Limiting Background by Tagging with Exclusive D0l Search for the decay B

1
Limiting Background by Tagging with Exclusive
D0l? (Search for the decay B???)

• Mousumi Datta
• University of Wisconsin, Madison
• Yibin Pan, Sau Lan Wu
• BABAR Collaboration Meeting
• Recoil Method Physics Parallel Session
• Dec. 10, 2003

2
Outline

• Motivation
• Dl?X tag VS D0l? tag
• D0l? tag selection
• Tagging efficiency (on B??? VS generic MC)
• Search for the decay B???
• Signal selection efficiency and background
  estimation
• Expected upper limit on signal
• Summary and outlook

3
Motivation and Overview
B-?X

• Due to presence of multiple neutrinos in the
  final state B??? decay has very little
  experimental constraint ? Use recoil method
  analysis
• After identifying the particles from the tag B,
  in the remaining of the event look for the
  signature of ? decay ? No extra particles except
  ? decay products and very little extra neutral
  energy in the calorimeter.

4
Existing Semi-leptonic B? Tags
Semi-leptonic B? sample
Tagging efficiency For B?K?? analysis (exclude
Ks?? mode) (5.25?0.14)?10-3 For B??? analysis
(including Qnet0 cut) 4.45?10-3
B- g D0 l- n X0 K- p
K- p p0
K- p p- p K0s p p-
X0 system nothing or ? or ?0 candidates from
possible D()0 ? D0 ?/ ?0 transitions. X0 is not
reconstructed

• For B??? decay only the leptonic ? decay modes
  are looked at
• Relatively high background level for hadronic ?
  decay modes

Currently used for B??? and B?K?? analysis
5
Exclusive D0l? Tags

• Cleaner subset of Dl?X tags
• Additional constraint from D0 reconstruction ,
  such as ?M(D0-D0), provide further background
  suppression
• The recoil side is cleaner compared to the recoil
  side of Dl?X tags
• For Dl?X tags the recoil side can contain
  neutrals from D0 decay
• For D0l? tags all the tag side neutrals are
  reconstructed and removed from consideration. Get
  better separation between signal and background.

Use D0l? tags to include hadronic ? decay modes.
Also provides a better signal to background ratio
for leptonic ? decay modes.
6
D0l? Selection

• Use DstarlnuUser package in release 12.5.2b
• Currently using 1999-2001 onpeak data and
  following MC samples

SP5 MC
7
Beta-level Selection Summary
ISL filter with minimum P of lepton 1 GeV
of charged tracks lt10
D0 D0 mass window 40 MeV for charged-modes 70
MeV for K??0 Vextex convergence
D0 D0-D0 mass difference 0.12 - 0.17 GeV for
D0 ?
0.13 - 0.17 GeV for D0 ?0 ?0 from D0
pi0SoftDefaultMass list (0.115lt ?0 Mass lt0.15
GeV) ? from D0 GoodPhotonDefault list
(cluster energygt100 MeV) 0.5 lt P of D0
candidate lt 2.5 GeV
D0l Lepton P gt 1 GeV passing Tight
electron/muon selection Vertex convergence for
D0l
8
Root-level Tag Selection

• Reject D0l candidates with lepton passing both
  Tight electron and Tight muon ID
• PID requirement for D0 daughter tracks
• Pions fail Tight electron/muon selection and
  KNNVeryLoose
• Kaons fail Tight electron/muon selection, pass
  KNNVeryLoose
• 0.01 lt LAT lt0.6 for neutrals from D0 and D0
• Cos?(D0-l) lt 0.0
• -1.1 lt Cos?(B-D0l) lt 1.1
• After applying above selection cuts select best
  candidate using
• D0 mass and D0-D0 mass difference
• Additional tag side cuts after selecting best
  candidate
• 0.135 GeV lt (D0-D0 mass difference) lt 0.15 GeV
• D0-?0 CMS angle lt 45 deg, D0-? CMS angle lt 90
  deg
• Total event charge 0

9
Tag Efficiency B??? VS Generic Sp4 MC
Purity After Best Candidate selection D0?0
62.57 D0? 48.57 After Qnet Cut D0?0
79.22 D0? 75
Select Best Candidate
10
Data-MC comparison after Beta-level selection
11
Data-MC comparison after best candidate
selection The difference in normalization mainly
coming from PID requirement for the D0 daughter
tracks and the lepton
12
of signal track
of signal ?0
Eextra (GeV) at Lab frame
Distribution of signal side variables after best
candidate selection
13
Signal Selection for l??, ??, ??0?

• One remaining track
• For Signal track with Plabgt0.6 GeV XY-DOCAlt1.0 cm
  

• e??
• Track passes VeryTight electron selection
• Track fails Tight muon selection and
  KNNVeryLoose kaon selection
• No signal side ?0 with mass between 0.115-0.15
  GeV
• Eremaining lt 0.3 GeV

• ???
• Track passes VeryTight muon selection
• Track fails Tight electron selection and
  KNNVeryLoose kaon selection
• No signal side ?0 with mass between 0.115-0.15
  GeV
• Eremaining lt 0.3 GeV

14
Signal Selection for l??, ??, ??0? (cont)

• ??0?
• Track fails Tight electron/muon selection and
  KNNVeryLoose kaon selection
• One/Two signal side ?0 with mass between
  0.10-0.16 GeV
• 0.55 GeV lt ??0 invariant mass lt 1 GeV
• Eremaining lt 0.3 GeV
• P of ?? lt 2.7 GeV
• -1.1 lt cos(?-??) lt 1.1

• ??
• Track fails Tight electron/muon selection and
  KNNVeryLoose kaon selection
• No signal side ?0 with mass between 0.10-0.16
  GeV
• P of signal track lt 2.7 GeV
• Eremaining lt 0.3 GeV

15
Signal Selection for 3??

• Three remaining charged tracks
• For each signal track with Plabgt0.6 GeV
  XY-DOCAlt1.0 cm
• All three tracks fail Tight electron/muon
  selection and KNNVeryLoose kaon selection
• No signal side ?0 with mass between 0.115-0.15
  GeV
• 0.55 GeV lt ?0 Mass lt 1 GeV
• P of ?0 gt 0.5 GeV
• 1.0 GeV lt a1 Mass lt 1.6 GeV
• P of a1 gt 1.0 GeV
• Vertex probability of three tracks gt 1
• Eremaining lt 0.3 GeV

P of a1? lt 2.7 GeV -1.1 lt cos(?-a1?) lt 1.1
16
Signal Selection Efficiency
j ? decay mode index, fj BF(??j)
17
Signal Efficiency Background Estimation
18
Expected Upper limit
Use statistical procedure used for Higgs search
in LEP. This procedure is also used in B???
analysis using hardonic tags
Without statistical/systematic error BR(B???) lt
2.18?10-4 at 90 CL
Including only statistical error BR(B???) lt
2.64?10-4 at 90 CL
19
Summary and Outlook

• Using D0l? tag can include less clean ? decay
  modes for B??? search
• Loss in tagging efficiency can be compensated by
  including more ? decay modes
• Goal is to combine the leptonic ? decay mode for
  Dl?X tag and D0l? tag ? get a better sensitivity
• D0l? tags can be used if analysis requires
  cleaner tag than Dl?X
• Example for B???, ????0? signal mode, if Dl?X
  tag is used than there is a possibility of
  selecting a tag side ?0 as the ? daughter
• Currently can use low track multiplicity cut ?
  helps reduce combinatorics (and ntuple size).
• If no such requirement is used then the
  combinatorics can be large (has not yet been
  checked)

20
Backup Slides
21
Double Tags at the Best Candidate Selection
Stage
22
Statistical Procedure for Combined Upper Limit

• Use statistical procedure used for Higgs search
  in LEP. This procedure is also used in B???
  analysis using hardonic tags

Likelihood Ratio estimator
Generate 5000 toy MC experiments for each
BR(B???) values (0.025-10)?10-4 in steps of
0.025?10-4