Search for B     with SemiExclusive reconstruction C.Cartaro, G. De Nardo, F. Fabozzi, L. Lista Università & INFN - Sezione di Napoli

04/10/03F.Fabozzi2 Searches for B    at BaBar Two analyses –based on data sample (81.9 fb-1) –First: reconstruct one B meson Semi-Exclusively In a Dl X mode (X = ,  0, nothing) –Then: the recoil is analyzed to search for a B    decay –  decays are reconstructed in the following modes: Semi  Leptonic Tags –   (e,  ) (e,  )  Semi  Exclusive Tags –  (e,  ) (e,  )  –  ( ,  0,  )  –The samples of the two analyses are statistically independent and the results have been combined

04/10/03F.Fabozzi3 B sample counting In Semi-Exclusive analysis we fully reconstruct a B meson in a B  DX had mode For the BR determination we need the number of B  B  events with a fully reconstructed B –determined with a fit to m ES –for the systematics we fit also with a Gaussian (4.5% less events) Too conservative? We’ll try to fit with a double Gaussian sideband peak m ES distribution fitted as Argus  Crystal ball B sample counting: –N B  B  = (1.67  0.09)  10 5

04/10/03F.Fabozzi4 Events preselection Preselections for the interesting channels –1 GTL & 0  0 –1 GTL & 1  0 –3 GTL & 0  0 Then fit to the m ES distributions –Crystal Ball + Argus Fits after the preselection used for –data-MC comparison –expected background prediction The shape of the Argus function after the preselection is in agreement with the shape after the full selection m ES data, 1GTL & 0  0 m ES data, 3GTL & 0  0 m ES data, 1GTL & 1  0

04/10/03F.Fabozzi5 Selections of tau decays Optimized for the best upper limit         events decay proceeds via two intermediate resonances, an a 1 and a  3 tracks and 0  0 P missing > 1.2 GeV/c Neutral Energy < 100 MeV Neutral Bumps < MeV < m(    0 ) < 950 MeV 1.1 GeV < m(  +     ) < 1.6 GeV |p 1 + p 2 + p 3 | > 1.6 GeV/c (c.m.s.) Lepton & kaon veto SemiExcl purity mode > 30% 1-prong events 1 track, 0  0 No K S Neutral Energy <110MeV Neutral Bumps < 1 P missing > 1.2 GeV/c Kaon veto Particle ID     only Lepton veto p c.m.s. > 1.2 GeV/c SemiExcl purity mode > 50%       0 events  decay proceeds via an intermediate  1 track and 1  0 p missing > 1.4 GeV/c Neutral Energy < 100MeV 0.55 GeV < m(    0 ) < 1 GeV SemiExcl purity mode > 50% Rejecting events with p miss in the beam pipe?

04/10/03F.Fabozzi6 Nominal Int. Pur. = 50 % Integrated purity cut For each selection we apply a cut on the “nominal”integrated purity of the B reco side –Different from the integrated purity that we see from the fit to m ES distribution –Not relevant from the point of view of the analysis but generated confusion during review Charged B sample

04/10/03F.Fabozzi7 Selection efficiencies 0.6% 0 0.1% 0 0        0 6.6% 0 1.2% 0 0    0 0.8%    0  0 0.4%21.6%2.7%0.1%     6.8%000      0 0.1%1.0%0.3%0         00.5%7.4%0     00.6%022.9%    e            e  Total efficiency = 11.3%

04/10/03F.Fabozzi8 Selected events B     B     B     B     Wrong sign control sample Unblinded data candidates

04/10/03F.Fabozzi9 Background from generic B  B  Mostly from semileptonic Vcb –Can we fight this background looking at the charged tracks list? 1-prong lepton 1-prong pi rho-channel Semilept. Vcb Semilept. Vub Semilept. Vcb Semilept. Vcb Hadronic Vcb Hadronic Vcb only 3 events selected 2 semilept. Vcb 1 hadronic charmless 3-prongs channel

04/10/03F.Fabozzi10 Systematic error (I) Uncertainty on N BB Uncertainty on signal selection efficiencies –Better understanding of systematics on neutrals Now: smearing procedure on the photon energy Cut on neutral energy in the EMC –Sensitive to the Data-MC agreement in the low-energy region (ex. simulation of machine background in the EMC) –Developing a module to study     events Tag one  (  -3prongs) and look the neutral energy distribution associated to the other  decay –Data-MC comparison

04/10/03F.Fabozzi11 Systematic error (II) Uncertainty on the expected background (b i ) estimate –Peaking background: from generic B  B  MC More generic MC is available  will reduce main systematic error –Continuum+combinatorial background: from data sideband scaling to signal region based on the fitted Argus shape Possible dependencies of the fitted Argus shape on the variables used in the selection taken into account –m ES fits in bins of a given variable »a correction factor for each variable Total correction as the product of the individual corrections –assumption that variables are uncorrelated –the effect of correlation between variables is small (total error on b i : 4.9  5.0, since the statistical error on b i is dominant We may think to remove highly correlated variables if they do not add in signal-to-background discrimination

04/10/03F.Fabozzi12 Branching fraction extraction Likelihood ratio estimator Q –Value of branching fraction  Minimum of -2logQ Br(B      )= 1.1  10   logQ min gives the statistical significance of the signal –~ 0.1  significance Unphysical region

04/10/03F.Fabozzi13 Upper limit determination Determination of C.L. and upper limit with a Toy MC –10000 random experiments generated for 400 values of the branching fraction Br(B      )< 7.7  (90%C.L.) Uncertainty on b i included COMBINED with semi- lept analysis: Br(B      )< 4.1  (90%C.L.)

04/10/03F.Fabozzi14 Expected sensitivity For each channel we fluctuate the observed events according to a Poisson distribution

04/10/03F.Fabozzi15 Conclusions Conference paper will be made public soon –Some delay due to discussion on if/how to quote a central value Another iteration of the analysis is starting –More signal and generic B  B  MC available –Better understanding of neutral systematics –Try to see if margins to improve background rejection Semileptonic Vcb is the main source