BaBar Collaboration Meeting

1

• Search For B??? Decay
• With SemiExclusive Reconstruction
• C.Cartaro, G. De Nardo, F. Fabozzi, L. Lista
• University INFN Napoli

2
Outline

• Analysis progress (BAD 389 v.3)
• New mES fit
• Data/Background comparison
• Sideband sample
• Continuum MC (preliminary)
• Systematics upper limit extraction
• Further studies
• Neutral energy (preliminary)
• Conclusions

3
Data Sample B Counting

• 2002 data processed and included in the analysis
• mES distribution fitted as Argus (sideband) ?
  Xball (peak)
• 3rd polynomial no longer used (impact on B
  counting lt 0.6)
• B sample counting
• NB?B? (1.66?0.09)?105
• Using a Gaussian to fit the peak
• NB?B? reduced by 4.2

4
Data - MC Comparison
Data

• After preselection (request on of GTL tracks
  and ?0)
• Better data/MC agreement using data sideband
• Continuum MC study started
• not in BAD 389 v.3

1GTL 0?0
1GTL 0?0
5
Data Sideband VS MC
3-prong 3? invariant mass
1-prong neutral energy
6
Data/Bkg Estimate Comparison

• Two procedures
• Bin-by-bin approach
• Each variable range is binned
• In each bin
• fit the mES distribution
• estimate the peaking and non-peaking background
  components
• Single-bin approach
• The mES fit is performed on the full variable
  range

7
Bin-by-Bin Approach Example EMC Energy
Sideband / peak argus area ratio
Data
8
Bin-by-bin VS Single-bin
??????0? Neutral energy
???e??? missing momentum
?????????? 2? invariant mass
GeV
9
Systematic Errors

• Continuum and combinatorial background
• Extracted from data sideband 5.21lt mES lt 5.26
  GeV/c2
• Scaled to peak area mES gt 5.27 GeV/c2 after
  preselection using single-bin approach
• Systematic correction
• The ratio between number of events passing each
  cut in the bin-by-bin / single-bin is taken as
  correction
• Assumed the correction can be independently
  applied on each variable, the total correction is
  the product of the corrections
• On such correction we assume 100 of systematic
  error
• Other systematic errors
• NB?B? reduced by 4.2 using Gaussian instead of a
  Xball to fit the mES peak
• Efficiencies not yet included but most of the
  code is available

10
Background
B? ? ? ? ?
B0 ? ? ? ?
NEW!
B? ? ? ? ?
11
Upper Limit Extraction
-2 Log Q
B ( B? ? ? ? ? ) lt 2.98 ? 10 ? 4
B ( B? ? ? ? ? ) lt 3.91 ? 10 ? 4
Dashed curve
Solid curve
We are blind on B???? ? but we can assume that
the number of observed candidates is equal to the
nearest integer to the expected background
prediction
We also include statistical and systematic
uncertainty on the background estimate
Notice More BB? MC may reduce the statistical
uncertainty on the background and improve the
upper limit!
Assuming gaussian instead of Xball peak, of
BB? reduces by 4.2 B ( B? ? ? ? ? ) lt 4.08 ?
10 ? 4
12
Upper Limit Distribution

• Generating data samples according to a Poisson
  distribution with the background prediction we
  get a distribution for the upper limit we can
  expect

• Mean value 3.6?10?4
• Most probable value 2.5?10?4

BR (10?4)
13
Further Studies on EMC Energy

• The worst data/MC disagreement is the neutral
  energy in ??????????
• Bin-by-bin approach doesnt help in this case

14
Improving Neutrals Quality

• Remove fake photons using Vub recipe to get a
  better Data/MC agreement
• EMC acceptance
• 0.41 lt ?EMC lt 2.54
• splitoff removal
• 0.05 lt LAT lt 0.5
• ?9/?25 gt 0.9

See for details http//www.slac.stanford.edu/
BFROOT/ www/Organization/CollabMtgs/200
2/detSep2002/Tues4d/ric.pdf
15
Neutral Energy Improvement

• Applied cuts
• ?EMC
• splitoff rm

• Applied cuts
• LAT
• ?9/?25

• Applied cuts
• LAT
• ?9/?25
• ?EMC
• splitoff rm

??????????
The variable is now less discriminating
16
Impact on U.L. Extraction

• Neutral energy with cuts on LAT and ?9/?25 (the
  most efficient for the 3-prong mode) modifies the
  expected background
• Increasing the expected background, the upper
  limit will increase too
• ?U.L. 0.35 ? 10 ? 4 -absolute-

B?B? continuum combinatorial Total Exp. Bkg
Raw Neutral Energy 0.8 0.6 1.4 ? 0.9
Improved Quality of Neutral Energy 1.3 1.2 2.5 ? 1.1
Not in BAD v.3
17
Conclusions

• The analysis is at good stage
• 2002 data processed and inserted in the analysis
• Few bugs fixed (background double-counting, ...)
• Good agreement Data/MC
• Tested using wrong sign sample (total event
  charge ? 0)
• Studied neutral energy quality
• Studied systematic effects
• Background estimate
• B counting
• Efficiencies not yet included completed
• Channel ??????0 ?0 ? doesn't improve the u.l.

18
Backup Slides

• Some more details.

19
B??? Decay
l
u
B?? ?
n

• Allows the measurement of fBVub
• Determination of fB given Vub (from independent
  measures)
• Standard Model Br O(10-5-10-4 ) for the ?
  channel
• Helicity suppression in the ? and e channels
• Br O(10-7) for muons and O(10-12) for electrons
• The best upper limits
• L3 Br( B ? tn) lt 5.7 ? 10-4
• P.L. B396 327 (1997)
• CLEO Br( B ? tn) lt 8.4 ? 10-4 (_at_ Y(4S)
  resonance)
• P.R.L. 86 2950 (2001)

20
SemiExclusive Reconstruction

• Starting idea
• To reconstruct as much as possible B mesons
  decays in the final states
• B? D0() n1p n2K n3 Ks n4p0
• where n11...5, n20...2, n3 0,1, n4 0,1
• The reconstruction ignores the intermediate
  resonances, for example B?DDs (Ds ?Fp) is
  reconstructed as B?DKKp
• All remaining tracks and neutral objects are
  associate to the recoiling B on which we look for
  B???
• Pro strong continuum and combinatorial
  background suppression
• Cons sensible reduction of the sample
  (?breco?0.2)

21
Data Sample B Counting

• Monte Carlo (SP4, Run-1 and Run-2 conditions)
• Signal43000 events B ? ?? B ? D0()?
• 74000 events B ? ?? B ? D0()a1/?
• Background72 106 generic BB?
• 67 106 generic B0B0
• 107106 continuum
• Continuum and combinatorial
• background also evaluated from
• data sideband
• Data
• 82 fb-1 from Run-1 and Run-2

Fit Argus(continuum) ? Crystal ball (peak) 3rd
polynomial no longer applied with the Argus fit
?

• mES (Ebeam-p2B)-1/2
• mES peak area is our B sample and is used as
  denominator for limit extraction
• it is not necessary to estimate the
  semiexclusive reconstruction efficiency

22
Event Preselection

• Channels under study
• ???e??? Br 18.0
• ??????? Br 17.5
• ?????? Br 11.1
• ??????0? Br 25.2
• ?????????? Br 8.95
• ??????0 ?0 ? Br 8.95
• Maintained but not included

Put new values!!!!!
PDG 2002
Data
1GTL 0?0

• Preselections
• 1 GTL 0 ?0
• 1 GTL 1 ? 0 or 2 ? 0
• 3 GTL 0 ? 0
• Fit to the mES distributions
• Crystal Ball Argus
• Fits after the preselection used for
• data-MC comparison
• expected background prediction

Generic BB
1GTL 0?0
23
Cut Optimization

• An automated procedure based on a scanning of the
  possible cut values has been developed
• determine the selection that minimizes the
  expected upper limit
• For each variable in a given selection
• Choose a possible cut
• Redo the selection and the toy MC for the u.l.
  extraction
• Take the cut value giving the lowest expected
  upper limit

24
Event Selection

• 1-prong events
• 1 track, 0 ?0
• No KS
• Neutral Energy lt110MeV
• Neutral Bumps lt 1
• Pmissing gt 1.2 GeV/c
• Kaon veto
• Particle ID
• ?????? only
• Lepton veto
• pc.m.s. gt 1.2 GeV/c
• SemiExcl purity mode gt 50

• ?????????? events
• decay proceeds via two intermediate resonances,
  an a1 and a ?
• 3 tracks and 0 ?0
• Pmissing gt 1.2 GeV/c
• Neutral Energy lt 100 MeV
• Neutral Bumps lt 1.5
• 600 MeV lt m(???0) lt 950 MeV
• 1.1 GeV lt m(?????) lt 1.6 GeV
• p1 p2 p3 gt 1.6 GeV/c (c.m.s.)
• Lepton kaon veto
• SemiExcl purity mode gt 40

• ??????0? events
• decay proceeds via an intermediate ?
• 1 track and 1 ?0
• pmissinggt 1.4 GeV/c
• Neutral Energy lt 100MeV
• 0.55 GeV lt m(???0) lt 1 GeV
• SemiExcl purity mode gt 50

25
Samples and Efficiency
Total Charge 3 prong events

• The total event charge is left as last cut in
  order to define two samples
• Q 0 signal sample
• Q ? 0 control sample B? ? ??

• Total efficiency on signal (MC) 13.0

channel
e??? ???? ??? ???0? ??????? ???0?0? ?????? ?0?
???e??? 25.9 0 0.8 0 0 0 0
??????? 0 9.5 0.6 0 0 0 0
?????? 0.2 2.5 24.8 1.3 0.3 0.1 0.1
??????0? 1.0 1.5 1.5 11.4 0.3 1.5 0.7
?????????? 0 0 0 0 5.0 0 0
selected
26
Background

• Continuum and combinatorial background
• Extracted from data sideband 5.21lt mES lt 5.26
  GeV/c2
• Scaled to peak area mES gt 5.27 GeV/c2 after
  preselection
• Peaking background
• Estimated from generic BB? Monte Carlo
• To avoid double counting of the combinatorial
  background the mES sideband in generic BB? MC is
  scaled and subtracted to the component in the
  signal region
• Scaled to the data peak area

???
e???
????
???0?
???????
27
Statistical Technique

• Using LEP technique to combine more channels
• Likelihood ratio estimator

• This definition can be extended to include PDFs

A Toy MC determines the log(Q) distribution for
different expected values of s to be compared
with the value obtained from Q data sample
The BF estimate is obtained in corrispondence of
the minimum in the log(Q) distribution if exists