1. LPNHE Richard Kass December 21 2005 1 Measurement of the Branching Fraction of B - → D 0 K* - J. Chauveau, M. John, R. Kass, G. Thérin Details of this analysis can be found in: BADs 697 & BAD1320 (PRD-RC draft) Also: BAD 1073 (GLW) (PRD-RC 72, 071103 (2005)) BADs 1141&1269 (ADS) (PRD-RC 72, 071104 (2005)) Outline of this talk Introduction Event Selection Backgrounds Fit strategy and fit results Systematic error estimate Summary/Conclusions

2. LPNHE Richard Kass December 21 2005 2 Why measure this branching fraction? Color favored b  c amplitudeColor suppressed b  u amplitude Cabibbo suppressed c  d amplitudeCabibbo favored c  s amplitude  This BF measurement is a by-product of the GLW and ADS analyses. This measurement uses ~2.7X more data than BaBar’s previous one Use 3 D 0 decay modes: K , K  0, K  Reconstruct K *- using only charged tracks K *-  K s  -      Ks   Experimentfb -1 BFx10 -4 Reference CLEO9.1 6.1 ± 1.6 ± 1.7 PRL 88,101803 (2002) Belle88 5.2 ± 0.5 ± 0.6 Belle-Con-0316 (2003) BaBar81.56.3±0.7±0.5PRD 69, 051101(2004) Previous B(B -  D 0 K *- ) Measurements BAD621

3. LPNHE Richard Kass December 21 2005 3 Definition of Some Variables B-B- D0D0 K*K* B rest frame -- KsKs K* K* rest frame -- KsKs B-B-  Helicity angle,  H, is the angle between the B and  in the K* rest frame Beam-energy substituted mass, m ES s=(CM energy) 2 0=e + e - system B=B-meson candidate *= CM quantity Energy difference,  E

4. LPNHE Richard Kass December 21 2005 4 Event Selection Criteria Differences between “original” (PRD 69, 051101(2004)) & this analysis: we do not always require GoodTracksLoose (sometimes use GoodTracksVeryLoose) original used Release 10, we use Release 12 (runs 1-3) & Release 14 (run 4)

5. LPNHE Richard Kass December 21 2005 5 Selection Criteria & Efficiencies original eff. (%) 12.8 3.5 4.0 PRD 69, 051101(2004) correction factors determined by comparing data (B -  D 0  - ) with MC correction factors determined using standard procedures/recipes Selection Efficiencies Handling of multiple B candidates in an event: occurs <25% of time choose candidate with smallest  2 : correct candidate picked 80% of time no bias introduced

6. LPNHE Richard Kass December 21 2005 6 Quick MC/Data Comparison Perform simple 1D unbinned Likelihood fit to m ES distributions after all cuts using Gaussian and Argus PDFs. Reasonable agreement red: signal green: B + B - blue: B 0 B 0 yellow: c c u u /d d /s s : white Two peaking backgrounds apparent: D 0 not correctly reconstructed → will add a PDF in ML fits non-resonant K  : B - →D 0 K s  - → will subtract from yields

7. LPNHE Richard Kass December 21 2005 7 K* and NR Background BW  resolution K  mass & helicity distribution for B -  D 0 [K s  - ] events Pre-selection mass window only ±125 MeV (analysis uses ±75 MeV) Too small for good estimate of NR background Clear asymmetry in helicity distribution. Extract NR contribution by fitting helicity distribution to spin 1 (K*) and spin 0 (NR) components S-wave + P-wave interference Require m ES > 5.27 GeV/c 2 combinatoric background

8. LPNHE Richard Kass December 21 2005 8 Accounting for Non-Resonant K  We model the amplitudes including NR-K  taking into account: Angular momentum difference between R and NR: K* is spin 1, cos  angular dependence NR-K  consistent with spin 0, flat angular dependence K  mass dependence: K* line shape described by a Breit-Wigner “BW(s)” NR-K  consistent with flat (or linear) mass spectrum Define the following quantities:  c e i  c =A(B - →D 0 K s  - )/A(B - →D 0 K * ): ratio of amplitudes for the 3-body background and the signal Including the helicity angle (  ) into the amplitude ( א) gives:   is the ratio of non-resonant to resonant B→DK* decays:  c =constant*  c

9. LPNHE Richard Kass December 21 2005 9 Efficiency corrections for Cos  Estimate efficiency as a function of cos  from the signal MC Assume the K* signal MC gives a cos 2  distribution with perfect detector resolution Efficiency correction Signal MC parabola uncorrected data+MC corrected data+MC cos  resolution ~0.004 red: signal green: B + B - blue: B 0 B 0 yellow: c c u u /d d /s s : white

10. LPNHE Richard Kass December 21 2005 10 Determination of  c resolution estimated from toy MC studies Also find:  c =(-90  37) 0 Reduce the event yields by (4  1) % to account for NR background Scan the likelihood function to determine  c consistency check: estimate  using B 0  D - (K s  - ) analysis of BAD820:  2 =0.023  0.036 Perform a 2D ML fit to m ES and cos  H (K*) to determine . m ES PDF=Argus  Gaussian, cos  H PDF previously discussed Must fix the value of  otherwise fit is unstable  small  2 cc Many checks on this procedure: toy MCs, MC without interference m ES sideband data

11. LPNHE Richard Kass December 21 2005 11 Event Yield Fit Procedure Event yields are determined from an unbinned maximum likelihood fit in the region 5.2  m ES  5.3 GeV/c 2 Choose simple PDFs to fit m ES distributions: A=Argus G=Gaussian Perform ML fits simultaneously in 3 regions. In each region fit K , K  0, K3  m ES distributions (k=1,2,3) 9 PDFs in all I) |  E| Sideband: -100  |  E|  -60 MeV & 60  |  E|  200 MeV pdf: A k II) D 0 Sideband: |m D -m D,PDG |  Take into account “doubly peaking backgrounds (DP) pdf: (N noP A+N DP G) k III) Signal region: |  E|  25MeV pdf: (N q q A+  N DP G+N sig G) k  scales the N DP found in D 0 the sideband fit.  =(D 0 signal region)/(D 0 sideband region) =0.24 for K , K3  and 0.88 for K  0

12. LPNHE Richard Kass December 21 2005 12 Results from Simultaneous ML Fits Do a simultaneous unbinned ML fit to all 9 samples    Gaussian mean ( MeV/c 2 ) 5279.53  0.255279.93  0.295279.97  0.28 Gaussian sigma ( MeV/c 2 ) 2.66  0.232.63  0.322.79  0.25 Argus parameter  -17.7  4.0-23.2  2.4-19.1  1.6 Argus endpoint ( MeV/c 2 )*5291.00 *Argus endpoint is fixed PDF parameters from the fits

13. LPNHE Richard Kass December 21 2005 13 Results from Simultaneous ML Fits D 0 sideband  E sideband D 0  K  0 D0KD0K D0KD0K D 0  K3  A simultaneous unbinned ML fit to all 9 samples Argus: 493.1  22.2 Argus: 1093.9  33.1 Argus: 2384.3  48.8 Argus: 212.8  15.2 N DP : 13.2  5.5 Argus: 516.9  22.9 N DP : 0.0  13.0 Argus: 2558.2  53.3 N DP : 31.0  17.8

14. LPNHE Richard Kass December 21 2005 14 Results from Simultaneous ML Fits Signal region D 0  K  0 D0KD0K D 0  K3  Argus: 151.4  13.3 Signal: 144.4  13.2 Argus: 770.6  30.5 Signal: 185.4  18.6 Argus: 923.7  32.4 Signal: 195.0  18.2 ~520 signal events

15. LPNHE Richard Kass December 21 2005 15 Branching Fraction Results    , efficiency (%) 13.304.608.82 B(D 0  X) (%) 3.8012.847.46 N, Yield (events) 144.4  13.2185.4  18.6195.0  18.2 B(B -  D 0 K *- )x10 -4 5.15  0.455.65  0.545.34  0.48 Statistical uncertainties only N(B + B - )=2.318x10 8 from 210.7fb -1 B(K *- )  B(K *-  [  +  - ] Ks  - )=0.230 f K* = 0.96  0.01  fraction of K*’s in K s  mass window PDG #’s

16. LPNHE Richard Kass December 21 2005 16 Branching Fraction Comparisons This Analysis: Runs 1& 2 Vs Runs 3&4 Old BaBar Vs New BaBar Analysis fb -1    Avg. BF OLD 81.5 5.8  1.05.8  1.28.7  1.56.3  0.7 NEW 211 5.15  0.475.65  0.575.34  0.505.31  0.30 Note: Older analysis did not correct for NR background

17. LPNHE Richard Kass December 21 2005 17 Systematic Errors Summary of Systematic Errors uncorrelated 3.1% 6.6% 4.7% uncorrelated 5.2% 6.2% 7.3% correlated standard recipes data (B -  D 0  - ) Vs MC study data & MC, vary cuts PDG BF uncertainties finite MC samples lumi script study data & MC, vary cuts

18. LPNHE Richard Kass December 21 2005 18 Branching Fraction Averaging Procedure Follow the procedure outlined in Lyons et al., NIMA 270, 110 (1988) Calculate a weighted average using the three measurements: B(B -  D 0 K *- )=w 1 B(B -  D K *- )+w 2 B(B -    D K *- )+w 3 B(B -  D K *- ) The weights (w 1, w 2, w 3 ) are calculated using the error matrix, V: V= V statistics + V systematics Calculate weights: Define vectors w and u: st=statistics syTOT=total systematics syC=correlated systematics Calculate variances:

19. LPNHE Richard Kass December 21 2005 19 Summary and Conclusions Using 211 fb -1 (runs 1-4) we have measured: B(B -  D 0 K *- )=(5.31±0.30±0.34)x10 -4 Previous BaBar result: 6.3±0.7±0.5 from 81.5 fb -1 (runs 1&2) BAD697V16 is available BAD1320V3 PRD-RC draft is available Would like to go to CWR (very) soon Would like this BF to make it into PDG2006

20. LPNHE Richard Kass December 21 2005 20 Extra Slides…

21. LPNHE Richard Kass December 21 2005 21 Toy MC verification of a 2-D fit to cos  H and m ES Need to fix  c to have a stable fit Guess  c from iso-spin similar B 0 →D - K S  analysis to be 10-15% Resolution on a  c measurement found to be: 0.34 for 280 signal events + 340 background events. corresponds to the fit sample after ADS selection applied toK ,K  0, K  modes 0.24 for 400 signal and 1500 background events corresponds to the fit sample after GLW cuts (BAD697)

22. LPNHE Richard Kass December 21 2005 22 For example, on the GLW selection with  c =0.13 2-D fit to cos  H and m ES D 0 modeN(signal) ξ mean (MeV)  (MeV/c 2 ) sin  c signif* GLW All 3401±25-25.4±115279.78±0.172.57±0.16-1.68±0.24 4.9  Many other fits in BAD697V16, all show large asymmetry Fit m ES sideband (<5.26 MeV/c 2 ) to Acos  2 +Bcos  +C → B=-1.6±1.1, signif=1.1 

23. LPNHE Richard Kass December 21 2005 23 Why no systematic for Double Peaking Background ? Our signal peak is about 520 events (three modes together) while the sum of the peak fitted in the D0 sideband is 44 +- 19 events. However, the sideband is about 4 times larger than the signal region, the subtracted background is about 11+-5 events under a peak of 520 signal events; i.e. 2%. Imagine our assumption is not correct and the background gaussian is, e.g., 50% larger (or smaller) than the signal. This only adds (subtracts) 1% onto the total yield. The addition of such a 1% correlated systematic into the final BF calculation, would go unnoticed (to 2 decimal places).