1. Evidence for the Flavor Changing Neutral Current Decays ICHEP 2002, Amsterdam, July 27, 2002 Jeffrey D. Richman UC Santa Barbara for the B A B AR Collaboration

2. Outline Theoretical predictions Theoretical predictions Experimental status Experimental status B reconstruction procedure, backgrounds, and event selection B reconstruction procedure, backgrounds, and event selection Fits and results Fits and results Conclusions Conclusions

3. Theory predictions: the standard model and beyond : sensitive to Wilson coefficients C 7, C 9, C 10 in OPE : depends almost entirely on |C 7 | New, heavy particles can affect the rate (2X) and decay distribs.

4. Theoretical predictions based on the standard model New calculations of QCD corrections predict too high a rate for B->K*  ; the necessary adjustment of T 1 form factor lowers the prediction for B->K*l + l -. dominant uncertainty: form factors

5. Generator-level distributions of from form-factor models Ali et al. 2000 (solid line) Colangelo 1999 (dashed line) Melikhov 1997 (dotted line) Shapes are very similar!

6. Decay rate vs. in the SM and SUSY J/   (2S)K q2q2 q2q2 SM nonres SUSY models Pole from K*  even in  +  - constructive interf. destructive SM theoretical uncertainty

7. Recent experimental results Belle (29.1 fb -1 ) [K. Abe et al., PRL 88, 021801 (2002).] BABAR (20.7 fb -1 )[B. Aubert et al., PRL 88, 241801 (2002).] BABAR (56.4 fb -1 ) [FPCP, DPF conferences] Today: Run 1+2 prelim. result (77.8 fb -1 => 84.4 M BB)

8. Decay modes and final states

9. B reconstruction with Typical resolutions:  (m ES )  2.5 MeV,  (  E)  25 - 40 MeV Fit region ( “ blinded ” ) Large radiative effects in electron channels (Geant signal MC)

10. Backgrounds and event selection Veto charmonium mass regions and J/  K->K*l + l - feedup Combine E miss, vertex information, and B production angle into a “B likelihood” variable. Jet-like event shape and M(Kl) [D->Kl into Fisher variable Veto and/or include in fit.

11. Charmonium veto regions in  E vs. m ll (Monte Carlo simulation) Nom. sig. region Radiation can pull m l+l- below the J/  mass, but  E becomes negative!

12. Check of charmonium tails in m l+l- Data=Points with error bars; MC=histogram  The MC simulates bremsstrahlung effects reasonably well. B A B AR preliminary

13. Charmonium control sample yields  MC normalization based on published B A B AR measurements.

14. vetoes in Double fake Double fake Triple fake Triple fake Typical mis-ID probabilities (momentum dependent): P(  e)=0.1%-0.3%; P(  )=1.3%-2.7% P(    e)=0.1%-0.3%; P(    )=1.3%-2.7% Reassign particle masses and veto D region in M(K  ) and M(K*  ) in muon channels.

15. Estimates of remaining peaking backgrounds  Except for contribution from B->K* , these backgrounds are computed by applying measured hadron->lepton fake rates to a MC sample 9X the data. Peaking backgrounds are not negligible in the muon channels.

16. Data:  E vs. m ES for modes Full area= fit region Small rectangle=nominal signal region B A B AR preliminary

17. Data:  E vs. m ES for modes Full area= fit region Small rectangle=nominal signal region B A B AR preliminary

18. m l+l- for events in nominal signal region (includes some background)  Events do not cluster on veto boundaries.

19. Fit method 2-dim, unbinned maximum likelihood fit in m ES vs.  E 2-dim, unbinned maximum likelihood fit in m ES vs.  E Signal shape:  Use Geant MC with shifts from J/  samples; Crystal Ball parametrization. Background shape: ARGUS shape in m ES ; exponential in  E. Possible correlation is considered in evaluation of systematic error. The background normalization and shape parameters float in the fit. The background normalization and shape parameters float in the fit.

20. Data: fit projections onto m ES and  E for and B A B AR preliminary

21. Data: fit projections onto m ES and  E for and B A B AR preliminary

22. Data: fit projections onto m ES and  E for and B A B AR preliminary

23. Data: fit projections onto m ES and  E for and B A B AR preliminary

24. Yields, efficiencies, systematic errors, and branching fractions B A B AR preliminary Low muon system efficiencies  install replacement detectors

25. Combined fits: all channels preliminary To combine the K*ll modes, we assume K*ee/K*  =1.2 (Ali et al.). To combine the K*ll modes, we assume K*ee/K*  =1.2 (Ali et al.). 4.4  2.8  Significance:

26. Combine channel fits The combined significance of the B->Kll channels is 4.4  including systematic errors). The combined significance of the B->Kll channels is 4.4  including systematic errors). There is an indication of a signal in B->K*ll, but the significance is only 2.8  including systematic errors  There is an indication of a signal in B->K*ll, but the significance is only 2.8  including systematic errors  This result is consistent with our result from 56.4 fb -1 (and is somewhat higher than our original upper limit based on 20.7 fb -1 ). This result is consistent with our result from 56.4 fb -1 (and is somewhat higher than our original upper limit based on 20.7 fb -1 ).

27. Conclusions We have updated our analysis with a sample of 84.4 M BB events, about four times the size of our original sample. We have updated our analysis with a sample of 84.4 M BB events, about four times the size of our original sample. We observe a B->Kl + l - signal (4.4  ) with branching fraction We observe a B->Kl + l - signal (4.4  ) with branching fraction The central value is higher than most theoretical predictions based on the SM, but the uncertainties are large (both expt and theory), so there is no inconsistency. The central value is higher than most theoretical predictions based on the SM, but the uncertainties are large (both expt and theory), so there is no inconsistency.

28. Conclusions (cont.) We observe events consistent with a B->K*l + l - signal, but the significance is only 2.8  We observe events consistent with a B->K*l + l - signal, but the significance is only 2.8  These modes will be studied for many years to come. We can expect These modes will be studied for many years to come. We can expect  better measurements and predictions for branching fractions  studies of kinematic distributions, which will provide additional sensitivity to new physics with less model dependence.

29. Backup Slides

30. Multiplicative systematic errors: efficiencies and event sample

31. Systematic errors on the fit yields Signal shape: allow 50% larger radiative tail; vary m ES,  E means from control sample values to MC values. Signal shape: allow 50% larger radiative tail; vary m ES,  E means from control sample values to MC values. Combinatorial background shape: fix m ES slopes to MC values; allow m ES slope to have quadratic dependence on  E. Combinatorial background shape: fix m ES slopes to MC values; allow m ES slope to have quadratic dependence on  E. Peaking background estimates: typically 100% uncertainties. Peaking background estimates: typically 100% uncertainties.

32. BaBar Event Display (view normal to beams) R drift chamber =80.9 cm (40 measurement points, each with 100-200  m res. on charged tracks) EM Calorimeter: 6580 CsI(Tl) crystals (5%  energy res.) Silicon Vertex Tracker 5 layers: 15-30  m res. Cerenkov ring imaging detectors: 144 quartz bars (measure velocity) Tracking volume: B=1.5 T

33. Control samples We make extensive use control samples in the data to check the MC, both for signal and background. We make extensive use control samples in the data to check the MC, both for signal and background.  Decays to charmonium.  Decays to charmonium. Each final state has a “signal-like” control sample that is identical except for the restricted range of q 2. Checks MC predictions for cut efficiencies.  “Large sideband”  “Large sideband” in m ES vs.  E plane: checks combinatorial background  Ke   Ke  : monitors combinatorial background

34. Check of E lep distributions using J/  K(*) samples Data=Points with error bars; MC=histogram  We understand lepton ID efficiencies vs. lepton momentum. B A B AR preliminary

35. Check of B likelihood distributions using J/  K(*) Data=Points with error bars; MC=histogram  Use this comparison to evaluate systematic errors on efficiencies. B A B AR preliminary Keep

36. Check of Fisher variable (continuum suppression) distributions using J/  K(*) Keep  Use this comparison to evaluate systematic errors on efficiencies.

37. Fit example: projections onto m ES slices  E<-0.11 GeV -0.11<  E<0.05 GeV  E>0.05 GeV B A B AR preliminary

38. Data: fit projections onto  E slices m ES <5.24 GeV/c 2 5.24<m ES <5.273 GeV/c 2 m ES >5.273 GeV/c 2

39. Forward-backward asymmetry vs. q 2 Polar angle of lepton in dilepton rest frame. q2q2q2q2 (GeV 2 ) A. Ali et al., PRD 61, 074024 (2000). standard model SUGRA models MIA SUSY