DØ Level 3 Algorithms Meeting:
20th March 2002 in the Farside
Talks, verbal reports
-
Report on 20th March trigger board meeting: Terry Wyatt.
-
Update on electron rejection factors as a result of looser shower
shape cuts: Ulla Blumenschein.
-
Update on p11.02 certification and leak finding for muons:
Martijn Mulders.
Memory leak in muon geometry is near to being fixed.
Should we ask for a pass release of p11.02 or should we push for
p11.04 to be made asap? Jon will discuss with the release managers.
P_t resolution for
loose muons is OK, although not quite as good as for tight muons.
Still no definitive understanding of why more than 50% of loose muons
in central region fail chi^2 cut.
Are BC-only segments really being found? All muon tracks that Martijn
sees have non-zero number of A-layer hits (scintillators/chambers?).
A status report on L3 muons has been promised for next week's trigger board.
-
Status of L3 vertex efforts: Per Jonsson.
Per showed
results of recent work at last week's b-id meeting.
In the short term what we really need is a primary vertex finder with
resolution at the few cm level.
Work on track-based primary vertex finding will be presented by Per
and/or Jon next week. It would be great if we could compare with the
hit-based primary vertex tool.
Discussion items introduced by Jon Hays
-
Centrally run jobs for "routine" checks of L3 releases.
Jon will put together, in consultation with the tool authors, a list of
certification files that should be run through l3fanalyze every time
there is a new release. We clearly save effort if one set of jobs is
run rather than each tool author duplicating this work. In the
short term each tool author will be responsible for making their
standard checks on the resulting rootuples.
In the longer term it would be highly desirable to put together a
standard set of macros/histograms that could be run centrally. These
will one day form the set of L3 histograms that will be examined
online by the shift crew. Jon will get this going on p11.03, with the
hope that by the time p11.04 is available we can get fast turn-around.
-
Calibration/alignment issues:
-
Making sure the correct calibration/alignment is used online
-
Keeping track of which calibration/alignment versions are used online for
which runs
This is a problem not just for L3 and has been discussed at several
special meetings.
One solution being considered is that COMICS will be responsible for
archiving the configuration downloaded at run start-up.
The rungrabber will then store this information in the runs database.
At the moment L3 doesn't fit into this picture, because we don't do a
COMICS download. Should this change? At the moment each L3 system
uses a different method of getting its calibration/alignment
information (if at all).
From a L3 point of view there are probably three paths we can go:
i) instrument our nodes/executables so that they understand how to talk
to COMICS so that we can send out configuration details : release string,
calibration version, etc.
ii) instrument our nodes/executables so that they can update the
Name-value server with the current versions of software etc.
iii) every time the configuration is updated produce a simple text file
containing name-value pairs for the various configuration parameters and
have that copied to the directory where the rungrabber looks for files.
i) and ii) have the advantage that it is slightly more automated and
error-proof than iii). iii) has the advantage of simplicity.
-
Related issue: how detailed does the calibration/alignment information
need to be for L3. Do we need to have the full offline quality
information or is some simpler approximation good enough? We should
think about this for each major detector component. In particular, we
hope to have presentation from Daniel and/or Robert at next week's
meeting on tracking detectors.
Scribe: Terry Wyatt.