Core Minutes 11/29/2005
Present: Joanne Bogart, Toby Burnett, David Chamont, Jim Chiang,
Seth Digel, Richard Dubois, Warren Focke, Riccardo Giannitrapani,
Tom Glanzman, Navid Golpayegani, Tony Johnson,
Heather Kelly, David Landriu, Chuck Patterson,
Igor Pavlin, Leon Rochester,
Tracy Usher
- Data Handling: (Tony)
-
Karen has been working on improvements to the Pipeline interface; Igor is
making data portal improvements.
- We're getting 10 new Linux servers: 4 gig, 4-processor systems which
should be more power than we need for a while. Will provide aliases so
users can refer to them by function rather than by node name. Then we
can implement load-balancing and fail-over transparently. There
will be a Confluence page in which plans for use of these servers
will be discussed. (Richard) We
also have 3 Windows Xenon servers which are about to be configured.
- (Richard) astroroot is a package which converts FITS data to ROOT, but
it only works on Linux. Could it be one of the functions of the Data Server
to use astroroot to make ROOT files? (Tony) The plan has been for the Data
Server to make either format available on the fly one way or another. astroroot
might (or might not) be used to do this.
GR tagging proposal: (Toby)
See the new GR tag thread in the Infrasoft mailing list. It suggests we move to
a scheme where in the future GlastRelease tags made on the MAIN branch would
all be of the form vNrM. For each such tag, a branch tag would automatically
be made. Any patches to the release would be made along such a branch.
Toby will make—scratch that; has made—a
Confluence
page in the Core area for continued discussion,
including use cases. (Navid) Impact on RM looks ok, but we need to be
careful about choice of branch names. Certain forms of branch tag will
confuse CMT.
- CalRecon: (David C.) Most of the planned
work is done and committed. He still wants to look at all packages
connected with ROOT storage: RootIo,
RootConvert, RootCnvSvc,
mcRootData,
digiRootData,... and associated unit tests.
For development he's been using ROOT 5 (in such a way that the old
dictionary scheme is the one being used) and it appears to behave
correctly with no special action on his part. Note this was on Linux
only; haven't tried on Windows.
- Testbeam support:
(Richard) He has asked
Leon to coordinate this effort. (Leon) One significant difference
from our usual style of event is that there may be more than one
primary particle. For simulation, plan to do a stand-alone G4 run
for particle generation. G4 has a native ascii output format, but we
would want the output to be in ROOT. Then it could be fed to Gleam
to propagate through the detector (Calibration Unit). There seems to
be no obvious difficulties with this approach. Generally speaking we
have preserved the ability to trace back to the correct (if more than
one) primary particle. Heather and Tracy concur.
(Richard) What about merit,
which implicitly
assumes a single primary particle? (Leon) Some adjustment will have to
be made; there will be additional information not captured by the
particle list. Perhaps for merit we should
declare the highest-energy primary particle to be the primary
particle.
(Richard) Another test beam issue is the upstream stuff which
needs to be modelled. He suggests we start a Confluence page for
discussion of all these (Testbeam software) issues.
Time scale is short: plan is due
mid– to late January.
- DC2 prep status:
- Classification (Richard, Tracy)
For now, Tracy has been using Toby's
scheme in a post-processing step. It also applies Bill's filter
cuts. TBD: output variables from this process. (Jim) Is there a
Confluence page or other documentation of just what the filter cuts are?
(Tracy) Currently, no. Best to just ask Bill. He is busy just now with
a meeting, but should be free soon.
- Sky model (Richard) made a test task, using Science Tools
checkout 3 source list. There are some issues having to do with handling
of time. Also the GRB source needs work (Johann has taken this on) and
the sky "looks funny". (Seth) Indeed it does: single-source runs end up
producing a blob. This could have to do with (improper) use of the
pointing history. (Toby) The two-step procedure whose purpose was to
allow us to reliably regenerate random sequences, hence events,
doesn't preserve sky orientation. We don't need the two steps for these
runs anyway. However we will need the two steps for background interleaving.
- Announcements: (Richard)
- Another gamma ray conference in
this
series will be taking place at a TBD location in the US in February, 2007, concentrating on GLAST.
- The Kavli Building will open in January or February of 2006; Kavli
people will have moved in by April. This will free up some space
for GLAST people in
Building 84 where several of us, including Richard, already are.
J. Bogart, Last Modified:
01-Jun-2010 15:47:32 -0700