1. A measurement of the B 0 B 0 oscillation frequency and determination of flavor-tagging efficiency using semileptonic and hadronic B 0 decays S. Bolognesi & M.A. Borgia for the CP-violation exam

2. Introduction  The strategy  The experimental environment S. Bolognesi & M.A. Borgia CP-violation exam

3. The measurement top contribute is dominantsensible to V td element of CKM matrix  one B reconstructed in a flavour eigenstate (B rec ) one B only tagged as B 0 or B 0 from its decay products (B tag ) mixed if same flavor / unmixed if opposite flavor  PDF for the two categories (mixed + / unmixed -) “dilution factor” due to mistag rate    distance between B rec and B tag decay (≈  B = 1.548±0.032 ps) if perfect flavour tagging ≈ time resolution function with parameters  B 0 B 0 mixing through NLO  EW diagrams involving exchange of up-type quarks S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 3

4. Likelihood and time independent analysis  Likelihood = sum over all events (mix. & unmix.) and over different tag types (with its own D i ) minimized to extract simultaneously  m d, D i (and some a i ) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 4  Time-independent analysis = neglecting background and assuming B rec correct identification, the observed time-integrated fraction of mixed events χ obs as a function of B 0 B 0 mixing probability χ d can be expressed as: where ω is the mistag rate, and χ d = ½ x d 2 /(1+x d 2 ) = 0.174 ± 0.009 and x d 2 = Δm d /Γ

5. BaBar detector  DCH + SVT detection and momentum measurement for charged particles  SVT vertex information  DIRC  z ≈ 50  m for B rec  z ≈ 100-150  m for B tag particle identification (charged hadrons)  DCH particle identification (dE/dx)  EMCphotons, electrons and neutral hadrons  IFR (RPC) muons and neutral hadrons S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 5 8.9 fb -1 @ Y(4s) + 0.8 fb -1 @ 40 MeV below Y(4s) (10.1 ± 0.4) × 10 6 BB pairs

6. Particle identification  Electrons track + EMC (shower shape, E/p) dE/dx in DCH Cherenkov angle in DIRC efficiency 92% mistag (  ) 0.3%  Muons interaction lenghts and # hits in IFR MIP in EMC efficiency 75% mistag (  ) 2.5%  Kaons where dE/dx Cherencov angle # photons efficiency 85% mistag (  ) 5% S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 6

7. Time resolution  (  t) dominated by  (  z)  (  z) dominated by  (  z Btag ) pseudo-track extrapolated from the interaction point in the Btag direction  reconstruct B rec  compute the B tag direction from the energy conservation  B tag vertex = intersection of pseudo-track with all the other tracks  B rec Beam Spot pseudo track (B tag ) zz  z ≈ 260  m  z) ≈ 180  m  Resolution function is the sum of three gaussians (3 parameters from MC 3 parameters from fit) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 7

8. Flavor tagging  4 strategies to define if Btag is B 0 or B 0 4 tagging cathegories  Lepton tag:presence of a prompt lepton (p CM >1.1 GeV against charm semileptonic decay)  Kaon tag:total kaons charge not 0  2 neural network cathegories: 5 neural network algorithm 4 based on tracks 1 exploits the charge of high momentum particles whose outputs are combained in a single full neural network tagger (x NT [-1,1]) NT1 NT2 S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 8

9. Analysis  Hadronic decay channels  Leptonic decay channels S. Bolognesi & M.A. Borgia CP-violation exam

10. Hadronic decays B 0 rec D* -  + ( /  + / a 1 + ) D0 -D0 - K +  - K +  -  0 K +  +  -  - K0s + -K0s + - D -  + ( /  + / a 1 + ) K +  -  - K0s -K0s - B 0 rec J/  K* 0 e + e - /  +  - (K 0 s →  +  -,  0 →  )  Usual cuts on intermediate/final particles:  resonances invariant mass (±2  )  vertex  2  threshold on momenta  opening angle between decay products S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 10  B 0 candidates characterized by  beam energy sobstituted mass: 5.270 < m ES < 5.290 (GeV)  E B0 – E beam in Y(4s) CM: |  E| < 3   E where   E = E resolution (19 -40 MeV)  Cuts against continuum (e + e - → qq)  normalized second Fox-Wolfram moment (R 2 =H 2 /H 0 ) < 0.5  large angle between thrust axis of B 0 and of the remaining tracks

11. Backgrounds (HD*) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 11 Data extracted from fit to the m ES distrbution * HD = Hadronic decays

12. Semileptonic decays S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 12  D 0 candidates:  combination with all charged tracks (p Tmin 50 MeV/c and charge opposite to that of the candidate K) => D* candidates  Usual cuts on intermediate/final particles:  invariant mass (±2  ) around nominal D 0 mass  vertex  2 > 1%  threshold on momenta  Mass difference:  m(D* - )-m(D 0 ) (± 2.5σ) of the nominal value  D* candidates:  D*-, p l > 1.2 GeV back-to-back => cosθ(D*- ) < 0  Neutrino existence consistency: (solving in the Υ(4s) system frame)

13. Sample composition S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 13  After cuts, 7517±104 B→D*lν events 3101±64 in the mode 1986±51 in the mode 2430±56 in themode  Mass difference distributions for each flavor tagging category Backgrounds are larger for semileptonic modes than for hadronic modes

14. Background  Combinatorial Due to falsely reconstructed D* candidates Estimated by fitting Δm(D*-D 0 ) distributions Gaussian + threshold function with a sharp rise followed by exponential tailoff Signal region within ±2.5σ of the peak in Δm(D*-D 0 ) Combinatorial background control sample provided by the sidebands region 0.150 < Δm(D*-D 0 ) < 0.160 GeV/c 2 S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 14  Three types of background to B 0 →D*lν : Combinatorial background Wrong-lepton background B + background

15. Wrong-lepton  Wrong-lepton:D* combination with wrong lepton Four potential sources: ”Fake lepton” (estimated selecting events in which a track candidate has failed very loose lepton criteria is substituted for the lepton candidate) Real D* from one B + real lepton from the other B (“uncorrelated lepton” bg) (estimated by parity-inversion of the lepton momentum in the Y(4s) frame => control sample) Events of the type B 0 →D*DX in which the D decays semi-leptonically produce a non-primary lepton (estimated with Monte Carlo, less than 1% => neglected) cc events producing real D* and lepton in back-to-back configuration. (estimated using combinatorial-subtracted off-resonance data) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 15

16. B + background Due to B-decays which involve additional final state particles ( B→D*(nπ)lν ) B 0 →D*(nπ)lν that pass selection criteria are considered as signal (they contribute to the measurement of Δm d and the additional low momentum π does not affect the tagging algorithm) B - →D *+ (nπ)l - ν considered as bg: they do not oscillate and must be corrected for in extracting Δm d and their mistag rate may differ from that of B 0 decays as well S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 16

17. Results  Likelihood fit results  Time integrated method results  Combined results S. Bolognesi & M.A. Borgia CP-violation exam

18. Backgr. treatment (LM*)  PDF must be extended with background contributions b = background sources i = tagging cathegories)  f = fraction of signal or background events  B = empirical description of  t distribution in background events (where  Fit to the background control samples (m ES sidebands) to determine time dependence, dilution factor, resolution function: three components for each background source zero lifetime: non zero lifetime, no mixing: non zero lifetime with mixing: S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 18 * LM = Likelihood method

19.  t distribution (LM*) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 19 * LM = Likelihood method HADRONIC SAMPLELEPTONIC SAMPLE

20. Time dependent asymmetry a(  t) (LM*) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 20 * LM = Likelihood method HADRONIC SAMPLELEPTONIC SAMPLE

21. Fit results (LM*) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 21 * LM = Likelihood method  Identical analysis procedure on MC data with detailed detector simulation:  fit results consistent with a priori insterted value and MC truth information  observed differences applied as a correction to the measured values

22. Systematic errors (LM*) S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 22 * LM = Likelihood method HADRONIC SAMPLE  Statistical error dominant, followed by MC correction uncertainties (  t for  m d )  Systematic error dominant due to big uncertanties in background characteristic (  t for  m d ) LEPTONIC SAMPLE

23. Time integrate (single bin) method  First aim: measurement of the mistag rate  Main feature: restriction of the sample to events in a single optimized Δt interval ( | Δt | < 2.5 ps because of Babar vertex resolution) Events with | Δt | > 2.5 ps have on average equal numbers of mixed and unmixed events => contribute nothing to the determination of the mistag rate Considering the different background contribution: f s, f β = fraction of signal and background source χ β = fraction of mixed events in each background source χ obs = Observed fraction of mixed events χ d = ½ x d 2 /(1+x d 2 ) and x d 2 = Δm d /Γ, while χ’ d takes into account the sample restriction S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 23 =>

24. Results  Hadronic Signal region defined as events with m ES > 5.27 GeV/c 2 Fraction of mixed events in the background determined by tag category using the sideband control sample, m ES < 5.27 GeV/c 2  Semileptonic - bg evaluated for each tag category and for each D 0 decay - mistag fractions calculated individually by tag category and decay mode using the Eq. shown - combination of the different decay modes, using the statistical errors to weight the individual results S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 24

25. Systematic errors Hadronic Semileptonic S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 25 Sources of systematic error for the mistag measurement on the hadronic and semileptonic samples

26. Comparison between the two methods S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 26 Combining results for the hadronic and semileptonic B samples for the likelihood fit method and for the single-bin method and taking into account the systematic errors Preliminary mistag rate Single bin fit uses a subset of the sample used for the other method The two sets of results are uncorrelated Good agreement between the two methods. Final result: Q ≈ 0.28

27. Final result S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 27  Hadronic sample:  Leptonic sample:

28. Back-up slides S. Bolognesi & M.A. Borgia CP-violation exam

29. Fox-Wolfram moments The Fox-Wolfram moments,, are defined by is the opening angle between hadrons and the total visible energy of the event are the Legendre polynomials To the extent that particle masses may be neglected,. It is customary to normalize the results to, i.e. to give. 2-jet events tend to give for even and for odd. S.Bolognesi & M.A. BorgiaCPV exam (27 July 2007) 29