablikimet al.(BESIII Collaboration), arXiv:2605.11464 [hep-ex]. 21

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.