Minutes of the Meeting on Trigger Examine                  Thursday, May 2, 2002
--------------------------------------------------------------------------------
In Attendance: Jim Linnemann (via video), Leslie Groer, Ron Lipton, Dugan
O'neil, Michiel Sanders, Horst Wahl, Bob Kehoe, Roger Moore, Stefan Gruenendahl,
Joe Steele, Pushpa Bhat, and a few others.
--------------------------------------------------------------------------------

Pushpa briefly outlined the objectives for Trigger Examine and the approaches
one can take in developing the Trigger Examine (see slides).  There were
discussions on what needs to be monitored, whether it is useful to monitor
trigger rates, wheter useful information can be gleaned from Examines on L1, L2
and for that matter L3 since we can only look at events ultimately passed by L3
filters (and therefore do rejection rates make much sense) and how much of
hardware monitoring is necessary and whether it should belong to Trigger Examine
or detector examine.  What needs to be monitored, are, obviously, all the
quantities calculated and used at each Trigger level.  There were arguments that
special runs (such as mark and pass or with special triggers) as well as using
events that come on complementary triggers might help understand trigger
performance.  There was agreement that debugging and monitoring of trigger
hardware and comparisons of its performance relative to precision readout of the
corresponding detector would be extremely important.  The detector/trigger
groups have to determine whether such monitoring should be included as part of
the detector examine or continue to run them as stand-alone tools.  Global
Trigger Examine, it was agreed, should concentrate more on monitoring the
algorithmic performance.  Currently independent packages for L1cal and L1muo
(and L2gbl) exist as examine packages.  If necessary, these could be continued
to be used as independent diagnostic tools.

The other main point for dicussion was the approach to take for developing the
Trigger Examine.  Pushpa and Michiel described the approach with Physics Examine
i.e., running the Reco + RecoAnalyze online and using ROOT browser + macros to
make and display histograms.  In real time, the macros are being run on the ROOT
file that is being written to by the RecoAnalyze code.  Currently, this is
posing some problems w.r.t. locking of file at appropriate times and is causing
the ROOT macro client to crash when it tries to access the file coincident with
time it is being written to.  Many suggestions of how to get around this problem
were discussed and some will be tried.  There was a clear concensus that this
approach would be the most efficient to build Trigger Examine since TrigSim +
TrigAnalyze that already exist can be used to produce ROOTtuples with all the
information to be monitored.

As a first step, we discussed making an L3Examine expeditiously starting with
L3fAnalyze.  Michiel will set this up in a few days time.  Dugan O'neil and Joe
Steele have volunteered to help with L2 part.  Ron Lipton is going to find
someone to coordinate putting all the pieces together for a Global Trigger
Examine.