SStudy of B and B s Decays at Belle
N. K. Nisar , ∗ Brookhaven National Laboratory,Upton, New York, 11973
E-mail: [email protected]
We report results on the search for B s → η η decay, and the searches for B → invisible and B → invisible γ decays. The former result is based on a data sample of . fb − recorded atthe Υ S resonance while the later results are obtained from a fb − of data sample collectedat Υ S resonance with the Belle detector at the KEKB ee − collider. We observe no significantsignal for the decays and set upper limit on their branching fractions at 90% confidence level of B B s → η η < . × − , B B → invisible < . × − and B B → invisible γ < . × − . On behalf of the Belle Collaboration ∗ Speaker © Copyright owned by the author(s) under the terms of the Creative CommonsAttribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0). https://pos.sissa.it/ a r X i v : . [ h e p - e x ] F e b tudy of B and B s Decays at Belle
N. K. Nisar
1. Introduction
In the Standard Model (SM), B s → η η decay proceeds via tree-level b → u and penguin b → s transitions. Penguin transitions are sensitive to Beyond-the-Standard-Model (BSM) physicsscenarios and could affect its branching fraction and CP asymmetry [1]. Once the branchingfractions for two-body decays B s,d → ηη, ηη , η η are measured, and the theoretical uncertaintiesare reduced, it would be possible to extract CP violating parameters from the data using theformalism based on SU(3)/U(3) symmetry [2]. The formalism requires at least four of these sixbranching fractions and the result on B s → η η is a potential input. The predicted branchingfractions of the decays B → invisible and B → invisible γ , where “invisible” defined as particlesthat leave no signal in the Belle detector, could be as high as − − − in the New Physics(NP) models [3, 4]. Decays with similar signature such as B → γν ¯ ν and B → ν ¯ νν ¯ ν arehighly suppressed in the SM [5–7]. A very low background from the SM indicates that a signal of B → invisible γ in the current B-factory data would indicate NP.
2. Belle detector
The Belle detector [8] was a large-solid-angle magnetic spectrometer that operated at the KEKBasymmetric-energy ee − collider [9]. The detector components include a tracking system comprisinga silicon vertex detector (SVD) and a central drift chamber (CDC), a particle identification (PID)system that consists of a barrel-like arrangement of time-of-flight scintillation counters (TOF) and anarray of aerogel threshold Cherenkov counters (ACC), and a CsI(Tl) crystal-based electromagneticcalorimeter (ECL). All these components are located inside a superconducting solenoid coil thatprovides a 1.5 T magnetic field. Outside the coil, the K L and muon detector (KLM) is instrumentedto detect K L mesons and to identify muons.
3. Search for the Decay B s → η η In this paper we report the preliminary result of the first search for the decay B s → η η usingthe full Belle data sample of . fb − collected at the Υ S resonance. The Υ S decays into B ∗ s ¯ B ∗ s , B ∗ s ¯ B s or B s ¯ B ∗ s , and B s ¯ B s pairs followed by the decays of the excited vector states to B s , by emitting a photon. Our data sample contains . ± . × B ∗ s ¯ B ∗ s pairs [10]. A setof Monte Carlo (MC) simulated events are used for the selection optimization and estimation ofreconstruction efficiency.We reconstruct η candidates using pairs of photons of energy that exceeds 50 (100) MeVin the barrel (end-cap) region of the ECL and requiring the invariant mass to be in the range ≤ M γγ ≤
580 MeVc . Candidates for the decay η → ππ − η are reconstructed using pairs ofoppositely-charged pions and η . We require the reconstructed η invariant mass to be in the range ≤ M ππ − η ≤
980 MeVc . To identify B s → η η candidates we use beam-energy constrained B s mass, M bc = √︃ E − p B s , the energy difference, Δ E = E B s − E beam , and the reconstructedinvariant mass of the η , where E beam , p B s and E B s are the beam energy, the momentum magnitudeand the reconstructed energy of B s candidate, respectively.The primary source of background are ee − → q ¯ q ( q = u, d, c, s ) continuum events. Becauseof large initial momenta of the light quarks, continuum events exhibit a “jet-like” event shape, while2 tudy of B and B s Decays at Belle
N. K. Nisar ] [GeV/c bc M5.3 5.35 5.4 5.45 ) E v en t s / ( . G e V / c D - - E v en t s / ( . G e V ) ) [GeV/c h - p + p M(0.92 0.94 0.96 0.98 ) E v en t s / ( . G e V / c Figure 1:
Signal-region projections of 3D fit to B s → η η data. Points with error bars represent data, bluesolid curves show the resulting fit-projection, while the red dash-dotted and blue dash-dotted curves showthe signal and background components. B ∗ s ¯ B ∗ s events are distributed isotropically. We use modified Fox-Wolfram moments [11], whichdescribe the topology of the event, to discriminate between signal and continuum background.To extract the signal yield, we perform an unbinned extended maximum likelihood fit to thethree-dimensional (3D) distribution of M bc , Δ E , and M ππ − η . MC sample is used to determinesignal and background probability density functions (PDF). We use B → η K S data recorded atthe Υ S resonance to adjust the PDF shape parameters used to describe the signal.To test and validate our fitting model, ensemble tests are performed by generating MC pseudo-experiments using PDFs obtained from the simulation and the B → η K S data. We use the resultsof pseudoexperiments to construct classical confidence intervals using Neyman construction [12].These confidence intervals are then used to prepare a classical confidence belt [13] and used tomake a statistical interpretation of the results obtained from fit to data.We obtain . ± . signal and . ± . background events from the 3D fit to data. Weshow the signal-region projections of the fit in Fig. 1. We observe no signal and estimate the90% confidence-level (CL) upper limit on the branching fraction of the decay B s → η η using thefrequentist approach [12] and the following formula: B B s → η η < N UL · N B ∗ s ¯ B ∗ s · ε · B dp , (1)where N B ∗ s ¯ B ∗ s is the number of B ∗ s ¯ B ∗ s pairs in the full Belle data sample, ε is the overallreconstruction efficiency for the signal B s decay, and B dp is the product of the secondary branchingfractions for all daughter particles in our final state. Further, N UL is the expected signal yield at90% CL obtained from the confidence belt, which is approximately 6 events. Using Eq. (1) weobtain a 90% CL upper limit on the branching fraction of B B s → η η < . × − . The totalsystematic uncertainty on the upper limit is estimated to be 17%.3 tudy of B and B s Decays at Belle
N. K. Nisar
Figure 2:
Projections of the fit result on cos θ T (left) and E ECL (right) for B → invisible. Points with errorbars are data, black solid line is the fit result, red dotted line is the signal component, green short-dashed lineis the generic B background component and blue dash-dotted line is the non- B background component.
4. Search for B decays to invisible final states γ These searches are based on a data sample containing × B ¯ B pairs accumulated at the Υ S resonance, corresponding to an integrated luminosity of fb − . Ten million MC simulatedevents for B → ν ¯ ν and B → ν ¯ νγ decays are generated and used to determine signal efficiencyand optimize signal event selection.Since the signal side particle, except photon, cannot be detected, the other B meson in the event( B tag ) is reconstructed. Then the signal is searched in the remaining part of the event. B tag mesonsare reconstructed from 494 hadronic decay modes by assigning signal probability to reconstructedparticles using a neural network (NN) package [14]. After reconstruction of B tag , no extra particlesbut photons are expected in the event. Thus events with extra tracks, π s, or K L s are rejected.The sum of all remaining energies of ECL clusters that are not associated with B tag daughtersand signal photons in case of B → invisible γ , denoted by E ECL , is a strong variable to identifysignal events. Since the distribution for signal events peaks at zero, the E ECL signal box is definedas E ECL < . GeV and the sideband is defined as . GeV < E
ECL < . GeV. Continuum eventsare the dominant source of background (Non- B ) followed by B ¯ B decay with a b → c transition(Generic B ). Two NNs are implemented to suppress these backgrounds.A two dimensional (2D) extended unbinned maximum likelihood fit is applied with fittingvariables E ECL and cos θ T to extract signal yield for the decay B → invisible. Here cos θ T is thecosine of the angle between the two thrust axes in the ee − c.m. frame. The two thrust axes aredefined as the directions that maximizes the longitudinal momenta of B tag daughters and particlesin the remaining part of the event. All PDFs are obtained from signal MC and off-resonance data.The projections of the 2D fit results are shown in Fig. 2 and the corresponding fitting yiels for eachcomponent are listed in Table. 1. No significant signal is observed and a 90% CL upper limit on thebranching fraction is estimated to be B B → invisible < . × − [15]. Systematic uncertaintyis estimated to be . using control samples B , ± → B ∗ , ± lν . B → invisible γ decays are searched by counting events in E ECL signal box in the bins ofsquared missing mass defined as: 4 tudy of B and B s Decays at Belle
N. K. NisarComponent YieldsSignal . . − . Generic B . . − . Non- B − . . − . Table 1:
Fitting yield ( B → invisible). N databkg,box N databox B → invisible γ . ± . M miss < c . ± . c < M miss <
10 GeV c . ± .
10 GeV c < M miss <
15 GeV c . ± .
15 GeV c < M miss <
20 GeV c . ± .
20 GeV c < M miss . ± . Table 2:
Estimated number of background events in the signal box ( N databkg,box ) and the number of events inthe signal box ( N databox ) for B → invisible γ and M miss bins. M miss = P beam − P B tag − P γ c , (2)where P beam , P B tag and P γ are four-momenta of ee − system, the B tag and the signal photon. Thenumber of background events in the E ECL signal box is estimated from the data sideband bymultiplying the fraction of events in signal box to the sideband, estimated in the MC. The countingresults in E ECL signal box and in bins of M miss are summarized in Table. 2. The observed numberof events is consistant with no signal. We set a 90% CL upperlimit on the branching fraction B B → invisible γ < . × − [15] with an associated systematic uncertainty of . .
5. Conclusions
In summary, we have used the full data sample recorded by the Belle experiment at Υ S and Υ S resonances to search for the decays B s → η η and B → invisible γ and no evidence is found.We set world’s first upper limits on the branching fraction of B s → η η and improved the existingupper limit on B → invisible γ . References [1] E. Kou et al. (Belle II Collaboration), Prog Theor Exp Phys , 123C01 (2019).[2] Y.-K. Hsiao, C.-F. Chang, and X.-G. He, Phys. Rev. D , 114002 (2016).[3] A. Dedes, H. Dreiner, and P. Richardson, Phys. Rev. D , 015001 (2001).[4] A. Badin and A. A. Petrov, Phys. Rev. D , 034005 (2010).[5] G. Buchalla and A. J. Buras, Nucl. Phys. B400 , 225 (1993).5 tudy of B and B s Decays at Belle
N. K. Nisar[6] B. Bhattacharya, C. M. Grant, and A. A. Petrov, Phys. Rev. D , 093010 (2019).[7] C. D. Lu and D. X. Zhang, Phys. Lett. B , 348 (1996).[8] A. Abashian et al. (Belle Collaboration), Nucl. Instrum. Methods Phys. Res. Sect. A , 117(2002); also see Section 2 in J. Brodzicka et al. , Prog. Theor. Exp. Phys. , 04D001 (2012).[9] S. Kurokawa and E. Kikutani, Nucl. Instrum. Methods Phys. Res. Sect. A , 1 (2003), andother papers included in this Volume; T. Abe et al. , Prog. Theor. Exp. Phys. , 03A001(2013) and references therein.[10] C. Oswald et al. (Belle Collaboration), Phys. Rev. D , 072013 (2015).[11] The Fox-Wolfram moments were introduced in G. C. Fox and S. Wolfram, Phys. Rev. Lett. , 1581 (1978). The Fisher discriminant used by Belle, based on modified Fox-Wolframmoments, is described in K. Abe et al. (Belle Collaboration), Phys. Rev. Lett. , 101801(2001) and K. Abe et al. (Belle Collabboration.), Phys. Lett. B , 151 (2001).[12] J. Neyman, Phil. Trans. Roy. Soc. Lond. A236 , 767, 333 (1937); Reprinted in
A Selection ofEarly Statistical Papers of J. Neyman , (University of California Press, Berkeley, 1967).[13] A. Stuart and J.K. Ord,
Classical Inference and Relationship , 5th ed., Kendall’s AdvancedTheory of Statistics, Vol. 2 (Oxford University Press, New York, 1991); see also earlier edi-tions by Kendall and Stuart.W.T. Eadie, D. Drijard, F.E. James, M. Roos, and B. Sadoulet,
Statistical Methods in Experi-mental Physics , (NorthHolland, Amsterdam, 1971).[14] M. Feindt, F. Keller, M. Kreps, T. Kuhr, S. Neubauer, D. Zander, and A. Zupanc, Nucl.Instrum. Methods Phys. Res., Sect. A , 432 (2011).[15] Y. Ku et al. (Belle Collaboration), Phys. Rev. D102