GlueX Offline Meeting, February 8, 2012

From GlueXWiki
Jump to: navigation, search

GlueX Offline Software Meeting
Wednesday, February 8, 2012
1:30 pm EST
JLab: CEBAF Center F326

Agenda

  1. Announcements
    1. sim-recon-2012-01-27 released: Mark
    2. RHEL6 build added to nightly line-up: Mark
  2. HDDM's C++ API: Richard
  3. Review of minutes from the last meeting: all
  4. Reconstruction sub-group reports
    1. Calorimeters: Matt
    2. Tracking: Simon
    3. PID: Paul M.
  5. Offline Software Review Planning: David
  6. Collaboration Meeting Offline Session.
  7. Action Item Review
  8. Review of recent repository activity: all

Communication Information

Video Conferencing

Slides

Talks can be deposited in the directory /group/halld/www/halldweb/html/talks/2012-1Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2012-1Q/ .

Minutes

Present:

  • CMU: Will Levine, Paul Mattione, Curtis Meyer
  • IU: Matt Shepherd
  • JLab: Eugene Chudakov, Mark Ito (chair), David Lawrence, Simon Taylor, Elliott Wolin, Beni Zihlmann
  • Regina: Andrei Semenov
  • UConn: Richard Jones, Igor Senderovich

Announcements

  • Mark reminded us that sim-recon-2012-01-27 has been released. It fixes the "memory explosion" problem with 90 degree tracks. It also includes the C++ HDDM API. See the release notes for a list of changes from the previous release.
  • Mark mentioned that Red Hat Enterprise Linux 6 has been added to the list of nightly-build platforms. Richard asked about the "center-of-gravity" of the Collaboration in operating system space. Mark answered at JLab the usual answer is the OS running on the JLab farm, presently CentOS 5. That having been said, most collaborators are running something more modern on their desktops and laptops, both at JLab and elsewhere. We don't have a "standard" OS and do not have plans to enforce one.

HDDM's New C++ API

Richard led us through his design note on the recently release C++ API for HDDM. From the note, improvements with respect to the C API we have been using are as follows:

  1. complete encapsulation of memory management
  2. getter and setter methods for attributes
  3. inheritance of parent attributes by child elements (read-only)
  4. ability to add or delete elements to/from an existing element list
  5. method to query an element for its parent element
  6. use of stl container and iterator semantics for element lists
  7. const_iterator and iterator classes, to allow data protection when appropriate

See the note for amplification and examples. We noted that the new API allows a much more compact and transparent access to event data than before.

Matt asked if this meant we would be writing our DST's in HDDM format. Mark said that he favored that approach, but that in any case we have to make an explicit choice of a supported data format. David suggested that we schedule a discussion for the next meeting.

Reconstruction Sub-Group Reports

Calorimeters

Matt gave the report.

  • Will and Matt have talked about BCAL reconstruction and dark noise.
    • The issue is understood, but more discussion is necessary.
  • Will and Matt revisiting b1pi events
    • Assess allegations of extraneous/split clusters
    • Compare new and old BCAL reconstruction algorithms
  • Will noted that to reconstruct photons a vertex is needed.
    • Vertex finding code not optimized yet.
    • He proposed an option to use the thrown vertex for reconstruction.
    • Paul volunteered to look into adding such an option to the DVertex factory.

Tracking

Simon made some announcements.

  • He has made improvements to deal with tracks that stop in the tracking volume, for both the FDC and the CDC.
  • He has written code that recovers tracks that fail because of a kink in the track, or from over-pruning.

PID

Though not strictly a PID topic, Paul reported progress on dealing with spiraling low momentum pions. He fits to a spiraling trajectory, eliminating extra track candidates.

Offline Software Review Planning

David called our attention again to the wiki page where information for the review is being collected. He will be giving a talk on the status of the planning at next week's collaboration meeting.

Collaboration Meeting Offline Session

We hashed out the final touches on the offline session.

Action Item Review

Mark went through a two-part exercise: (1) updating the status of items in the Mantis Bug Tracker and (2) encouraging others to do the same.

Review of Recent Repository Activity

We noted that Richard recently added makefiles to the BMS system to allow user control of building GPU code in the AmpTools package. Igor reported success with using them. Matt asked for some details from the UConn folks.

New Action Items

  1. Schedule a discussion of DST/reconstruction output format. -> Mark
  2. Add an option to use truth information in the DVertex factory. -> Paul