Monday Morning Meeting August 25th, 2014

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

Attendance

  • Socorro: Jeff, Juergen, Rob, George, Lindsey, Pam, Susan, Tak, Jim, Sanjay, Urvashi, Kumar
  • CV: Darrell, Dave
  • Garching: Sandra, Justo
  • SCO:

News / Meetings / Visitors

  • Rob on vacation starting Thurs, 8/28. Back Monday, 9/8.
  • CUC Face-to-face: Sept 22,23
  • ALMA ICT Face to Face CV: Oct 13
  • PLWG/ PLDeveloper Meeting: Oct. 20(?)

Build, Release, Testing

  • OSX Mostly working on 10.8
    • 10.9 issues with plot MS
    • Need to get help desk -> Susan to get system from Lindsey
  • New Pre-release is mostly doc changes

  • Can make switch to RH-6?
    • Make the switch after 4.3 release.

Architecture

  • Technical Forums: VI/VB2 Unification THIS WEEK!
    • 15:00 UT

Development

  • HPC meeting this week and next?
    • Proceed with known tasks, and review status on 9/11/14?
  • Statistics meeting this week and next?
    • Dave to start a draft design based on collected requirements?
    • Review proposed design on 9/9/14?

Pipeline

* ALMA Pipeline: Nearing use * VLA: In commissioning

AOB

Developer Reports

Monday Meeting
  • Sanjay Bhatnagar
    • Fix but in MSSelection
    • Imager
  • Sandra Castro
    • Have implemented the last bits of MMS heuristics in mstransform. For the last case (when combinespws=True and time averaging across scans), the task will check if the subMSs of the input contain all the selected spws. If yes, it checks if they contain the selected scans. If yes, it checks if the duration of the scans are >= timebin. If yes, run in parallel. If any of these checks fail, it will process the input MMS as a monolithic MS and create an output MS. This will interrupt the MMS flow of the data reduction! (CAS-6543).
    • Finished the implementation of a strict parsing of input parameters for flagdata's list mode. I still need to make these changes available to flagcmd. Committed to trunk in r30824. (CAS-6791).
    • Started to implement cvel2 (CAS-6835).
    • Many discussions with Justo on the heuristics of MMS in mstransform.
    • Found a bug when doing separation of spws and time averaging at the same time in mstransform. Created CAS-6889.
  • Lindsey Davis
    • pipeline development planning
      • worked on the development plan for the November pipeline release
      • wrote jira tickets for pipeline targets
    • low calibrator snr heuristics development
      • clarified simple wide to narrow spw mapping pipeline heuristics with PWG
      • started work on hifa_spwphaseup task
    • pipeline deployment
      • followed up on some pipeline build and deployment issues at JAO, DSOC
    • pipeline release testing
      • followed up on some EVLA pipeline error trapping and plotting issues
    • followup on OODT pipeline integration issues
    • ADASS registration and abstract preparation
  • Kumar Golap
    • Completed and tested cube Data and Image partitioning using coordinatesystems
    • Started to look into improving primary beam selection
      • Found that our FFTW implementation is not optimal ..making 200 beams of complex beam of 4000x4000 size and FFTing it on a 16 core machine took 270 s with our FFTW library.. but it takes some 40s with FFTPack + OpenMPing (over the 200 beams).
  • Justo Gonz├ílez Villalba
    • Completed most of the mstransform refactoring to adress issues related with transformation of sub-tables:
      • Remove duplicated rows in FEED, SOURCE, SYSCAL and FREQOFFSET sub-tables after combining SPWs
      • Multiplex rows in FEED, SOURCE, SYSCAL and FREQOFFSET sub-tables after separating SPWs
      • Add support for optional columns for SPECTRAL_WINDOW, DDI, FEED, SOURCE, SYSCAL and FREQOFFSET sub-tables
      • Merge SOURCE, SYSCAL and FREQOFFSET sub-tables from all sub-MSs after partition
      • Pending: CALDEVICE and SYSPOWER subtables
  • Jim Jacobs
    • Ger on CASA Core unification
      • Issue with QWT version on RH-6
    • NOAJ Interested in using Aligned arrays in VI/VB2
    • CARTA Software Design Notes
  • Jeff Kern
    • Continued effort on pipeline deployment
    • Usual administrative tasks
  • Susan Loveland
    • Viewer: CAS-6807, CAS-6843, CAS-5689
    • Plotms: CAS-6813, CAS-6820, CAS-6840, CAS-6841, CAS-6842, CAS-6870
  • David Mehringer
    • Fixed broken imsmooth test that was introduced in a change I made previous week.
    • Participated in unified stats framework discussion. Began enumerating design notes based on requirements and discussion.
    • Hopefully resolved intermittent specfit test failure by initializing polynomial spectral component correctly.
    • Resolved as no longer reproducable, CAS-5350: tProfileFit1D fails intermittently
    • Discussed with tester why the median beam in an even set of beams had to exist in the dataset and could not be an "average" of two beams. CAS-6727 imhead should be able to return the median of perplanebeams
    • Replaced make_pair() call with std::pair constructor, removed spurious "!" from conditional re: CAS-6861 Address C++11 issues in MSMetaData
    • Investigated UML modeling/diagramming tools for linux. Settled on VioletUML for modeling classes in unified stats framework.
  • George Moellenbrock
    • WEIGHT_SPECTRUM, WEIGHT_SPECTRUM, WEIGHT_SPECTRUM, including chan-dep statwt python prototype
    • stats framework discussions
  • Dirk Petry
  • Urvashi Rao
    • Checked in working code for old vi/vb and non-working code for new vi/vb. Dev can continue on the side on new vi/vb without disrupting everything else.
    • Evaluated work remaining to make the cube partitioning function usable in a parallel run. Simplest solution is to change its output to conform to what the framework expects. Worth testing before embarking on the longer route.
    • Halfway through connecting the old interactive clean GUI to the new framework. Looks like it's possible to use it with a reasonable compromise between old and new functionality (including runtime parameter mods per field).
    • Some user support
  • Darrell Schiebel
  • Rob Selina
    • Deployed recent pre-release builds (tars) in the AOC.
    • Collected the requirements for the statistics framework (input from the team)
    • Worked on the requirements capture and task list for first HPC release. Conclusion on path forward for the preferred MPI implementation.
  • Takahiro Tsutsumi
    • Mask handling code for refactored imager: checked in makeMask() and other related mask conversion functions. Still working on to add few more functions.
    • Investigated the issue that buildCoordinateSystem() does not return the correct spectral coordinate for optical velocity mode
Friday NAOJ Meeting
  • Kanako Sugimoto
    • Started discussion of whether or not removing old SD tasks from CASA 4.3
    • Modified simulator tasks to use new SD task ( CAS-6859) -> Ready To Test
    • Use antenna position rather than observatory position in SD imaging ( CAS-6859): Discussion with Kumar. He did the actual work, and I fixed the other bug. -> Ready To Test
    • fixed a bug of spw parameter in SD imaging ( CAS-6795) -> Ready To Test
    • Update of ATM library ( CAS-6729): Contacted Dominique again but not response yet.
    • follow-up of the MS transition meeting: discussing how to proceed with Takeshi and Kohji
  • Wataru Kawasaki
  • Takeshi Nakazato
    • [In Progress] CAS-6596: update unit test according to change in sdaverage
    • [In Progress] CAS-6702: implemented raster row selection to sdflag with helper module to detect raster rows
    • Implementation planning for MS transition
  • Shinnosuke Kawakami
    • CAS-6461(asyc); ready to test, but I did.
    • CAS-5552(imcollapse) and CAS-5553(imregrid); problem of remaining table cache (memory leak) on NFS drive. It doesn't occur on a locak drive. So Dave and I think this is not issue because CASA doesn't recommend NFS environment.

-- KanaSugimoto - 2014-08-21-- KanaSugimoto - 2014-08-21
Topic revision: r14 - 2014-08-25, JeffKern
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback