Minutes of Fermi/GLAST Core Software Meeting Tuesday 18 August 2009
Present: J.Chiang, J.Vernaleo, C.Patterson, E.Winter, H.Kelly,
R.Dubois, K.Lo, A.Borgland, L.Rochester, N.Golpayegani, T.Burnett,
W.Focke, T.Glanzman
ScienceTools
Still running v9r15p4. Report in confluence
FSSC
EricW: Science Tools v9r15p2 stable, all testing so far has passed.
Still ingesting data...
John: Ingesting last 3 weeks of backfill reprocessed data. About 95%
of photon data already ingested.
Public data release still scheduled for Monday 24 August 2009.
Documentation
from Chuck: Still working on the pyLikelihood Tutorial, and hit a
rough spot with the BinnedAnalysis. Currently, I'm rerunning the
prerequisite Science Tools, using 6mos of data instead of one day in
an effort to get an acceptable fit. Liz Hays is trying to get binned
analysis working with the FSSC tools, and we're exchanging
information. She suggested that I compute my exposure map cube using
bin energy EDGE instead of CENTER, and, while it resolved a problem I
was experiencing, it did not improve the fit. Liz also reports that
the BinnedAnalysis does not perform well if the bins are starved for
data.
(Hopes to complete this tutorial later this week.)
Richard: iLat data access broke due to building in the (old) line-mode
astroserver client. Need to track down Gino/Andrea to update.
GR Status
Richard: Pass6 ToT bug fix reprocessing had a problem. (This is a test
of ~1 months data to verify the effect of the ToT fix prior to public
data release.)
Leon: v15r47p15 bad version of classifyAlg, no updated CTB variables
were written out. Working on two possible solutions. One way or
another we must recreate the MERIT files for the 500 runs constituting
this test.
Anders: prefer rebuilding GR with fix and reprocess using the same
mechanism as the original P90-MERIT task.
—
Richard: A topic has been raised amongst the Poobahs: analyzing the
effect of various types of LAT failures and consequent impact on
operation, processing and science. Will be session at upcoming
collaboration meeting to discuss further. If/when a failure occurs,
will need to perform simulations accounting for the crippled part(s)
of the LAT. Need to get tools ready to perform sims quickly, if
needed.
Anders: Important to know rates after various failure scenarios, effects
on flight software, science, etc. How to recover effective area?
Richard: We have had algorithms for TKR and CAL failures, but not ACD.
Need expert support for this including E.Charles for ACD (who will be
in Pisa last 3 months of the year)
—
Heather: new root 5.20 has been built which contains a new version of
the xroot client which is expected to solve a pathological re-read
problem in the client that causes overloading of xroot servers. Wilko
and Tom will test.
—
Heather: Will have help from Kim Lo. He will build Externals
02-05-03 for all platforms. Welcome Kim!
—
Some problem with Liz and change in overlay file location? She should
talk with Leon...
Tom: Plans afoot to move Overlay files from /u31/lsrea to xroot
(including spreading them across various servers to avoid server
overload)
—
Heather: New Science Tool build awaiting outstanding issue resolution...
ReleaseManager
Navid: ST v9r15p4 rebuild complete, but not yet tested. (An SCons
issue prevented the previous attempt from succeeding.)
Tag collector nearly complete. Still needs improvements to user
interface. Expect to finish up soon.
SCons
Some issues over the past week with tagging (container packages),
should now be fixed.
Joanne away for one more week. Still laundry list of things to do.