Core Minutes 8/7/2007

• Collaboration Meeting: See the agenda with links to all talks for details. (Toby) The emphasis was on performance studies; there was very little on infrastructure. Riccardo Rando gave a nice talk comparing Pass 4 and Pass 5, showing that Bill had been successful in increasing effective area without any unpleasant side effects. (Leon) The Beamtest report included a discussion of response of the CsI to particles. There is a discrepancy of at least 5% with MC.
• ACD geometry: (Joanne) An extra lunch-time get-together on ACD geometry was held during the Collaboration Meeting for concerned parties. Topics discussed include
  • Top screws This has been implemented in a recent tag of xmlGeoDbs and Heather has upgraded ACD code as needed to handle both old and new-style geometries. Data collected by Luis Reyes was used to locate the screws. Since then Robert Johnson has extracted what is almost certainly better information from design drawings. Although the differences are not large, the old offsets should be replaced with Robert's numbers. [Post-meeting: replacement done in xmlGeoDbs v1r43p2]
  • Side screws They can be handled in the same manner. Code will already handle them correctly; only xmlGeoDbs has to be updated. Joanne needs data, preferably from drawings. This will probably take a while, but side screws are of much less importance to the simulation anyway. Other items will be tackled first.
  • Side shingling Robert and Bill agree the fake shingling (tested and ready to go, but not yet used in production jobs) is adequate for MC but not for analyzing real data; we need something better in place before the big background run, even if that entails a modest delay in the start of that run. There are two kinds of change needed for an accurate model: the tiles need to be located correctly, including the tilt of rows 1 and 2, and some tiles need to be represented by a volume with a trapezoidal face (not currently supported by our stuff) rather than a rectangle. Both changes involve significant work for Joanne and for Eric C. Positioning the tiles correctly is more critical. If we continue to model them with boxes the gaps at the edges where sides come together will be too large, but not by an intolerable amount and the error will be in the right direction (simulated model will have worse performance than the real thing). Hence the plan is to try to get just this change in for the big background run; that is, end of this month or early in September.
  See Robert's Collaboration Meeting talk for more information on these issues.
• Classification tree variables: (Tom) Certain new CTB variables were not in the merit file as advertised because we are still running Pass 4. After running a second stage of classification (Pass 5 rather than Pass 4) they are. All output files are kept so anyone who wants the Pass 4 variables can get them. (Tracy) The Service Challenge Steering Committee wanted to stay with Pass 4 through the Collaboration Meeting at least. It may be that we can change to Pass 5 as the default now; this is easy to do though it will involve a new tag of GlastClassify.
• ISOC Operations Simulation readiness (Heather) In connection with an operations simulation exercise planned for October, Anders would like to know about the status of a couple items:
  • EBF files: (Tom) There is need to produce ebf (raw data) files during routine Monte Carlo simulations for Flight Software and Level 0 studies. The FSW folks want triggered events but not filtered (OBF), while the Level 0 studies wish to see only filtered events. With help from Tracy and Heather, the following job options snippet, defining two instances of EbfWriter, was used with success.

    Filter.Members={"EbfWriter/EbfFirst", "OnboardFilter", "EbfWriter/EbfSecond", "Count/downlink"};
    
    EbfFirst.FileName="$(ebfFileNameT)";
    EbfFirst.WriteEbf=true;
    EbfFirst.LdfFormat=false;
    
    EbfSecond.FileName="$(ebfFileNameF)";
    EbfSecond.WriteEbf=true;
    EbfSecond.LdfFormat=false;

    Two ebf files were produced of substantially different sizes, and they have been forwarded to Bryson for evaluation.
  • New FSW, OBF: (Tracy) We are actively working on modifications necessary to update the FSW part of the OnboardFilter used in Gleam to the current flight release (B1-0-1). A meeting was held during the collaboration meeting between the GSW and FSW folks where it was agreed to incorporate into the FSW code the modifications necessary to build on both linux and windows in the GSW environment. At this meeting it was felt that the FSW build tool (CMX) could not be used to provide libraries to GSW, instead the libraries would need to be built in the FSW environment (as they are now).

    This work has begun and a start at documenting the changes that will be necessary is now in Confluence (a child page of the "Post 55 Day Run Wish List" page).

    Stay tuned!

• L0 Interleave status: (Tracy) Significant progress has been made, but this is now on hold until FSW/OBF is dealt with.
• Event collection: (Heather) RootIo has been updated to handle writing Event Collections using Eric's metaRootData package (this package has also been modified to compile on Windows, including its example applications). Reading from a meta ROOT file is almost finished and Heather hopes to finally commit all to CVS today (Tuesday - not Thursday!!!).
• SCons: (Navid) Progress is ongoing. See the SCons demo page for the latest information. Recent work includes getting SWIG to work properly and modifications to the CVS archive (a copy, not the real one for the time being) to allow SCons to interact with checkout packages in a natural way. The technique uses soft links to make the connection between checkout package and the packages it references. Next he plans to work on getting SCons to do CMT-style overrides of package version.
• Acknowledgements: These minutes are brought to you thanks to written contributions from Heather, Navid, Tracy and Tom.

previous

minutes index

next

J. Bogart, Last Modified: 21-Aug-2007 13:37:38 -0700