User Tools

Site Tools


2020 Mini-workshop #2

A zoom link will be provided via email to the difx-users list. The meeting will be recorded and made available shortly after conclusion.

  • Proposed date: Tuesday December 1 1500 UTC
  • Proposed title: Mixed-polarisation correlation and post-correlation in DiFX
  • Proposed agenda:
    • Moderator: Geoff Crew
    • PolConvert Update (Geoff Crew, 20 minutes)
    • difx2fits alternative: (Leonid Petrov, 20 minutes)
    • Discussion of polarisation handling in DiFX and post-correlation (20 minutes)
    • Overview of VGOS mode issues (John Barrett, 10 minutes)
    • Discussion of sideband issues highlighted by VGOS - how to avoid in the future (20 minutes)
    • Planning date/time/topic for next mini-workshop (10 minutes)
      • Proposal: 1 or 2 weeks later. Select topic from available list or a new idea.



Note for Adam: recording not possible without host permission!

Geoff Crew's talk

Comprehensive slides available here

drivepolconvert provides simplified interface to underlying polconvert. Should be operator friendly.

converting polconvert to python3 in progress and almost complete.

Leonid Petrov's talk

Comprehensive slides available. here

Some discussion around difference between XY and HV only being important for long baselines - hence not properly distinguished in most connected element arrays or software packages.

Can FITS-IDI be modified to allow proper description of mixed pol cases using POLTYP[AB] fields already available in standard? Consult with Eric Greisen.


The main action seems to get Eric G. to accept Leonids proposal. Mark Kettenis is looking at putting polconvert into CASA itself (ie. MS based)

John Barrett's talk

Slides available here.


  • create a 2.5.4 tag
  • then 2.6.2[3?]; potential changes might help?
  • try using zoom bands
  • change the RDBE-G/R2DBE-G (probably the real fix, and is perhaps a config flag change only)

Some discussion on how to get the 4th bullet into the field system…. Should capture 2.5.4 for Bonn and USNO to capture state, at least? Then investigate zoom and verify. DiFX 2.6.3 is (testing aside) ready to go, go ahead and put changes into 2.6. Then test that with zoom Then work on the config fix.

How to fix this in the future. Obviously a set of test data. DiFXdiff should be run on captured data. (Do this at Bonn.) Synthetic data generation for testing. GPU program should provide us with some automated tests.

next meeting topics

NA to get the short straw on timezone, ideally held before christmas

  • Jan on output bands
  • Lenoid on Calc to VTD library
  • Jakob on (short) simulation tool of VLBI raw data


Geoff, Leonid, John, asowink, Christian Plotz, Cormac, Frederic Jaron, Greg Lindahl, Harro Verkutier, Helge Rottmann, Jakob Gruber, Mark Kettenis, Michael Dutka, Walter Brisken, (and maybe a few more in and out).

difx/meetings/2020miniworkshop2.txt · Last modified: 2020/12/02 13:58 by cormac