A Search for B Recoiling Against BD0l

1
A Search for B??? Recoiling Against B-?D0l-?

• Mousumi Datta
• on behalf of
• Yibin Pan, Sau Lan Wu
• University of Wisconsin, Madison
• Physics Meeting, August 2, 2004
• Review Committee Dieter Best, Fabrizio Bianchi,
  Homer Neal, Eli Rosenberg
• Conference note BAD 980, Supporting Document
  BAD 773

2
Outline

• Physics motivation
• Existing BaBar analyses and results
• Analysis Overview
• Signal efficiency
• Background estimation from MC and data side-band
• Control samples and validation
• Systematic studies
• Limit setting procedure
• Results
• Combining results with statistically independent
  existing BaBar analysis

3
Motivating the Search

• Purely leptonic B decay. Standard Model branching
  ratio
• Provide direct measurement of B meson decay
  constant fB
• fB 0.916 0.032 GeV (PDG 2004, Lattice
  QCD)
• Extract Vub / Vtd by combining branching
  ratio measurement with results from B mixing
• Sensitive to charged Higgs, leptoquarks.

4
Branching Ratio Expectations

• Helicity Suppressed
• Standard model estimate using 2004 PDG values
  
• Existing upper limits at 90 CL
• BABAR BR(B ???) lt 4.2 10-4 (submitted
  to PRL)
• LEP (L3) BR(B ???) lt 5.7 10-4

t m e 1 5 10-3 1 10-7
fB 0.196 0.032 GeV Vub (3.67
0.47) 10-4
BR(B ? ??) (9.3 ? 3.9) 10-5
5
Analysis Strategy

• In B-factory environment
• ee- ? ?(4S) ? BB-

B???, B?X (Xanything) Main ? decay modes
??(e, ?)?(e, ?)?? ??(?, ??0, ???)??
B-?X

• First reconstruct one of the B-mesons, referred
  as tag B or tag side.
• Make requirements on the remaining
  tracks/neutrals that constrain them to be
  consistent with ? decay. This remaining part of
  the event is referred as signal side.

6
Existing BABAR Analyses (81.9 fb-1)
Semi-Leptonic Tag Analysis (X. Chen, M. Datta, Y.
Pan, S. Sekula, J. Wimmersperg, S. L. Wu) BAD
417, 598 Tag side
B? ? D0l?X
(semi-leptonic) (X ?, ?0 or nothing)
Signal Side ??
(e, ?)?(e, ?)?? (hadronic decays
of ? not considered) Upper Limit at 90 C.L. BR (
B ? ?? ) lt 6.7 ? 10?4
Hadronic Tag Analysis (C. Cartaro, G. De Nardo,
F. Fabozzi, L. Lista, S. Robertson) BAD 389,
596 Tag side B??D()0Xhad (hadronic) (Xhad
hadrons) Signal Side ??(e, ?)?(e, ?)?? ??(?,
??0, ???)?? Upper Limit at 90 C.L. BR ( B ? ?? )
lt 4.2 ? 10?4
Combined upper limit at 90 CL BR ( B ? ?? ) lt
4.2 ? 10?4
Submitted to PRL (hep-ex/0407038)
7
Analysis Overview
Data and MC Sample Run 1-3 data processed with 12
series release, SP5 MC On-resonance
112.5 fb-1 Off-resonance
11.9 fb-1 BB-
325.6 fb-1 B0B0
402.1 fb-1 cc
117.9 fb-1 uds
138.4 fb-1 ??-
235.5 fb-1 B??? VS generic 310K B???
VS Dl?X 118K
Tag Side

• Decay Branching
• Mode Fraction ()
• ?? e?? 17.84 ?? ???
  17.36 ?? ?? 11.06 ?? ??0?
  25.42 ?? ???? 9.16

Signal Side
8
Tag B Reconstruction (I)

• Global Event Selection
• Qnet 0
• Less than 11 ChargedTracks

• D0 Selection
• Mreco MPDG lt 40 MeV (K?, K3?, Ks??)
• Mreco MPDG lt 70 MeV (K??0)
• Consistent particle ID
• Vertexing, mass constraint fit

• D0 Selection
• ?0 from pi0SoftDefaultMass list,
  
• ? from GoodPhotonDefault list
• Kinematic fit

9
Tag B Reconstruction (II)

• Lepton Selection
• PID eMicroTight or
• muMicroTight
• P gt 1 GeV, vertexing

• -1.1lt cos?B,D0l lt 1.1

• Best candidate selection based on D0 mass and
  ?M
• Additional requirement on ?M and CM angle between
  D0 and ?0/?
• 0.135 GeV lt ?M lt 0.150 GeV, ?(D0,?0)lt60? for
  D0?D0?0
• 0.130 GeV lt ?M lt 0.155 GeV , ?(D0,?) lt90? for
  D0?D0?

Corrected tag reconstruction efficiency on B???
VS generic MC ?tag (1.818 ? 0.074 (stat)
)?10-3
Raw ?tag in signal MC corrected by the data-MC
ratio for on-peak data, and generic MC scaled to
data luminosity. Systematic error associated with
tagging efficiency correction will be discussed
later in the talk.
10
Signal Selection

• Signal-side track and (or) ?0 multiplicity
• PID in combination of acceptance and veto mode
  using the selectors
• PidLHElectrons (e)
• MuMicroVeryTight (?)
• KNNVeryLoose (K)
• piLHTight (?)
• Remaining neutral energy (Eextra )
• Sum of the CM energy of the neutrals (from
  CalorNeutral list), which are not associated with
  tag-side or the ?0 candidate from the ??0? decay.

11
Background Suppression

• Main source of background is BB-
• Tag B meson correctly reconstructed
• Undetected particles on the recoil side
• Events with KL and/or neutrino, frequently tracks
  and (or) neutrals pass outside detector
  acceptance.
• Continuum Bkg in hadronic ? modes
• Un-modeled events (most likely from two photon
  processes)

• Requirement on missing mass removes the
  un-modeled processes
• P requirement for the ? daughters
• Requirement on intermediate resonances
• ?? selection 0.55lt M??0 lt 1.0 GeV, -1.1lt
  cos??-? lt 1.1
• a1? selection 0.55 lt M??- lt1.0 GeV, P??-
  gt0.5 GeV, 1.0ltM3?lt1.6 GeV,
• P3?gt1 GeV, Pvtxgt1,
  -1.1ltcos??-a1 lt1.1

12
Summary of Signal-Side Selection
13
Signal-Side Selection Efficiency
Total signal-side efficiency for each selection
?ij is the efficiency of the selection i for the
MC ? decay mode j, fj BR(??j)
No systematic correction applied to efficiencies
listed in the table
14
Background Estimation from MC
No systematic correction is applied to MC. All
errors are statistical only.
15
Background Estimation from Eextra Side-band
The Eextra shape in the MC is used to extrapolate
the data side band to the signal region. RMC
NMC,SideB/NMC,SigR NExpData,SigR NData,SideB .
RMC
16
Validation of Eextra Distribution
Eextra simulation in MC effects both signal
efficiency and background estimation

• Use double-tagged events for validation of
  Eextra simulation in the signal MC
• Eextra distributions of double tagged events for
  data and MC are in reasonable agreement

• To validate background estimation from Eextra
  side-band extrapolation apply the method on
  various background control samples, such as,
  Events with two signal side track, non-zero net
  charge.
• Differences between expected and observed number
  of events are less than 1? for most of the cases.
  The largest deviation is within 2?.

17
Systematic Uncertainties

• Estimation of number of BB pair systematic
  error of 1.1
• Tagging efficiency determination
• Sources of uncertainty in signal efficiency
  determination
• Tacking efficiency -0.8 correction factor
  and 1.4 systematic error per track.
• Particle identification for the signal track(s)
• Obtained using the tables containing data-MC
  ratios of PID selector efficiencies or
  mis-identification rates.
• Custom PID tables are made take account of the
  correlation between PID selectors.
• Eextra simulation
• ?0 multiplicity Use official neutrals
  correction recipe
• Background estimation from Eextra simulation

18
Tagging Efficiency Systematics

• Use double-tag yield in data and MC
• Number of double tag events (N2) is related to
  the tag reconstructed efficiency (?) and total
  number of BB- events (N) as follows
• N2
  ?2N
• From double tag yield in data (407.0?20.2) and
  MC (434.4?12.4) obtain correction factor for
  tagging efficiency
• ?data/?MC 0.969 ? 0.029
• The correction factor in agreement with the one
  obtained from data-MC normalization difference.
• The 3.1 error obtained form double-tag method
  is used as the systematic error.
• ?tag (1.82 ? 0.074 ? 0.055)?10-3

19
Systematic Uncertainty from Eextra Modeling

• The multiplicity of low energy clusters in MC is
  higher than that is data.
• The difference in multiplicity between data and
  MC is about -0.46 for neutral clusters in the
  energy range of 20-30 MeV.
• For higher energy neutrals the difference in
  multiplicity is less than -0.20.
• From Eextra subtract 30 MeV in every two events
  and 80 MeV in every five events.
• Obtain correction factor and systematic error
  from the change in efficiency or background
  estimation caused by the shifts

20
Systematics (Cont)
Systematic correction for background
estimation e?? mode (1.02?0.04), ??? mode
(1.13?0.06) ?? mode
(1.12?0.03), ??? mode (1.09?0.04),
3?? mode (1.07?0.03)
21
Limit Setting Procedure (LEP Higgs method, used
in the B??? analysis using hadronic tags)

• Using a likelihood ratio estimator to combine
  different channels

• Statistical and systematic uncertainties on
  expected backgrounds are included in the
  likelihood definition by convoluting with a
  Gaussian G(bi,?bi), where bi is the expected
  background and ?bi is the uncertainty on
  background expectation.

• Branching fraction upper limit calculated by
  running toy MC for different branching fraction
  hypothesis.

• The confidence level (C.L.) for certain signal
  hypothesis is computed as

22
Nominal Upper Limit at 90 C.L.

• Upper limit is calculated for the case when
  observed number of events in data is equal to the
  expected number of background events
• Including modes with worse signal to background
  ratio actually slightly improves sensitivity.

BR(B???) (?10-4)
23
Physics Results
Branching fraction upper limit at 90 C.L.
BR(B???) lt 4.3 ? 10-4 Central Value
24
Eextra Distributions (Un-blind)
25
Combined Results

• The hadronic tag sample used for B??? search is
  statistically independent of the D0l-? tag
  sample.
• These two statistically independent samples are
  combined by first calculating the likelihood
  ratio estimator QL(sb)/L(b) for each sample.
• A combined likelihood ratio estimator is created
  by taking the product of the semiptonic (Qsl) and
  hadronic (Qhad) likelihood ratio estimator
  Qcomb Qsl ?Qhad
• Combined branching fractio upper limit at 90
  C.L.
• Central Value

26
Summary

• Search for B??? is performed in the recoil of
  exclusive semi-leptonic decay D0l? using Run 1-3
  dataset. Obtained branching fraction upper limit
  at 90 C.L.
• Combined with statistically independent hadronic
  tag sample, obtained upper limit at 90 C.L
• Preliminary result aiming for ICHEP, 2004
• Thanks to the review committee (Eli, Fabrizio,
  Dieter, Homer), AWG (Leptonic bc) and CW
  reviewers.
• Outlook and Plans
• Add Run 4 data.
• Cut optimization (Suggestion from the CWR)
• Improve background rejection, background
  estimation for ???? mode
• Publication