1. Results from BaBar on the Decays B  Kl + l - and B  K*l + l - FPCP-2002, U.Pennsylvania John J. Walsh INFN-Pisa

2. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 2 Outline Introduction Introduction Analysis Overview Analysis Overview Control Samples Control Samples Results Results

3. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 3 B  K (*) l + l - in the SM and Beyond Flavor changing neutral current: proceeds via loop or box diagrams  quite small SM branching ratios Flavor changing neutral current: proceeds via loop or box diagrams  quite small SM branching ratios Massive particles can contribute to the loop/box: top quark, Higgs, SUSY  sensitivity to New Physics Massive particles can contribute to the loop/box: top quark, Higgs, SUSY  sensitivity to New Physics

4. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 4 Branching Fraction Predictions in the Standard Model New Ali et al. predictions are lower! (All predictions exclude J/  contribution.)

5. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 5 Decay rate vs. q 2 in the SM and SUSY J/   (2S)K q2q2 q2q2 SM nonres SUSY models Pole from K*  even in  +  - constructive interf. destructive

6. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 6 Recent Experimental Results Belle has published a signal based on 29.1 fb -1. Our upper limit from Run 1 (based on 20.7 fb -1, submitted to PRL):

7. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 7 Cherenkov Detector (DIRC) Silicon Vertex Tracker (SVT) Instrumented Flux Return (IFR) CsI Calorimeter (EMC) Superconducting Coil (1.5T) Drift Chamber (DCH) e - (9 GeV) e + (3 GeV) BaBar Detector @ PEP II

8. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 8 B Meson Reconstruction at  (4s) Define 3 regions in  E, m ES plane: Define 3 regions in  E, m ES plane:  A – Signal region  B –Fit region  C – Large Sideband region (*)  measured in  (4S) rest frame E i  E * beam  Improve resolution B B Typical resolutions:  (m ES )  2.5 MeV  (m ES )  2.5 MeV  (  E)  25 - 40 MeV  (  E)  25 - 40 MeV full fit region is blind

9. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 9 Analysis Strategy I Lepton and kaon ID, candidates formed for the different channels: Lepton and kaon ID, candidates formed for the different channels:  B +  K + l + l -, where l is either e or   B +  K* + l + l -, where K* +  K     and K 0      B 0  K 0 l + l -  B 0  K* 0 l + l -, where K* 0  K    Apply selection to suppress backgrounds from: Apply selection to suppress backgrounds from:  Continuum events – event shape  BB events – vertexing, E miss  B  J/  (  l + l - )K decays – exclude regions in  E, m(l + l - ) plane  Peaking backgrounds (small) Selection criteria optimized on simulated or “large sideband” events. The full fit region is blind. Selection criteria optimized on simulated or “large sideband” events. The full fit region is blind. This talk: 56.4 fb -1 on peak

10. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 10 Analysis Strategy II Monte Carlo is used for signal efficiencies and to estimate contributions from the peaking backgrounds. Monte Carlo is used for signal efficiencies and to estimate contributions from the peaking backgrounds. We use control samples in the data to check the MC: We use control samples in the data to check the MC:  Decays to charmonium. Each signal-mode final state has a “signal-like” control sample that is identical except for the restricted range of q 2. (Also a serious background!)  “Large sideband” in m ES vs.  E plane: checks comb. bkgd.  K ( * ) e -   combinations: checks comb. bkgd. The signal is extracted from a 2-D fit to the m ES vs.  E plane. Both the background normalization and its shape float. The signal shapes are taken from MC + tuning from J/  samples. The signal is extracted from a 2-D fit to the m ES vs.  E plane. Both the background normalization and its shape float. The signal shapes are taken from MC + tuning from J/  samples.

11. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 11 B  J/  (  l + l - )K : Background Source Actually, this channel is “part” of the signal, with q 2 = m(  2 Actually, this channel is “part” of the signal, with q 2 = m(  2 However, we are not interested in this part of the signal and it must be removed by direct veto. However, we are not interested in this part of the signal and it must be removed by direct veto. When the leptons from J/  ->l + l - radiate or are mismeasured, the event shifts in both m(  ) and in  E. Remove these events from BG region as well: simplify fit in m ES vs.  E plane Remove these events from BG region as well: simplify fit in m ES vs.  E plane Nominal signal region

12. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 12 B  J/  (  l + l - )K : Control Sample Kinematics very similar to the signal Kinematics very similar to the signal Sample of such events can be used to verify efficiencies of essentially all selection criteria Sample of such events can be used to verify efficiencies of essentially all selection criteria Excellent agreement found in data/MC comparison Excellent agreement found in data/MC comparison Points: data Histo: MC M(l + l - ) E.g. study tails in M(l + l - ) distribution

13. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 13 J/  Sample: signal-like log L B off resonance J/  and Large Sideband Control Sample Study: B Likelihood Variable keep Large SB Sample: background- like log L B -10 4

14. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 14 Charmonium Control Samples: Yields in Data vs. Simulation

15. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 15 Unblinded Run 1+2 data EE m ES

16. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 16 Unblinded Run 1+2 data EE m ES

17. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 17 Run 1,2 Unblinded: m ES 2D fit projections after  E cut: e: -110<  E<50 MeV  : -70<  E<50 MeV Preliminary !

18. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 18 Run 1,2 Unblinded:  E 2D fit projections after m ES cut 5.2724<m ES <5.2856 GeV

19. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 19 Signal Candidate Properties M(ll) – no apparent pileup near the J/  vetoes M(ll) – no apparent pileup near the J/  vetoes 2 of these consistent with D mass M(Kl) – possible background from B  D , D  K , both  ’s mis-id’d as electrons. (Note, this peaking BG is explicitly vetoed in K  channel). M(Kl) – possible background from B  D , D  K , both  ’s mis-id’d as electrons. (Note, this peaking BG is explicitly vetoed in K  channel). Simulation predicts 0.06 events of this background for this channel Simulation predicts 0.06 events of this background for this channel Studies of electron mis-id probabilities show no indication of problem. Studies of electron mis-id probabilities show no indication of problem. Nevertheless, include systematic error to account for possibility that 2 of these events are BG. Nevertheless, include systematic error to account for possibility that 2 of these events are BG. Preliminary !

20. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 20 Fit Results I Results of max likelihood fit in  E – m ES plane for the 8 channels Results of max likelihood fit in  E – m ES plane for the 8 channels Preliminary !

21. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 21 Fit Results II Combining channels: m ES and  E projections for Kll and K*ll Combining channels: m ES and  E projections for Kll and K*ll B(B  K*ee)/B(B  K*  )=1.21 from Ali, et al, is used in combined K*ll fit. Preliminary !

22. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 22 Systematic Uncertainties Systematic errors on the efficiency  Trk eff.  Model dependence Systematic errors on the fit yields  Signal shapes  Background shapes oincludes peaking background uncertainty Largest sources ~ 7 – 11 % ~ 0.5 – 2.0 events

23. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 23 Signal Statistical Significance Statistical significance of signal is computed: Statistical significance of signal is computed:  Toy MC: fit to background-only toy experiments and observe how often we obtain signal larger than observed signal in data.  Consider change in ln L when fixing the signal component to zero in fit (Gaussian approximation).  Results: Kl + l -  5.0  if systematics included  still > 4  K*l + l -  3.5  Based on the K*l + l - result we place an upper limit for this channel: Based on the K*l + l - result we place an upper limit for this channel: @ 90% CL Preliminary !

24. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 24 Comparison to Our Run 1 Result Run 1: Run 1: Run 1+2: Run 1+2: All data fully reprocessed for Run 1+2 results: improvements in tracking, vertex detector alignment, etc.  resulted in migration of events in/out of signal region. Sensitivity of this analysis is mostly unchanged by the reprocessing (some improvement in K s modes). All data fully reprocessed for Run 1+2 results: improvements in tracking, vertex detector alignment, etc.  resulted in migration of events in/out of signal region. Sensitivity of this analysis is mostly unchanged by the reprocessing (some improvement in K s modes). Migration of events into/out of signal region checked with control samples  results are compatible Migration of events into/out of signal region checked with control samples  results are compatible The probability for a Kll branching fraction at our new value to give our Run 1 result is at the 2-3% level. Preliminary !

25. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 25 Conclusions We have studied the channels B  Kl + l - and B  K*l + l - using 56.4 fb-1 of data at the BaBar experiment at PEP-II. We have studied the channels B  Kl + l - and B  K*l + l - using 56.4 fb-1 of data at the BaBar experiment at PEP-II. We obtain the following results: We obtain the following results: The statistical significance for B  Kl + l - is computed to be > 4  including systematic uncertainties. The statistical significance for B  Kl + l - is computed to be > 4  including systematic uncertainties. Preliminary !

26. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 26 Peaking Backgrounds Usually due to particle mis-idenficiation, e.g.: Usually due to particle mis-idenficiation, e.g.: Mis-id’d as muons  K  background Since mis-id probability is higher for muons than for electrons, explicit vetoes are applied for the muon modes. Since mis-id probability is higher for muons than for electrons, explicit vetoes are applied for the muon modes. Summary of peaking backgrounds as obtained from high statistics Monte Carlo sample. Summary of peaking backgrounds as obtained from high statistics Monte Carlo sample. These are included in fit to extract signal. These are included in fit to extract signal.

27. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 27 BaBar Run 1 Analysis (20.7 fb -1 ) Projections of the 2D fit onto m ES after a  E cut.

28. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 28 Belle results (29.1 fb -1 ) Bkgd shape fixed from MC

29. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 29 Continuum Background Suppression Continuum suppression: exploit fact that continuum events are more jet-like than BB events Continuum suppression: exploit fact that continuum events are more jet-like than BB events  R 2 : W-F 2 nd moment  Cos  thrust : angle of candidate thrust axis  Cos  B : angle of B in CM  m Kl : Kl invariant mass Combine optimally using Fisher discriminant Combine optimally using Fisher discriminant Put plot here. Put plot here. e+e+ e-e- e+e+ e-e- qq Signal B Other B

30. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 30 Generator-level q 2 Distributions from Form-Factor Models Ali et al. 2000 (solid line) Colangelo 1999 (dashed line) Melikhov 1997 (dotted line) Shapes are very similar!

31. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 31 Particle Identification Electrons – p* > 0.5 GeV shower shapes in EMC E/p match Muons – p* > 1 GeV Penetration in iron of IFR Kaons dE/dx in SVT, DCH  C in DRC E/p from E.M.Calorimeter Shower Shape e e   1 < p < 2 GeV/c 0.8 < p < 1.2 GeV/c E/p > 0.5 e  e   c from Cerenkov Detector e  0.5 < p < 0.55 GeV/c dE/dx from Dch 0.8 < p < 1.2 GeV/c

32. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 32 The DIRC is able to identify The DIRC is able to identify particles via a measurement particles via a measurement of the cone angle of their emitted of the cone angle of their emitted Cherenkov light in quartz Cherenkov light in quartz Provides good  /K separation for Provides good  /K separation for wide momentum range wide momentum range (up to ~4 GeV/c) (up to ~4 GeV/c) Particle Quartz bar Cherenkov light Active Detector Surface DCH DIRC Kaons with DIRC

33. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 33 Measure Angle of Cherenkov Cone in quartz –Transmitted by internal reflection –Detected by PMTs Particle Identification (DIRC) ( Detector of Internally Reflected Cherenkov Light ) Particle Quartz bar Cherenkov light Active Detector Surface

34. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 34 DIRC  c resolution and K-  separation measured in data  D *+  D 0  +  (K -  + )  + decays DIRC  c resolution and K-  separation measured in data  D *+  D 0  +  (K -  + )  + decays Particle Identification (DIRC) cont’.  (  c )  2.2 mrad >9>9 2.5  K/  Separation

35. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 35 J/  Control Samples: Lepton energy distributions Points: data Histo: MC Electron channels

36. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 36 J/  Control Samples: Lepton energy distributions Muon channels Points: data Histo: MC

37. K (*) ll Results from BaBar, J. Walsh, FPCP-2002 37 e + e -   (4s)  BB data used for this talk e + e -   (4s)  BB data used for this talk Run 1: 20.6 fb -1 (1999-2000) Run 1: 20.6 fb -1 (1999-2000) 23 million BB events Run 2: 55 fb -1 (2001-2002) Run 2: 55 fb -1 (2001-2002) 60 million BB events (so far) (so far) e + e - annihilation e + e - annihilation 40 MeV below  (4s) 40 MeV below  (4s) Run 1: 2.6 fb -1 Run 1: 2.6 fb -1 Run 2: 6.2 fb -1 Run 2: 6.2 fb -1 Data Sample This talk: 56.4 fb -1 on peak