Monday Morning Meeting May 4, 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, George, Tak, Lindsey, Sanjay, Urvashi, Juergen, Martin, Pam
• CV: Akeem, Darrell, Andy, Ville, Mark
• Garching: Dirk, Julian,
• SCO:

News / Meetings / Visitors

• Distinguished Performance Award
• CASA Ops Review
• ALMA Observing Modes Meeting

Build, Release, Testing

• CASA 4.4
  • Pre-release packages are available on cluster
  • Performance issues: changes pushed into release branch on Friday. * To be tested today

Architecture

• Technical Forums:

HPC

• OpenMP threading issues in casacore

Development

• CASA-4.5 Planning

Pipeline

• CASA 4.3.1 / EVLA pipeline commissioning: waiting for CASA 4.4 setjy flux scaling fix validation; patch CASA 4.3.1 ?
• ALMA interferometry pipeline acceptance testing: simple spw mapping fix, flagging tbuff patch
• Development
  • Web log development for session support completed and tested
  • Imaging prototype with tclean completed
  • Error handling improvements for session support
  • Flagging code refactoring started
  • Plotms cache loading inefficiencies

AOB

• -ffast-math

Developer Reports

Monday Meeting

• Sanjay Bhatnagar
  • Feed Selection in MS Selection
  • Tests on large A array imaging
  • Planning Issue: FFT Server -> API control of threading behavior.
• Sandra Castro
• Lindsey Davis
  • flagdata / tbuff / flagging command compression discussions with Jeff, Sandra, Julian, Brian, ALMA PWG
  • error handling improvements for observing session support
  • ALMA observing session / full polarization data testing
  • plotms testing with Pam, Brian, and others
  • flagdata, spw mapping patches to support ALMA interferometry acceptance testing
• Pam Ford
  • CAS-7373 (Resolved) - colorize by field issued fixed by VIVB2 implementation in plotms
  • CAS-7482 (Resolved) - plotms will use FLOAT_DATA column when default DATA column does not exist in single-dish MS
  • CAS-7428 (Resolved) and CAS-7481 (fixed) - pipeline plotms write-lock issues fixed
    • sd pipeline issue revealed Cube resize error, created CAS-7488 for MSTransformBufferImpl bug
  • CAS-7511 (Resolved) - discussion of antenna selection syntax
  • Tested and Resolved:
    • CAS-7390 (Virtual model data), fixed related bug in plotms
    • CAS-7412 (Improve performance with subchunk counts)
      • creating VI and getting VisBuffer structure performance now comparable to 4.3, verified by Lindsey
  • Tested loading requested metadata axes only
    • no performance improvement, data cubes seem to be the big performance issue according to Jim
    • plotms always loads time, field, spw, channel, frequency, corr, scan, ant1, ant2, baseline, flag, observation, and intent axes
  • Plotms being tested by Drew Brisbin in CV
    • Fixed plotms bug with time- and channel-averaged weight and sigma axes (CAS-7118 and CAS-7119)
  • Made release-4_4 developer build for testing/patching
  • Finally built casa on OSX 10.10 for CAS-4592 and CAS-7246. HALLELUJAH
  • Investigated running plotms in parallel on same MS, created CAS-7504 for MSTransformIterator write-lock
  • Investigated plotms correlation selection issue discovered by Tak in CAS-7434, created CAS-7512 for MSTransformBufferImpl bug
• Kumar Golap
• Justo González Villalba
  • Further improved performance of MSTransformIterator initialization to be in line with the old plain VI/VBJim Jacobs
  • Fixed bug in frequency calculation for channel average
  • Minor fixes in connection with default virtual model data column
  • Tested OpenMP improvements introduced in the last casacore commits
• Jeff Kern
  • ALMA Obs Mode Review
• Susan Loveland
  • CARTA save/restore state
  • CAS-5689: Profile fitting incorrect for particular cube and settings.
• David Mehringer
• George Moellenbrock (2 weeks)
  • CAS-7478: uvcontsub segv; unreproducible except with user's typo in fitspw specification (followup today)
  • CAS-7271: Mis-calibration occurred when field changes not recognized as "new cal", and gainfield='nearest' (fixed)
  • CAS-6995: 'nearest' interp in frequency mis-calculated for LSBs (fix pending)
  • Cal lib "quick reference" nearly ready for v4.4 testing (also a facility for translating old applycal parameters to callib)
  • Polarization and other user support for JAO, Socorro, Japan, etc.
  • Catching up on various writing projects: Summer School book Calibration chapter, Lin feed Poln Memo, prep for Interferometry and Calibration lectures at Canadian Workshop (August)
• Dirk Petry
• Martin Pokorny
  • Progress on prototype for parallelized iteration control in tclean: existing serial, continuum and cube parallel imager, normalizer and deconvolver components now can plug into refactored framework. Focus next on iteration control.
• Urvashi Rao
  • Writing functional (python level) tests for tclean use cases and converting all existing tool level examples to task tests.
  • Other : Turned off more log messages from 'ia' tool use within old clean. Started build of release-4_4 on el6 desktop and got stuck. Went through Jeff's 4.5 todo list and gave tclean related feedback.
• Darrell Schiebel
• Ville Suoranta
• Julian Taylor
  • minor tweaks to mpi launcher (CAS-7415)
  • invalid timerange fixes to manual flag compression (CAS-7349)
  • setup serial vs parallel performance monitoring for alma cycle 2 testcase (CAS-7483)
  • filed casacore PR with booltobit performance improvement (CAS-7406, gh-101)
• Takahiro Tsutsumi
  • Makemask help doc update and some more testing to make sure everything is working as expected
  • Setjy bug fixes related follow-up discussions/testing
  • Found an issue with the task interface: setting in a global parameter can modify the default parameter behavior in subsequent task run with arguments inside if the global parameter is the task's sub-parameter and not explicitly set in the task command (e.g. setjy's fluxdensity parameter)

Friday NAOJ Meeting

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