MEASUREMENTS OF RARE B MESON DECAYS AT BELLE

Transcription

1 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 MEASUREMENTS OF RARE B MESON DECAYS AT BELLE Brendan C. K. Casey Brown University Providence, RI 296 Representing the Belle Collaboration ABSTRACT We report recent measurements of rare B meson decays by the Belle collaboration including branching fractions and time-independent partial-rate asymmetries. Emphasis is placed on the penguin mediated b s transition, particularly, the recent discovery of the leptonic penguin decays, b sl + l, in both semi-inclusive and exclusive channels. Comparisons are made between experimental results and standard model predictions for several penguin mediated modes. We also report new measurements of B + ωl + ν and B D s π + that are important for determining the CKM matrix element V ub. TTH7 627

2 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 1 Introduction At lower energies, such as the b mass scale, contributions to measurable processes from new phenomena are expected to be small. We would probably only see these effects in rare decays where the magnitude of new physics contributions may be comparable to the Standard Model(SM) contribution. We can see new physics in these modes if the deviations from predictions are larger than both our measurement error and the error in SM calculations. Most B meson decays involving the penguin mediated b s amplitude can be measured to high precision at B-factories and, in certain cases, their rates are calculated including next-to-next-to leading order corrections. 1 This allows these modes to be used as ideal tools for limiting, or perhaps finding, physics beyond the Standard Model. We investigate these in three categories: radiative b sγ penguins, leptonic b sl + l penguins, and hadronic b sq q penguins, focusing on time-independent measurements. Another presentation at this conference focuses on time-dependent measurements. 2 Rare decays mediated by the b u process can be used to measure the CKM matrix element V ub. 3 Here we will give an overview of two new measurements, B + ωl + ν and B Ds π +, that will be useful cross checks to more standard methods of determining V ub. We also present the first observation of the decay B Ds K +. 2 KEKB and Belle The analysis is based on data taken by the Belle detector 4 at the KEKB e + e storage ring. 5 The data set consists of 78.1 fb 1 on the Υ(4S) resonance corresponding to 85. ±.5 million B B events. An off-resonance data set of 8.8 fb 1 was taken 6 MeV below the Υ(4S) resonance to perform systematic studies of the continuum e + e q q background where q is either a u, d, s, or c quark. Most results pressnted here have only been analyzed using a subset of this data. KEKB collides 8 GeV electrons and 3.5 GeV positrons that are stored in separate rings, producing an Υ(4S) system that is boosted by γβ =.425 along the beam axis. KEKB has achieved grater than cm/s 2 instantaneous luminosity, the highest luminosity ever achieved. The Belle detector is a general purpose magnetic spectrometer with a 1.5 T axial magnetic field. Charged tracks are reconstructed using a 5 layer Central Drift Chamber (CDC) and a 3 layer double sided SiliconVertex Detector (SVD). Candidate electrons and TTH7 628

3 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 photons are identified using an 8736 crystal CsI(Tl) calorimeter (ECL) inside the magnet. Muon and KL candidates are identified using resistive plate chambers embedded in the iron magnetic flux return (KLM). Hadron and auxiliary lepton identification is provided by an array of 1188 Silica aerogel Čherenkov threshold counters (ACC) and a barrel of 128 time-of-flight (TOF) plastic scintillator modules. 2.1 Particle Identification and Continuum Suppression Charged π and K mesons are identified by their energy loss (de/dx) in the CDC, their Čerenkov light yield in the ACC, and for particles with momenta below 1.5 GeV, their TOF information. For each hypothesis (K or π), the de/dx, ACC, and TOF probability density functions are combined to form likelihoods, L K and L π. K and π mesons are distinguished by a cut on the likelihood ratio L K /(L K + L π ). A similar likelihood ratio including calorimeter information, such as shower shape and calorimeter energy versus drift chamber momentum, is used to identify electrons. Muons are identified based on their penetration depth into the KLM and the goodness of the match between the KLM hits and the CDC track. Each analysis makes a separate cut on the likelihood ratio depending on the particular backgrounds specific to that analysis. The efficiencies for all species are usually above 9%. The hadron misidentification rate (K as π or π as K) is below 1%. The fake rates for muons are below 2% and the fake rates for electrons are below.5%. Continuum background is reduced using event shape variables. Most analyses quantify the event topology with modified Fox-Wolfram moments 6 defined as h so l = i,j p i p j P l (cos θ ij ), h oo l = j,k p j p k P l (cos θ jk ), where i enumerates B signal candidate particles (s particles) and j and k enumerate the remaining particles in the event (o particles); p i is the ith particle s momentum, and P l (cos θ ij ) is the lth Legendre polynomial of the angle θ ij between particles i and j. The h so l terms contain information on the correlation between the B candidate direction and the direction of the rest of the event. The odd h oo l terms partially reconstruct the kinematics of the other B in the event while the even terms quantify the sphericity of the other side of the event. We create a six-variable Fisher discriminant called the Super TTH7 629

4 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 Super Fox Wolfram B Flight Direction SFW cosθ B Likelihood Ratio: SFW cosθ B R Fig. 1. Continuum suppression variables: SFW (top left), cos θ B (top right), and the combined likelihood ratio (bottom). The solid curves are the signal PDFs derived from MC. The dashed curves are the continuum PDFs derived from sideband data. The open points are the B + D π +, D K π + data sample. The solid points are off-resonance data. TTH7 63

5 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 B K + π - before R cut B K + π - after R cut Events / 2.5 MeV Events / 2.5 MeV m bc (GeV) m bc (GeV) Fig. 2. The beam constrained mass distribution for the B K + π data sample before and after requiring the likelihood ratio cut LR >.8. Fox-Wolfram defined as SF W = l=2,4 where α l and β l are the Fisher coefficients. ( h so ) α l l + h so l=1 4 ( h oo ) β l l, h oo The SF W variable can be combined with the B flight direction with respect to the beam axis, cos θ B, or other suppression variables, to form a single likelihood L B B = L(SF W ) B B L(cos θ B ) B B... for signal and an equivalent product for continuum, L q q. Continuum background is suppressed by cutting on the likelihood ratio LR = L B B L B B + L q q. These variables are shown in Fig. 1. The signal probability density functions (PDF) are derived from Monte Carlo (MC); the continuum PDFs are taken from sideband data discussed below. The SF W PDFs are modeled as the sum of a simple Gaussian and an asymmetric Gaussian 7 for both signal and continuum; the cos θ B PDF is modeled as a second order polynomial for signal and is flat for continuum. We make separate requirements on LR for each mode depending on the expected background determined using sideband data. As an example, Fig. 2 shows the B K + π data sample before and after imposing the LR >.8 requirement. TTH7 631

6 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 3 Penguins The flavor changing neutral current process b s can not occur via a first order, or tree, diagram in the SM. This process does occur as a b tw transition followed by a t + W s transition as shown in Fig. 3. This is the so-called penguin transition. Since the W and t are virtual particles, we can replace either or both of them with new physics particles. One typically chooses SUSY partners and replaces the W with a charged Higgs or replaces both the W and t with a chargino and stop leading to new physics penguins. Although these processes do not take place at first order in the SM, there are scenarios where they can be first order transition leading to new physics trees. t b s W Standard Model Penguin b ~ t s χ New Physics Penguin µ e b s New Physics Tree Fig. 3. Penguin diagrams A penguin diagram is usually completed by radiating either a photon, gluon, or Z off of one of the particles in the loop. However, as experimentalists, its easier to classify these in terms of the final state. I refer to b sγ as radiative, b sq q as hadronic, and b sl + l as leptonic penguins. In general, the hadronic penguins include both the gluon and Z radiation, and the leptonic penguins include both Z and photon radiation as well as higher order box diagrams. There is a hierarchy in rates with radiative penguins dominating and leptonic penguins trailing. Typical branching fractions are TTH7 632

7 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 B radiative b sγ hadronic b sq q leptonic b sl + l Here, the name of the game is searching for new physics by comparing measured rates to SM predictions and looking for discrepancies. This can be done in two ways. First, hadronic uncertainties are canceled by integrating over the hadronic part of the final state. This leads to high precision calculations that can be compared to inclusive rate measurements. Second, one can search for asymmetries due to interference in modes that contain only one SM penguin amplitude. In this case, the first order SM expectation is zero asymmetry and higher order corrections usually lead to small asymmetries. Therefore, measurements of relatively large asymmetries in these modes is evidence for new physics. 3.1 Radiative Penguins The semi-inclusive B X s γ signal was first published by CLEO in and later confirmed by ALEPH 9 and Belle 1 publications in 1998 and 21 respectively. The Belle measurement selects a photon with energy ranging from 2.1 GeV to 2.9 GeV and combines it with a recoil mass system consisting of one neutral or charged kaon and up to four pions allowing only one π. The signal yield is extracted from the beam constrained mass distribution, m bc = Ebeam 2 p2 B, with background parameters determined from the SF W sideband. Belle measures B(B X s γ) = (3.36 ±.53 ± ) 1 4. The beam constrained mass, photon energy, and recoil mass spectra are shown in Fig. 4. The word average of CLEO, ALEPH, and Belle is (3.22 ±.4) 1 4 which can be compared to the next-to-leading order SM prediction of (3.35 ±.3) 1 4. Thus, for the inclusive mode, we have reached the current sensitivity to test for new physics and we see no deviation from the SM within the errors. BaBar has also released preliminary measurements of this rate based on an inclusive method that also agrees with the previous measurements. 11 The beauty of this is that all four experiments apply very different analysis techniques with different background treatments and arrive at the same result. To improve on the sensitivity for new physics for inclusive radiative penguin modes, one must have a better understanding of the recoil mass spectrum, among several other TTH7 633

8 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 events/(5mev/c 2 ) (a) (b) beam constrained mass (GeV/c 2 ) Events/1MeV m b (GeV/c 2 ) Photon energy (GeV) Events/(1MeV/c 2 ) Recoil mass (GeV/c ) TTH7 Fig. 4. Top: The beam constrained B mass distribution (a) compared with the total q q background and b c (open histogram); (b) after background subtraction compared with the signal MC expectation (hatched histogram). Middle: The photon energy spectrum, background subtracted and corrected for the cut-off on the recoil mass. The data points are compared with signal MC expectations for three different values of the b quark mass. Bottom: The recoil mass distribution after background subtraction compared with the signal MC expectation (hatched histogram). 634

9 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 things. Belle has embarked on a program to measure the various exclusive final states in the recoil mass spectrum. We currently have measurements of the Kπγ (Fig. 5) and Kππγ (Fig. 6) final states. 12 We see clear signals for K (892)γ, K 2 (143)γ, and inclusive K + π π + γ with the following branching fractions: final state B( 1 5 ) K (892)γ 3.91 ±.23 ±.25 K + (892)γ 4.21 ±.35 ±.3 K2 (143)γ K + π π + γ 2.4 ±.23 ±.25 The significance is not sufficient to disentangle the substructure of the K + π π + intermidiate states, however, it is clearly dominated by K π and Kρ. These final states account for (35 ± 8)% of the total b sγ spectra. Events/(5MeV/c 2 ) data total K*(141)γ qq E sideband 5 Events/(2.5MeV/c 2 ) M(Kπ)(GeV/c 2 ) data total qq Events total K 2 *(143)γ K*(141)γ M bc (GeV/c 2 ).5 1 cosθ hel Fig. 5. Top: Kπ invariant mass, bottom left: beam constrained mass, bottom right helicity distribution for B K + π γ candidates. The helicity distribution is background subtracted. TTH7 635

10 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 events/(2.5mev/c 2 ) data total qq BB b sγ events/(1mev/c 2 ) data qq BB b sγ 1 5 events/(5mev/c 2 ) M bc (GeV/c 2 ) data total Kργ qq b sγ + BB events/(1mev/c 2 ) M Kππ (GeV/c 2 ) data total K*πγ qq b sγ +BB M Kπ (GeV/c 2 ) M ππ (GeV/c 2 ) Fig. 6. Top left: beam constrained mass, top right: K + π π + invariant mass distribution, bottom left: K + π mass distribution, bottom right: π + π mass distribution for B K + π π + γ candidates. TTH7 636

11 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, Leptonic Penguins The excellent agreement between experimental measurements and SM predictions for radiative penguin decays has already been used to place strong constraints on new physics. 13,1 A provocative interpretation is that there is either more than one new amplitude and the combined effects cancel, or the new amplitude is twice as strong as the SM contribution, but in the opposite direction, leaving the square of the amplitude unchanged. These scenarios are not so far fetched if one considers the zoo of new particles expected to appear near the TeV scale. Since b sγ is a two body decay, the kinematics are fixed. Kinematic distributions such as the photon energy or recoil mass spectrum tell much more about the motion of the quarks within the B meson than they do about new physics. However, the leptonic penguins, b sl + l are three body decays. In these modes, kinematic distributions such as the dilepton invariant mass spectrum and the lepton forward-backward asymmetry with respect to the B flight direction, are sensitive to new physics. These distributions, along with the integrated rate, provide enough constraints to determine both the magnitude and phase of the new physics contributions, solving many of the ambiguities still existing in the b sγ data. Also, b sγ kinematic parameters can be used to tune the parameters of the B meson such as the effective masses and relative motion of constituents, 14 further reducing the theoretical uncertainties. At Belle, we have a program for both semi-inclusive 15 and exclusive 16 reconstruction of b sl + l. The semi-inclusive analysis follows the b sγ analysis. We require either an e + e, µ + µ, or e + µ combination with p e >.5 GeV, p µ > 1 GeV, and m(l + l ) > 2 MeV. The recoil system contains one neutral or charged kaon and up to four pions allowing one π and having total mass less than 2.1 GeV. In the instance of multiple candidates per event, the best candidate is selected based on E = E B E beam and the B flight direction. The signal characteristics are modeled using a NNLO calculation of Ref. 1 for the dilepton mass spectra. The recoil mass is modeled as the sum of exclusive kaon, K (892), and inclusive hadronization for recoil mass above 1.1 GeV based on the model in Ref. 17. The three components are normalized so that the inclusive and exclusive rates agree with the predictions of Ref. 1. The model errors are determined to be 13% for di-electron final states and 14% for di-muon final states. Backgrounds are dominated by double semi-leptonic B B events where either both B mesons decay semi-leptonicly or a b c semi-leptonic transition is followed by a TTH7 637

12 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 Arbitrary unit Visible Energy (GeV) Fig. 7. Visible energy in the event used to distinguish events with neutrinos. The solid histogram is B X s l + l signal Monte Carlo. The dashed histogram is double semi-leptonic B B events. c s semi-leptonic transition. The presence of two undetected neutrinos allows these events to be reduced using missing energy and missing mass variables as shown in Fig. 7. Another large source of background arrises from B charmonium decays where the charmonium state decays to dileptons. These are removed by vetoing regions of the dilepton mass spectra allowing generous room for the radiative tails as shown in Fig. 8. A final background arrises from B X s π + π decays where both pions are misidentified as leptons. We determine the size of this background by removing the lepton ID cuts and measuring the number of B X s π + π candidates in our data sample. These events are shown in Fig. 9 We then multiply this by the double misidentification probability to estimate the number of background events in our sample. The estimated background is 2.6 events in the µ + µ channel and.1 events in the e + e channel. The entire reconstruction procedure is tested by removing the charmonium veto and measuring the rates for B J/ψX s shown in Fig. 1. These distributions are used to determine parameters of the signal fitting functions as well as provide an overall cross check of the analysis. The resulting beam constrained mass distributions are shown in Fig. 11. We see significant signals in both the e + e and µ + µ channels, corresponding to the following branching fractions TTH7 638

13 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 Fig. 8. Dilepton invariant mass for e + e (top) and µ + µ bottom for b c cx Monte Carlo. The shaded regions are removed from the data set. 5 Data X S π + π Fig. 9. Beam constrained mass distribution for B X s π + π data events used to determine the backgrounds from double misidentification of π + π as l + l. TTH7 639

14 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 Fig. 1. Beam constrained mass distributions for B J/ψX s in the e + e channel (top) and µ + µ channel (bottom) used to calibrate the analysis procedure. TTH7 64

15 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 B( 1 6 ) (m(l + l ) >.2 GeV) X s e + e 5. ± X s µ + µ 7.9 ± X s l + l 6.1 ± These can be compared to the SM calculation of (4.2 ±.7) 1 6. The lack of signal events and goodness of fit of the background function in the e + µ channel is also an indication that we understand the combinatorial as well as B X s π + π backgrounds. The dilepton and recoil mass spectra are shown in Fig. 12. We see no indication of new phenomena at the current level of precision. 4 (a) X s e + e - (b) X s µ + µ - 3 Entries / (2.5 MeV/c 2 ) (c) X s l + l - (d) X s e + µ - + c.c M bc (GeV/c 2 ) Fig. 11. Beam constrained mass distributions for the various B X s l + l modes. The solid lines indicate the fit results and the dotted lines show the sum of the background components. In the exclusive channels we see signals in both the Ke + e and Kµ + µ modes but do not see significant signals in the K channels leading to the following branching fractions TTH7 641

16 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 3 (a) (b) 2 Entries / (.2 GeV/c 2 ) (c) (d) M ll (GeV/c 2 ) M Xs (GeV/c 2 ) Fig. 12. SM expectations for the (a) dilepton and (b) recoil mass spectra; the observed (c) dilepton and (d) recoil mass spectra (circles). The histograms in (c) and (d) show the SM expectations after all the selections are applied; histograms are normalized to the expected branching fractions. The dotted line in (d) indicates the cut on the recoil mass. TTH7 642

17 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 B( 1 7 ) Ke + e Kµ + µ ±.8 Kl + l ±.6 K l + l < ( 9% C.L.) The SM expectations are (3.5 ± 1.2) 1 7 for Kl + l and (1.4 ±.4) 1 6 for K l + l. Contrary to expectations, we see more kaons than K mesons, however the errors are still large and this would be interpreted as a breakdown in our knowledge of the form factors rather than a breakdown in the SM. 3.3 Hadronic Penguins and Direct CP Violation With the exception of completely inclusive measurements, predictions for hadronic penguin mediated processes are typically held back by large hadronic uncertainties. This is because the radiated q q pair can not be decoupled from either the s quark or the spectator quark in the B meson. However, there are several states that are supposedly dominated by a single penguin amplitude and can therefore be used to search for new physics via interference. This interference is usually manifested in a partial-rate asymmetry defined as A CP (f) = N( B f) N(B f) N( B f) + N(B f), where B represents either a B or B + meson, f represents a flavor specific final state, and B and f are their conjugates. The decay B K π is dominated by a b sd d penguin with no tree contibutions and thus one expects very small to no partial rate asymmetry in this mode. One could imagine a scenario in which the K π final state receives contributions from other Kπ final states through rescattering. In this case, we expect the asymmetry in K π to be less than or equal to the asymmetries in the other Kπ modes. Combining measurements from Belle, BaBar, and CLEO leads to a 9% confidence limit of.15 < A CP <.2 for the K π ± final state. We would not expect anything much larger than this in K π. With a 29.1 fb 1 data set, Belle reported a partial-rate asymmetry of 18 A CP =.46 ±.15 ±.2 (29.1 fb 1 ). TTH7 643

18 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 K π - K π + Events / 2 MeV 49 ± 8 evts. Events / 2 MeV 18 ± 6 evts. E (GeV) E (GeV) K S π - 5 K S π + Events / 2 MeV Events / 2 MeV E [GeV] E [GeV] Fig. 13. E distributions for B KSπ for B (left) and B (right) candidates. The top row is the 29.1 fb 1. data sample; the middle row is the 78.1 fb 1 data sample. The sum of the signal and background functions is shown as a solid curve. The dashed curve represents the signal component, the dotted curve represents the continuum background, and the hatched histogram represents the charmless B background component. The bottom distributions shows the asymmetry as a function of time. TTH7 644

19 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 The E distribution for this surprising result is shown in Fig. 13. We have now applied the analysis to the full data set. The resulting asymmetry is A CP =.2 ±.9 ±.1 (78.1 fb 1 ). The updated E distribution along with the asymmetry as a function of run period is shown in Fig. 13. Although we can not say anything about new phenomena in this mode, we did learn a lesson in statistics. I have received a poignant comment on this from a senior summer school participant which I put forward as a corollary to the laws of statistics: A three sigma effect has a 5/5 chance of going away. Two other interesting pure penguins are the B φk and B K γ modes. 19 Although K γ is a radiative penguin, the exclusive mode is grouped in this category because it shares the hadronic uncertainties with its hadronic cousins through the B K form factor. The beam constrained mass distributions for B K γ are shown in Fig. 14. The partial-rate asymmetries are A CP ( 9% C.L.) K γ.22 ±.48 ± < A <.62 φk.7 ±.9 ±.5.18 < A <.16 We see no evidence for non-zero asymmetries, however, these measurements constitute the strongest bounds on DCPV in pure penguin modes. 4 New Results on V ub Two of the most important measurements expected from B-factories are turning out to be two of the hardest measurements to make. They are measuring the magnitude and phase of the CKM matrix element V ub. Here, we focus on the magnitude V ub. There are three rates proportional to V ub 2 illustrated in Fig. 15. The purely leptonic decays B lν are not yet measured although we expect to reach the sensitivity for B τν relatively soon. The semi-leptonic B Xlν is the classic approach and follows closely to successful extractions of V cb. However, there are large backgrounds from related B TTH7 645

20 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 (a) K - π + γ (b) K + π γ 4 Yield 165 ± 14 ± 2 Yield 186 ± 15 ± 3 4 fit result total b.g. fit result total b.g. 2 2 Entries/(MeV/c 2 ) (c) K S π γ Yield 6 ± 8 ± 1 fit result total b.g. (d) K S π + γ Yield 58 ± 8 ± 1 fit result total b.g Entries/(MeV/c 2 ) 1 (e) K π γ (f) K + π γ 1 Yield 34 ± 6 ±.5 Yield 27 ± 6 ±.5 fit result total b.g. fit result total b.g Beam constrained mass (GeV/c 2 ) Fig. 14. Beam constrained mass distributions for B K γ TTH7 646

21 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 meson decays that are difficult to control and there are also theoretical uncertainties in both the exclusive and inclusive channels. There are also pure tree hadronic processes that are proportional to V ub 2. These suffer the same theory problems as exclusive semileptonic decays via form factor uncertainties and also can not be clearly classified as pure tree due to rescattering and uncertainties involving higher order diagrams. However, the lack of neutrinos in the final state makes these measurements extremely clean relative to their semi-leptonic counterparts. Since there is no golden-mode for measuring V ub, we must measure it in as many ways as possible in an attempt to understand both the theoretical and experimental systematic errors. b u b b b u W l ν l ν u s c u V ub Leptonic semileptonic hadronic Fig. 15. Processes with rates proportional to V ub Exclusive Semi-leptonic Decays Exclusive semi-leptonic analysis at Belle revolves around the neutrino reconstruction method. Taking advantage of the well defined center of mass kinematics we can define the neutrino four momentum using the total reconstructed energy and momentum in the event: E miss = E beam E rec, p miss = p beam p rec. This is then combined with a recoil system to form a B candidate. Backgrounds are reduced by making consistency cuts on the neutrino mass, the event charge, the direction TTH7 647

22 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 of the missing momentum, and that the event contains only a single lepton. BELLE Entries / (1 MeV/c 2 ) (a) Entries / (1 MeV) (b) m bc (GeV/c 2 ) E (GeV) Fig. 16. The (a) beam constrained mass and (b) E for selected B candidates. The points with error are data after subtracting continuum and fake lepton background events. The dashed histogram is B X c lν background. CLEO, BaBar, and Belle have all used this analysis where the recoil system was either a π + or a ρ meson. Belle has now extended this to the ω system. 2 The advantage being the narrow ω resonance adds an extra tool in separating out backgrounds from other charmless recoil systems. The B reconstruction variables are shown in Fig. 16. After making cuts on the B reconstruction variables, we fit the π + π π invariant mass plot shown in Fig. 17. This yields 73 ± 15 ω candidates in the B signal region. From sideband studies, we determine that 19% of these are from continuum ω production leaving 59 ± 15 B + ωl + ν events corresponding to a branching fraction of B(B + ωl + ν) = (1.4 ±.4 ±.3) 1 4. This can be compared with other Belle semi-leptonic branching fractions B(B π l + ν) = (1.35 ±.11 ±.21) 1 4, B(B + ρ e + ν) = (1.44 ±.18 ±.23) 1 4. Further studies are required to extract V ub from this mode. TTH7 648

23 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 BELLE Entries / (1 MeV/c 2 ) m(π + π π ) (GeV/c 2 ) Fig. 17. The π + π π invariant mass plot for candidates in the B signal window. 4.2 Hadronic Decays Recently, attention has been placed on extracting V ub from the hadronic decay B D s π +. The quark level diagram for this decay is shown in Fig. 18. As mentioned above, this mode has similar theoretical uncertainties to the exclusive semi-leptonic modes but the backgrounds can be easily identified and controlled since we fully reconstruct the final state. The sister decay is B D + s K. This is expected to be much smaller since it can only take place via an exchange diagram at the quark level. However, rescattering among the Dh and D s h (h = K, π) final states could enhance the rate. Backgrounds in these modes fall in three categories. Backgrounds containing real D s mesons are from continuum D s production and have no peaking structure in E. Backgrounds from real B decays into the same final state peak in E but are flat under the D s mass distribution. Finally, backgrounds from B Dπ decays can migrate into the D s mass range if one of the final state pions is misidentified as a kaon. These are explicitly removed with veto cuts after flipping the mass assignments on individual tracks. All of these backgrounds can be separated in a two-dimensional fit to the E and m(d s ) distributions. The fit contains a two-dimensional Gaussian for signal, one dimensional Gaussians in each variable to account for the backgrounds listed above, and a final flat background. The D s mass distribution is shown in Fig. 19. The E distributions is shown in Fig. 2. We find D s π candidates leading to a branching fraction of B(B D + s π ) = ( ±.7) 1 5. TTH7 649

24 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 B - b - d W u -c D- s s π+ B - b - d c - D+ s s s -u K - + D π- c- d d ū c - D+ s s s -u K - Fig. 18. Quark level diagrams for B D s π(k) decays. Events/(.5 GeV/c 2 ) 15 D s + K D + s π M(D S ) (GeV/c 2 ) Fig. 19. The D s mass spectra for D + s K (top) and D + s π (bottom) in the B signal region. TTH7 65

25 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 Events/(.1 GeV) D s + K - D s + π E (GeV) Fig. 2. The E spectra for D + s K (top) and D + s π (bottom) in the D s signal region. Surprisingly, we see a larger signal of events in the D s K channel leading to a branching fraction of B( B D s K + ) = ( ± 1.3) 1 5. As a cross check, we apply the fit in the D mass region also shown in Fig. 19. The branching fractions we obtain for D + π and D + K both agree within one sigma with the PDG values. Further information can be found in Ref. 21. A method has been proposed 22 that relates the ratio of rates of D + s π to D + s D to V ub /V cb 2. Using PDG values for the B D + s D branching fraction and V cb we obtain V ub = ( ) 1 3, consistent with other results from semi-leptonic decays. 5 Conclusions The Belle experiment is leading the way into a new era of precision SM testing and searches for new physics with leptonic penguin decays. Its interesting to note that we are now measuring significant signals with rates at the level. We are also providing the strongest limits on DCPV in pure penguin modes through the partial-rate asymmetry limits on φk and K γ. TTH7 651

26 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 We now have a new measurement of V ub using hadronic decays. This demonstrates the experimental feasibility as well as simplicity of this class of measurements. Its now up to theorists to deal with the complications arising from these modes. We have also demonstrated that the complicated B backgrounds are greatly reduced in the B ωlν channel with respect to previously measured exclusive semi-leptonic decays. I would like to thank the SSI22 organizing committee, particularly Maura Chatwell, for putting on an excellent summer school and providing outstanding hospitality. References [1] A. Ali, E. Lunghi, C. Greub, G. Hiller, hep-ph/1123; A. Ali, P. Ball, L. T. Handoko, G. Hiller, Phys. Rev. D 61, 7424 (2). [2] W. Trischuk. Talk at this conference. [3] M. Kobayashi, T. Maskawa, Prog. Theor. Phys. 49, 652 (1973). [4] Belle Collaboration, K. Abe et al., Nucl. Inst. Meth. A 479, 117 (22). [5] E. Kikutani ed. KEKBAccelerator Papers, KEK Preprint , to be published in Nucl. Inst. Meth. A. [6] G. Fox, S. Wolfram, Phys. Rev. Lett. 41, 1581 (1978). [7] A Gaussian distribution with a different σ above and below the mean of the Gaussian. [8] CLEO Collaboration, M. Alam et al., Phys. Rev. Lett. 74, 2885 (1995). [9] ALEPH Collaboration, R. Barate et al., Phys. Lett. B 429, 169 (1998). [1] Belle Collaboration, K. Abe et al., Phys. Lett. B 511, 151 (21). [11] P. Bloom. Talk at this conference. [12] S. Nishida, M.Nakao et al. Belle preprint 22-11, hep-ex/2525, to appear in Phys. Rev. Lett. [13] A. L. Kagen and M. Neubert, Eur. Phys. J. C 7, 5 (1999). [14] CLEO Collaboration S. Chen et al., Phys. Rev. Lett. 87, 25187(21), D. Cronin- Hennessy et al., Phys. Rev. Lett. 87, 25188(21). [15] J. Kaneko et al. Belle preprint 22-28, hep-ex/2829, submitted to Phys. Rev. Lett. TTH7 652

27 XXX SLAC Summer Institute (SSI22), Stanford, CA, 5-16 August, 22 [16] K. Abe et al. Phys. Rev. Lett. 88, 2181 (22); A. Ishikawa et al., Belle-Conf [17] A. Ali, G. Hiller, T. Handoko, T. Morozumi, Phys. Rev. D 55, 415 (1997). [18] B. C. K. Casey et al. Belle preprint 22-24, hep-ex/279, to appear in Phys. Rev. D [19] M. Nakao et al., Belle-Conf [2] C. Schwanda et al., Belle-Conf [21] P. Krokovney et al. Belle preprint 22-25, hep-ex/2741, submitted to Phys. Rev. Lett. [22] C. S. Kim, Y. Kwon, J. Lee, W. Namgung, Phys. Rev. D (21). TTH7 653