Strategy for single top analysis in the e jets channel

1
Strategy for single top analysisin the e jets
channel
B.Andrieu, E. Busato (LPNHE, Paris) special
thanks to M. Kado (LAL, Orsay)
2
General remarks

• Signal vs background
• Low cross section signal (handful of events) ?
  Maximize selection efficiency
• High cross section background ? After e jets
  preselection, first use most efficient selection
  criteria (b-quark jet tagging) before
  topological cuts
• Background not precisely predicted by theory and
  Monte Carlo simulation? Avoid restricting phase
  space too early in order to keep handle on
  background
• Selection of b-quark jets
• No 100 efficient, background-free, b-quark jet
  tagging method ? Use (as many as possible)
  different subchannels (tagging methods, number of
  tags, )
• But these different subchannels are based on the
  same sample ? Use exclusive samples to be able
  to combine independent limits in the most
  efficient and simple way

3
Strategy for b-quark jet tagging

• FIRST use Soft Muon Tagging ? Select events with
  gt 1 SMT jet
• Well-established, simple and robust b-quark
  tagging method
• Experience from first-pass p13 analysis
• Possible gain on trigger efficiency using e-mu
  trigger jointly withe jets trigger
• If possible, separate again according to 2nd jet
  tagging
• gt 2 SMT jets
• Exactly 1 SMT jet gt 1 JLIP-tagged jet
• Exactly 1 SMT jet 0 JLIP-tagged jet
• THEN use a second tagger (JLIP) ? Select events
  with gt 1 JLIP jet SMT veto
• Use only e jets trigger for this channel
• If possible, separate again according to 2nd jet
  tagging (JLIP)

4
Final selection?

• Topological cuts?
• Optimize cuts to obtain the best limit in each
  channel separately or use the same cuts for all
  channels?
• Combine statistically independent limits from
  different channels
• Knowing the efficiencies of topological cuts on
  signal and background,is it possible to use a
  Matrix Method to estimate the contribution of
  signal and background for each channel?
• OR Combine the variables with the highest signal
  vs background separating power into a likelihood
  ratio?
• Estimate a limit for each channel and combine
  these statisticallyindependent limits?
• OR combine all channels using the
  likelihood-ratio distribution andderive a single
  limit for the e jets channel?

5
Summary

• First attempt to define a strategy for e jets
  analysis
• Combination of different tagging methods
  relatively clear
• Strategy for final selection and result still to
  be discussed
• Eventually, plans with evolve as analysis
  progresses and time schedule becomes tighter!