Monday Morning Meeting - 10 August 2015

• DIAL-IN NUMBERS & PASSCODES:
• USA Toll Free Number: 866-901-8266
• USA Toll Number: +1-203-566-3863
• PARTICIPANT PASSCODE: 725760

Attendance

• Socorro: Jeff, Susan, Jim, Lindsey, Sanjay, Urvashi, Kumar, Martin, George
• CV: Akeem, Ville, Mark, Darrell
• Garching: Sandra, Julian, Dirk
• SCO:

News / Meetings / Visitors

• VLASS is ramping up
• Dual mode in ALMA

Build, Release, Testing

• Issues for the release:
  • Ephemeris Objects
    • Nearest in gainfield
    • Import ASDM issues with RefCodes
  • tclean
    • Make the sorted subtable in memory rather than on disk
      • Throw this switch
    • Issue with clearmod in tclean, open's source table but does not release lock.
    • Clark Clean: One is memory based and one is disk based
      • Yes to be cleaned up in 4.6
  • Issue with canned model for VLA
    • To be investigated about which position is being used.
  • BDF Flags in ALMA
    • Michel is looking at the issues

Architecture

• Technical Forums:

HPC

• Testing and bug fixing phase. All implementation is in place for 4.5.
• HPC-related blocker ticket, CAS-7705 (tclean exhibits excessive number of open() calls against MS subtables): James saw that running tclean in sequential on an MMS was slower than on a normal MS. Julian has seen that VisibilityIteratorReadImpl::setTileCache opens each SubMS once for each of the 8 sorting columns. This may explain the slowness in tclean on an MMS.
• Next step for 4.5 is to write documentation for the users on how to run CASA in parallel.

Development

• Sorted Column

Pipeline

• Multiple-spw science target continuum images, first continuum subtracted science target cubes with tclean
• CASA 4.5. integration issues: BDF flags, WVR uncorrected vs corrected flags, ephemeris tables, tclean mosaics

AOB

Developer Reports

Monday Meeting

• Sanjay Bhatnagar
• Sandra Castro
  • Fixed a bug on the creation of a symlink for the SYSPOWER sub-table on MMSs.
  • Created MMS for Jenkins to run on tasks automatically.
  • Changed the error handling in flagmanager when user asks to save a version name that already exists. The task will issue a warning and rename the existing flagversions to a ".old.timestamp" name.
  • Verifying the flagging of shadowed antennas when ref frame is other than J2000. Todd claims that it is not flagging one MS on APP after my fix.
• Lindsey Davis
  • continued work on AQUA report generation, most flagging steps have entries
  • CASA 4.5. testing and bug fixing
• Pam Ford
• Kumar Golap
  • Reqal Gridder not yet checked in, giving different results
• Justo González Villalba
• Jim Jacobs
• Jeff Kern
  • Various Management documents
  • SKA Architecture Report
  • Beginning 4.6 Planning
• Susan Loveland
• David Mehringer
• George Moellenbrock
  • On "vacation" for 3 of the past 4 weeks
  • Most common parametrized types now work via CalLibrary
  • CalLib parsing now occurs in CalibratingVi2 code so clients need only specify the file name
  • Nuts and bolts work continues on cal solve migration to VI2: calibration code is in good shape; still some difficulty layering ViImplementations (maintaining control of the calibration-aware layer outside the VI2). Have decided to punt on this for v4.5.
  • Various weight-related discussions w/ SD folks and Crystal. Found global factor of 2 error in traditional statwt (due to combining real and imag parts in the std estimation), as well as digital factors as large as sqrt(2) missing from the nominal initialization. (Context here is the case of dataset combinations, so absolute weights matter.)
  • Also various work on Penticton Interferometry workshop talks and ALMA polarization CASAguide and other testing.
• Dirk Petry
• Martin Pokorny
  • Continuing work on visstat2; primarily work on the tool-level code this week, some work on code using the statistics framework.
• Urvashi Rao
• Darrell Schiebel
• Ville Suoranta
• Julian Taylor
• Takahiro Tsutsumi
  • Worked on modification to mask parameter interface (i.e. moving mask parameter out of sub-parameters of niter>1 ). Still needs some more work especially to make sure it works properly for multi-field case
  • Added parallel cube tclean to use virtual concat. but ran into a problem of table-lock in accessing some of sub-cube images to be concatenated
  • Investigated a new multiple channel selection issue reported by Dirk M.
  • ~ 0.5d on science: reading and commenting on a draft of science paper using ALMA archive data

Friday NAOJ Meeting

• Kanako Sugimoto
• Wataru Kawasaki
• Masaya Kuniyoshi
• Takeshi Nakazato
• Renaud Miel