HLT Studies

1
Muon PRS Group Darin Acosta (Muon
Guy) University of Florida
2
Whos Involved?

• In Europe
• Ugo Gasparini et al. Generator studies, L2
• Norbert Neumeister DB prodn, L1, L2
• Anna Vitelli DB prodn, barrel muon
  reconstruction
• Several others with lesser involvement
• In the U.S. (Endcap Muon group)
• D. Acosta, S.M. Wang (UFlorida) L1
• R. Breedon, T. Cox (UCD) geometry, endcap
  recon.
• B. Tannenbaum (UCLA) L1
• R. Wilkinson (Caltech) endcap recon., general
  guru

3
HLT Milestones

• November 1999
• EMU missed completely. Insufficient time to
  develop code for geometry, digitization,
  persistence, and L1 from scratch in 6 months
• July 2000
• EMU sort of met it. All pieces were in place,
  and preliminary results shown, but not enough
  time for tuning (PT assignment), debugging, and
  optimization (code was too slow for out-of-time
  pile-up).
• September 2000
• Close to having all above issues resolved for
  dedicated L1 studies for the L1 TDR (due by Nov.)
• But, goal line keeps moving more demands on L1
  trigger threshold definition (90 effic. vs. 50)

4
U.S. Experience (1)

• EMU group works well together
• Lots of e-mail exchanges, video conferences, and
  meetings to keep everyone informed and on-track
• A real user group where you can post a message
  (even dumb ones) and get help
• EMU code development for ORCA is entirely a U.S.
  effort, with only minor consultation from Anna,
  Norbert, Vincenzo,
• We are heartened by Sarahs comments
• US muon has this ORCA experience, which is
  one of the reasons I think they are so
  successful!
• Because

5
U.S. Experience (2)

• This development has been painful
• Large C/ORCA/SCRAM learning curve to climb
  (makes it hard for newcomers to contribute)
• ORCA is a moving target
• Version doesnt stay stable long enough to
  finish bug fixes, tuning, and collect results.
• e.g. ORCA3 ? ORCA4 was a major switch after the
  November 1999 HLT meeting. It took many months
  before basic code was ready, which left little
  time for L1 developers to complete their work.
  We still had to rush for the July HLT meeting
  despite it being 8 months later!
• ORCA 4.2 ? ORCA 4.3 has also been frustrating.
  Objectivity schema changed after July meeting,
  and it took a while before it was made backward
  compatible with previous DBs so that studies
  could continue to prepare for L1 TDR studies
• Clearly development must occur, but it would
  help users if the underlying architecture would
  stay stable

6
U.S. Experience (3)

• Some of this pain stems from none of the EMU
  developers being based at CERN
• Attending the weekly RPROM meeting is not
  enough. Must have a door to knock on to get
  advice.
• Long distance code development is slow, but most
  of us have no other choice.
• Teaching duties, other active experiments, etc.
• Thus, EMU is not 100 ready at time of
  milestones
• Preliminary results shown by European colleagues
  (using our code) mislead CMS Collaboration
• i.e. Poor L1 efficiency in DT/CSC overlap, poor
  L1 PT resolution lots of bad PR that
  demoralizes group
• Having a larger U.S. base of users and
  developers would improve efficiency of ORCA
  development

7
Recent Results
Nevertheless, we have obtained some results on L1
performance
95
Good efficiency in DT/CSC overlap region
8
More Work Needed
L1 PT resolution shows difference between ? and
?-
?
?-
9
Near-Term L1 HLT Goals

• Must get Endcap Muon code perfected so that we
  can prepare efficiency and rate plots for the L1
  TDR that are representative of the system
  performance
• Last few bugs must be fixed
• HLT Muon databases will be re-made with latest
  code
• No new MC needs to be produced in U.S. right now
• Caltech, UCD, UFlorida have contributed CMSIM
  samples in the past
• Objectivity database formatting Ntuple
  production done at CERN without U.S. EMU
  involvement
• Norbert and Anna have done a lot of work on
  this, and have prepared the Ntuples used by the
  Muon groups. We should be doing more of this,
  particularly in preparing standard analysis jobs
  and Ntuples

10
Long-Term Plans

• It would be nice to get the underlying EMU and
  L1 code finished so that we can move on to
  physics and HLT algorithms!
• Hasnt happened in parallel because the EMU
  users are also the developers
• My personal selection of topics
• A fast L2 muon reconstruction package
• Between that used by L1 and that used in offline
  analysis
• A study of the reconstruction of muons in b-jets
• Many important signals couple to bs Higgs,
  SUSY,
• Accompanying punch-through debris will
  deteriorate performance
• Studies of your favorite physics topic that
  couples to muons
• Higgs H?4l, H?????, WH ???bb, ZH ???bb
• SUSY ??0 ? 3l MET

11
Summary

• Clearly the Muon PRS group could use more
  involvement from people in the U.S.
• Current members are over-stressed with demands
• Many more things we could be doing but havent
• L2 algorithms, physics studies,
• Might be a good training ground for new students
  before they commit to an analysis on CDF/D0
• But there is a significant learning curve to
  overcome to become effective
• Requires a lot of time, a lot of patience
• Weve started a good collaboration in the U.S.
• Future members shouldnt have to suffer as much
  as we did!