Core Minutes 12/13/2005
Present: Joanne Bogart, Philippe Bruel,
David Chamont, Jim Chiang,
Seth Digel, Richard Dubois, Warren Focke,
Riccardo Giannitrapani,
Tom Glanzman, Navid Golpayegani,
Heather Kelly, David Landriu, Julie McEnery, Chuck Patterson,
Igor Pavlin, Leon Rochester, Tom Stephens,
Tracy Usher
- DC2:
- Reprocessing (Richard, Tom G.) We're getting ready to
reprocess the billion-event background events, half of
them coming from Lyon. The code includes Leon's latest
improvements. 6 of the
20,000 runs failed. Reprocessing should only fail if the
ROOT input files weren't created (same whether run was done
at SLAC or Lyon). One simple check we can make is to make
sure we have the same number of events after reprocessing.
Tom has written a script to ensure that
we don't exceed the 1000-jobs-per-user limit.
- Sky model testing (Richard) Looks good so far. The pieces
have been tested individually, but more testing has to be done
on the integrated object.
More (Richard) We need a step to create FITS files:
level one photons and pointing and livetime. How much data do
we want? Original goal was a day, when a day meant 1 billion events,
but now it means 4 billion.
Variable documentation:
(Leon) New CT variables from Bill aren't documented.
(Heather) Would like to see types included in n-tuple doc.
Decision was made to add a column for this [and as of this writing
it appears Leon has already done so].
- Beamtest status: (Leon) Mission not quite
accomplished:
- Some code assumes origin is between towers, but for this geometry
y=0 is in the middle of a tower.
- With current geometry arrangement can't move ACD tiles everywhere
we would like to. This may be due to the way volumes are nested, a scheme
appropriate for the full geometry but not necessarily what we want here.
- Visualization is coming along nicely; will not be a show-stopper.
(Richard) Still to go: dealing with ancillary upstream stuff; figuring
out how to get data from CERN to SLAC.
Benoit is about to import a new app for the SPS source into the
beamtest06 package.
- I&T matters:
(Joanne) Revised ACD geometry with overlapping of top tiles in
correct direction—required to make sense of data taken with
ACD—has been committed and tagged for a while; no known
problems with it. (Heather) is about to incorporate it into new EM
tag, mostly based on v7r3p3. [She has gone with xmlGeoDbs v1r26, which
includes the needed fix, but does not include any of the recent updates
in support of the Beam test.]
(Warren) There has been a delay in getting ACD hooked up having to
do with cooling, so there will be no data until after Xmas.
- OB: It was so early when the hot topics
above had been covered that when Richard asked if there was any other
business, there was:
- (Tracy) The discrepancy Julie saw in trigger status bits
between runs of p3 and p4 has been traced to a problem with one
of Bill's new variables, which
occasionally (523 instances in a run of 40,000 events) takes
on an illegal value, thereby disqualifying the event for some
purposes. See a fuller discussion in the
LAT
Performance Analysis mailing list.
- (Julie) Convention in FITS files for numbering [Tracker] layers
doesn't match what's done elsewhere. (Jim) The plan
is to just refer to front and back layers in FITS, so this won't
be an issue.
- (Jim) IRF functions look ratty. Any suggestions on how to improve
things are most welcome.
J. Bogart, Last Modified:
01-Jun-2010 15:47:36 -0700