EEarly charmless B physics at Belle II Eldar Ganiev , β University and INFN, Trieste, Italy
E-mail: [email protected]
We report on the ο¬rst measurements of branching fractions, CP-violating charge asymmetries,and longitudinal polarization fractions in charmless π΅ decays at the Belle II experiment. We usea sample of electron-positron collisions collected in 2019 and 2020 at the Ξ₯ ( π ) resonance thatcorresponds to 34.6 fb β of integrated luminosity. The results are compatible with the knownvalues, which indicates a good understanding of detector performance. On behalf of the Belle II collaboration β Speaker Β© 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 ] J a n arly charmless B physics at Belle II Eldar Ganiev
1. Introduction
The physics of charmless π΅ decays is an essential portion of the Belle II program. The expectedlarge yields will enable signiο¬cant advancements in the understanding of quark dynamics, includingan improved determination of the CKM phase πΌ / π [1], a conclusive test of the πΎ π isospin sum-rule [1, 2], a thorough investigation of CP-violating asymmetries localized in the phase space ofthree-body π΅ decays [1], and measurements of the decay-time-dependent CP violation in almostpure penguin π β πππ channels, such as π΅ β ππΎ and π΅ β π (cid:48) πΎ decays [1].Belle II [1] is a magnetic spectrometer, designed to reconstruct the products of electron-positroncollisions produced by the SuperKEKB asymmetric-energy collider, located at the KEK laboratory,Japan. The Belle II detector started its collision operations on March 11, 2019. The sample ofelectron-positron collisions used in this work corresponds to an integrated luminosity of 34.6 fb β [3]and was collected at the Ξ₯ ( π ) resonance as of May 14, 2020. We aim to perform ο¬rst measurementsin charmless π΅ decay modes, which oο¬er ideal benchmarks to test tracking, reconstruction of neutralparticles, particle identiο¬cation, vertexing, and advanced analysis techniques capabilities.We focus on decays with branching fractions of 10 β , or larger, into ο¬nal states suο¬cientlysimple to obtain visible signals in the analyzed data set with a relatively straightforward recon-struction. The target decay modes are two-body π΅ β πΎ + π β , π΅ + β πΎ + π , π΅ + β πΎ π + , π΅ β πΎ π , π΅ β π + π β , π΅ + β π + π decays; three-body π΅ + β πΎ + πΎ β πΎ + , π΅ + β πΎ + π β π + decays, and quasi two-body π΅ + β π ( ) πΎ + , π΅ β π ( ) πΎ , π΅ + β π ( ) πΎ β ( ) + , π΅ β π ( ) πΎ β ( ) decays. In what follows, charge-conjugate modes are implied, and πΎ β+ , and π indicate the πΎ β ( ) + , and π ( ) mesons, except when otherwise stated.The principal challenge is to overcome the initial (cid:46) β signal-to-background ratio with aselection suο¬ciently discriminating to isolate an abundant signal. The dominant backgroundsarise from random combinations of particles produced in ππππ‘πππ’π’π π + π β β π Β― π ( π = π’, π, π , π )events. We use two variables known to be strongly discriminating between signal and continuum:beam-energy-constrained mass and energy diο¬erence, deο¬ned as π bc β‘ βοΈ π /( π ) β ( π β π΅ / π ) and Ξ πΈ β‘ πΈ β π΅ β β π /
2, where β π / πΈ β π΅ and π β π΅ are the reconstructedenergy and momentum of π΅ meson candidates, all in the Ξ₯ ( π ) frame. For further discrimination,we use a binary boosted decision-tree classiο¬er that makes a non-linear combination of about30 kinematic, decay time, and event-shape variables. We train the classiο¬er to identify statisticallysigniο¬cant signal and background features using unbiased simulated samples.
2. Two- and three-body decays
We search for the decays π΅ β πΎ + π β , π΅ + β πΎ + π , π΅ + β πΎ π + , π΅ β πΎ π , π΅ β π + π β , π΅ + β π + π , π΅ + β πΎ + πΎ β πΎ + , and π΅ + β πΎ + π β π + [4]. We form ο¬nal-state particle candidates byapplying baseline selection criteria and then combine candidates in kinematic ο¬ts consistent withthe topologies of the desired decays to reconstruct π΅ candidates.In the simulated samples, for each channel, we simultaneously vary the selection criteria oncontinuum-suppression output and charged-particle identiο¬cation information to maximize S/ β S+B,where S and B are signal and background yields, respectively, estimated in the signal region. The π selection is optimized using the π΅ + β π· (β πΎ + π β π ) π + control channel reconstructed in2 arly charmless B physics at Belle II Eldar Ganievsimulated and collision data. After applying the optimized selection, we restrict samples to onecandidate per event, choosing it randomly.Some channels show a non-negligible fraction of self-cross-feed , i.e. misreconstructed signalcandidates formed by misidentiο¬ed (swapped mass assignments) signal particles or combinationsof signal and non-signal particles. We include the self-cross-feed component in our ο¬t model byο¬xing its proportions to the expectations from simulation.For the three-body decays, simulation is also used to identify and suppress contamination frompeaking backgrounds. We exclude the two-body invariant mass ranges corresponding to π· , π π ,and π π decays for π΅ + β πΎ + πΎ β πΎ + and π· , π π , π π , π½ / π , and π ( π ) decays for π΅ + β πΎ + π β π + decays. In addition, we veto the genuine charmless π΅ + β πΎ β ( ) π + subcomponent to allow for aconsistent comparison with the known value [5].We determine signal yields from maximum likelihood ο¬ts of the unbinned Ξ πΈ distributions ofcandidates restricted to the π bc > 5.27 GeV/ π and | Ξ πΈ | < 0.15 GeV region. Fit models are obtainedempirically from simulation, with the only additional ο¬exibility of a shift of the signal-peak position,which is determined in data. Examples of the Ξ πΈ distributions with ο¬t projections overlaid areshown in Figure 1. - - - [GeV] E D C and i da t e s pe r . G e V Belle II (preliminary) -1 L dt = 34.6 fb (cid:242)
DataTotal fit + c.c. - p + K ο¬ B + c.c. - p + p ο¬ B BackgroundSXF - - - [GeV] E D C and i da t e s pe r . G e V Belle II (preliminary) -1 = 34.6 fb t d L (cid:242) DataTotal fit + c.c. p + K ο¬ + B + c.c. p + p ο¬ + B Background - - - [GeV] E D C and i da t e s pe r . G e V DataTotal fit + c.c. p S0 K ο¬ B Rare backgroundContinuum background
Belle II (preliminary) -1 = 34.6 fb t d L (cid:242) - - - [GeV] E D C and i da t e s pe r . G e V Belle II (preliminary) -1 = 34.6 fb t d L (cid:242) DataTotal fit + c.c. + p - p + K ο¬ + B + c.c. + p - K + K ο¬ + B + c.c. + p - p + p ο¬ + B pp K ο¬ h c X ο¬ B Background
Figure 1:
Distributions of Ξ πΈ for (top left) π΅ β πΎ + π β , (top right) π΅ + β πΎ + π , (bottom left) π΅ β πΎ π ,and (bottom right) π΅ + β πΎ + π β π + candidates reconstructed in 2019β2020 Belle II data. The "SXF" labelindicates self-cross-feed. The projections of unbinned maximum likelihood ο¬ts are overlaid. We determine each branching fraction as B = π /( ππ π΅π΅ ) , where π is the signal yieldobtained from the ο¬t, π is the reconstruction and selection eο¬ciency, and π π΅π΅ is the number ofproduced π΅ Β― π΅ pairs, corresponding to 19.7 million for π΅ + π΅ β and 18.7 million for π΅ π΅ pairs. Theselection eο¬ciencies are determined from simulation. For those eο¬ciencies, where simulation maynot accurately model data, we perform dedicated checks on control samples and assess systematicuncertainties. We obtain the number of π΅ Β― π΅ pairs from the measured integrated luminosity, the π + π β β Ξ₯ ( π ) cross section (assuming that the Ξ₯ ( π ) decays exclusively to π΅ Β― π΅ pairs), and the3 arly charmless B physics at Belle II Eldar Ganiev Ξ₯ ( π ) β π΅ Β― π΅ branching fraction [6]. For the branching fraction measurement of π΅ + β πΎ π + and π΅ β πΎ π , we consider a 0.5 factor to account for the πΎ β πΎ probability.We also measure CP-violating asymmetries A CP = A β A det , where A is the observedcharge-speciο¬c signal-yield asymmetry, and A det is the instrumental asymmetry due to diο¬erencesin interaction or reconstruction probabilities between opposite-charge hadrons. We determine A from a simultaneous non-extended likelihood ο¬t of the unbinned Ξ πΈ distributions of bottomand antibottom candidates decaying in ο¬avor-speciο¬c ο¬nal states. We evaluate the instrumentalasymmetries A det ( πΎ + π β ) = β . Β± .
003 and A det ( πΎ π + ) = β . Β± .
022 by measuring thecharge-asymmetry in abundant samples of π· β πΎ β π + and π· + β πΎ π + decays, respectively,assuming no CP violation. Moreover, we estimate the instrumental asymmetry related to chargedkaon reconstruction alone A det ( πΎ + ) = β . Β± .
022 by combining all inputs in the relationship A det ( πΎ + ) = A det ( πΎ + π β ) β A det ( πΎ π + ) + A det ( πΎ ) , where A det ( πΎ ) estimated following Ref. [7].Finally, we assess the main systematic eο¬ects, such as coming from tracking, π reconstruction,particle identiο¬cation, and shape modelling. The measured branching fractions and CP-violatingasymmetries are summarized in Table 1. The results are compatible with known values.
3. Quasi two-body decays
We search for the decays π΅ + β ππΎ + , π΅ + β ππΎ β+ , π΅ β ππΎ , and π΅ β ππΎ β , followed by the π β πΎ + πΎ β , πΎ β β πΎ + π β , πΎ β+ β πΎ π + , and πΎ β π + π β decays [8]. We ο¬rst reconstruct chargedpion and kaon candidates by applying simple track-quality and particle-identiο¬cation selections. Wecombine them into intermediate-resonance candidates, which are required to meet invariant-massconditions. Then, reconstructed particles are combined into π΅ meson candidates that are required tosatisfy π bc > 5.25 GeV/ π and | Ξ πΈ | < 0.2 GeV. For each decay mode we accept one signal candidateper event retaining the π΅ candidate with the highest vertex ο¬t probability. The fraction of self-cross-feed candidates is ο¬xed to the expectations from simulation. To reduce remaining background, weο¬rst apply high-eο¬ciency continuum suppression selection. Then, for each individual ο¬nal state,we train and optimize a multivariate boosted decision tree classiο¬er.We determine signal yields with an unbinned extended maximum likelihood ο¬t. For eachcomponent, we compose the single event likelihood as the product of one-dimensional probabilitydensity functions for each of the observables. The observables in the ο¬t are π bc , Ξ πΈ , πΆ (cid:48) out (outputof the continuum suppression classiο¬er), π ( πΎ + πΎ β ) (invariant mass of the π candidate), cos π π» , π (cosine of the helicity angle of the π candidate), π ( πΎ + π ) (invariant mass of the πΎ β candidate),cos π π» ,πΎ β (cosine of the helicity angle of the πΎ β candidate). The last two observables are relevantonly for the π΅ β ππΎ β modes. Fit models are obtained from simulation. Figures 2 and 3 showexamples of data distributions with ο¬t projections overlaid.Finally, we calculate branching fractions and the fraction of longitudinally polarized events π πΏ = π πΏ / π πΏ π πΏ / π πΏ + π π / π π (if applicable), where π πΏ and π π correspond to the number of longitudinallyand transversely polarized signal candidates, respectively, and π πΏ,π is the corresponding selectioneο¬ciency determined from simulation. The longitudinal (transverse) polarization state correspondsto both π and πΎ β having zero (opposite) spin projections along the decay axis.We evaluate systematic uncertainties, such as those coming from tracking, particle identiο¬ca-tion, and shape modelling. The ο¬nal results, summarized in Table 1, agree with known values.4 arly charmless B physics at Belle II Eldar Ganiev
Figure 2:
Distributions of discriminating observables of the π΅ β ππΎ candidates reconstructed in 2019β2020 Belle II data. The projection of unbinned maximum likelihood ο¬t is overlaid. Figure 3:
Distributions of discriminating observables of the π΅ β ππΎ β candidates reconstructed in2019β2020 Belle II data. The projection of unbinned maximum likelihood ο¬t is overlaid.
4. Summary
We report on ο¬rst measurements of branching fractions, CP-violating charge asymmetries, andlongitudinal polarization fractions in charmless π΅ decays at Belle II. We use a sample of 2019 and2020 data corresponding to 34.6 fb β of integrated luminosity. From the maximum likelihood ο¬t, wedetermine signal yields for the decay modes π΅ β πΎ + π β , π΅ + β πΎ + π , π΅ + β πΎ π + , π΅ β πΎ π , π΅ β π + π β , π΅ + β π + π , π΅ + β πΎ + πΎ β πΎ + , π΅ + β πΎ + π β π + , π΅ + β ππΎ + , π΅ + β ππΎ β+ , π΅ β ππΎ ,and π΅ β ππΎ β . The results agree with known values and show a good understanding of detectorperformance oο¬ering a reliable basis to assess projections for future reach.5 arly charmless B physics at Belle II Eldar Ganiev
Table 1:
Summary of branching fraction, CP-violating asymmetry, and longitudinal polarization fractionmeasurements, where the ο¬rst contribution to the uncertainty is statistical, the second is systematic.
Decay mode Branching fraction Γ β CP-violating asymmetry Longitudinal fraction π΅ β πΎ + π β . Β± . Β± . . Β± . Β± .
008 - π΅ + β πΎ + π . + . β . Β± . . + . β . Β± .
022 - π΅ + β πΎ π + . + . β . Β± . β . + . β . Β± .
024 - π΅ β πΎ π . + . β . Β± . π΅ β π + π β . + . β . Β± . π΅ + β π + π . Β± . Β± . β . + . β . Β± .
123 - π΅ + β πΎ + πΎ β πΎ + . Β± . Β± . β . Β± . Β± .
022 - π΅ + β πΎ + π β π + . Β± . Β± . β . Β± . Β± .
023 - π΅ + β ππΎ + . Β± . Β± . π΅ β ππΎ . Β± . Β± . π΅ + β ππΎ β+ . Β± . Β± . . Β± . Β± . π΅ β ππΎ β . Β± . Β± . . Β± . Β± . References [1] E. Kou et al. , The Belle II Physics Book,
PTEP (2019) no.12, 123C01[ hep-ex/1808.10567 ].[2] M. Gronau, A precise sum rule among four π΅ β πΎ π
CP asymmetries,
Phys. Lett. B (2005) no. 1, 82.[3] F. AbudinΓ©n et al. (Belle II Collaboration), Measurement of the integrated luminosity ofthe Phase 2 data of the Belle II experiment,
Chin. Phys. C (2020) no. 2, 021001[ hep-ex/1910.05365 ].[4] F. AbudinΓ©n et al. (Belle II Collaboration), Measurements of branching fractions and CP-violating charge asymmetries in charmless π΅ decays reconstructed in 2019β2020 Belle II data,[ hep-ex/2009.09452 ].[5] P. A. Zyla et al. (Particle Data Group), Review of Particle Physics, PTEP (2020) no.8,083C01.[6] A. J. Bevan et al. (Belle and BaBar Collaborations), The Physics of the π΅ factories, Eur. Phys. J
C74 (2014) 3026 [ hep-ex/1406.6311 ].[7] A. Davis et al. (LHCb Collaboration), Measurement of the instrumental uncertainty for πΎ β π + -pairs at LHCb in Run 2, [ LHCb-PUB-2018-004 ].[8] F. AbudinΓ©n et al. (Belle II Collaboration), Rediscovery of π΅ β ππΎ (β) decays and measure-ment of the longitudinal polarization fraction π πΏ in π΅ β ππΎ β decays using the Summer 2020Belle II data-set, [ hep-ex/2008.03873hep-ex/2008.03873