Bs signal selection

1
Bs signal selection

• Donatella Lucchesi
• University and INFN Padova
• for
• Bs Mixing group

CDF Collaboration Meeting June 7, 2006
2
The Blessed Selections hadronic decays
Signal Bs?Ds? Ds ???
Bs?Ds? Ds ??? ??KK Ds ?K0K K0 ?K? Ds
?3? Bs?Ds3? Ds ??? Ds ?K0K
Partially reconstructed B mesons
B0 ?D-?
Combinatorial background
3
Hadronic Bs decays summary
B ?D0? 26,000
B0 ?D-? 22,000
Decay channel Yield
Bs?Ds? Ds ??? 1587
Bs?Ds? Ds ?K0K 762
Bs?Ds? Ds ?3? 627
Bs?Ds3? Ds ??? 428
Bs?Ds3? Ds ?K0K 200
Total 3604
4
Blessed selection Bs semileptonic Decays
Bs?DslX Ds ??? ??KK Ds ?K0K K0 ?K? Ds
?3?
Total yield Dsl 37,000 events
5
What can we do to improve it?

• Include PID in the selection
• Use NN to discriminate signal from background
• Add new decay modes

6
Including PID in the selections
Bs?Ds? Ds ?K0K K0 ?K? has reflection from
B0?D? D ?Kpp when you give K mass to p
Standard analysis use D veto MKpp MDgt24
MeV
A. Heijboer
7
Including PID in the selections
Aart proposal MKpp MDgt18 MeV OR CCL?0
19.2?23.0 Give this to NN
A. Heijboer
J. Mulmestadt
Johannes idea
LR cut event by event increases Ds
16?40
In progress the Bs significance evaluation
8
Neural Network selections
B. Casal, K. Anikeev
Use ROOTSNNS v3.0

• Train NN to separate signal from bckg (side
  bands)
• Optimize to maximize
• Fit mass distribution using the actual mixing
  fitter

Rectangular cuts
NN cuts
Increase 33 like adding 78 of more data
9
Neural Network selections summary
B. Casal, K. Anikeev
57.78/43.301.3 improvement ?A
Use wrong charge events as background in the
future?
10
Partially Reconstructed B
Jeff Miles
B?D?, Dp ????0 , D?Dg/p0 treated as hadronic?

• Large yields (same order of fully reconstructed)
• High k-factor, lose soft g/p0
• Lower S/B

Depend on mass
Different K def. investigated finally use the
standard one
11
Partially Reconstructed B
Amanda Deisher
B0?D -D0? -?
5.00 ltmasslt5.35 GeV/c2
Low combinatorial background
ct 2 464.6 14.2 13.5 mm
Above 5.17 GeV Dp and DK
12
New Decay Modes
Simone Pagan Griso
Bs?Ds? Ds ?KsK Ks ??? has reflection from
B0?D? D ?3p as for Ds ?K0K K0 ?K?
Different topology easy to separate
270 events 1fb-1
Using PID
16 increase
13
Semileptonic decay modes
V. Tiwari, G. Gomez-Ceballos

• Improve event selection including PID Bs?Dslx
• Ds?fp combinatorial bck reduce by factor 2
• Ds ?KK yield increased by factor 2.5 and
• combinatorial bck reduced by factor 2
• Include lSVT dataset
• Remove the ct cut

Noticeable gain
14
Summary

• The number of events can be increased by
  including
• new decay modes, partially reconstructed events
• and by optimizing cuts.
• Plan for the summer
• Include partially reconstructed modes and
  optimize cuts.
• The actual gain on Dms significance will be the
• combination of all the effect.
• Expected improvements on sA
• 1.3 for the hadronic decay modes cuts
  optimization
• Something for semileptonic decays
• Under evaluation for adding partially
  reconstructed
• hadronic decay