[Difx-users] Missing baselines with R-X/Y data

Thu Feb 4 12:09:59 EST 2021

Hi Adam, all

Finally some data so I can return to the mixed-pol issue again. To remind you, my problem is that when correlating R-pol (1Gbps) data from ONSALA60 against X/Y-pol (2Gbps) data from ONSA13NE and ONSA13SW, I must define “zoom bands” to get visibilities on all baselines. Not sure if the problem is in DiFX, or difx2mark4 (or elsewhere).  I have uploaded a tar.gz archive with input/output files (excluding 18GB of VDIF data) at
https://chalmersuniversity.box.com/s/9ed0t61bk04ag88is2ovhlm0er4qd00n
There you can check logfiles, input files etc. If you want the data, I can share it (no time right now to remember how to split out a small chunk of data for testing, so used one full 30 sec scan). I can also try various things on my end if you have suggestions.

Here are the steps to reproduce with DiFX 2.6.2:

Unzip archive,
Edit "*.files" with correct paths to "data" directory
cd zoom
Edit "on1028_1.machines" accordingly (e.g. computer name)
vex2difx -v -v -v -d *v2d --force (need force as this is mixed circular/linear data)
calcif2 *.calc
startdifx -n -f *.input -v
I now get messages, see logfile, like
"Thu Feb  4 17:56:58 2021   0 skirner WARNING  Baseline 0 frequency 0 points at two different frequencies that are apparently identical - this is not wrong, but very strange.  Check the input file"
The message is correct, I have identical zoom windows - otherwise I don't get all baselines. To see result, I run
difx2mark4 -v -b X 2300 14000 *.difx -s ../station_code_file.txt
and check the output on screen. All baselines are included:
[...]
      n120[SX] 448
      n120[ST] 896
      n120[XT] 448
      n120[SS] 448
      n120[XX] 448
      n120[TT] 448
[...]

Same procedure without zoom-windows, I get a different set of baselines

[...]
      n120[SX] 448
      n120[XT] 448
      n120[SS] 448
      n120[XX] 224
      n120[TT] 448
[...]
Curiously, the "ST" (which is the ONSA13NE-SW baseline) is missing, and the XX (ONSALA60) is half (224).

Not sure the zoom-window version is correct, but I need the ST-baseline, and it seems to work, so I use zoom-windows. But, I'm curious: why does this happen?

Kind regards
Eskil

From: Difx-users <difx-users-bounces at listmgr.nrao.edu> on behalf of Eskil Varenius via Difx-users <difx-users at listmgr.nrao.edu>
Reply to: Eskil Varenius <eskil.varenius at chalmers.se>
Date: Monday, 30 November 2020 at 11:51
To: Adam Deller <adeller at astro.swin.edu.au>
Cc: "difx-users at listmgr.nrao.edu" <difx-users at listmgr.nrao.edu>
Subject: Re: [Difx-users] A few surprises

Hi Adam,

Apologies, but I will have to defer sending additional details as the data in question have already been removed after correlation. (I would like to verify that the files I send indeed produce the errors I claim, so I don't send files with other typos). Also, I have too many other todo-items to follow this up right now, but we regularly observe with similar settings so I should get another data set to play with in a few weeks. When I do, I will check the input-file differences and let you know, and keep the data on disk for additional tests. But I did experience the same thing for all similar data so far. Will let you know once I have new data.

Cheers

Eskil
On 2020-11-30 11:43, Adam Deller wrote:
Hi Eskil,

Thanks for the heads up, the first two are surprising (I'm never surprised by gaps in documentation!)

I've got no experience with the intel compilers, so will pass on that one.  For the second, what is the difference in the .input file produced by vex2difx in the two cases?  That might help shed some light.

Cheers,
Adam

On Mon, 30 Nov 2020 at 21:18, Eskil Varenius via Difx-users <difx-users at listmgr.nrao.edu<mailto:difx-users at listmgr.nrao.edu>> wrote:
Dear difx developers,
I have noted a few surprises which I would like to report:

- Compiling DiFX 2.6.1/2.6.2 with intel 2019 compiler, mpifxcorr runs
but refuses to actually read any data. It just sits there, never
starting the correlation. Using intel 2018 compiler instead, everything
works nicely.

- Correlating mixed-polarisation data (antenna ON: R-pol, OE and OW: X/Y
pol) delivers correlation products on the mixed circ/linear baselines
(R-X and R-Y pol), but not on the linear/linear baselines (XX,XY,YX,YY).
However, if I add zoomwindows - describing the same setup already in the
VEX file, so no actual zooming - then I get also the linear polarisation
products. (I know mixed-pol support is limited, but might as well note
the behaviour.)

- "difx2mark4 --help" v1.6 does not say it needs .calc files, but seems
it does. So documentation should be amended?

Kind regards
Eskil

_______________________________________________
Difx-users mailing list
Difx-users at listmgr.nrao.edu<mailto:Difx-users at listmgr.nrao.edu>
https://listmgr.nrao.edu/mailman/listinfo/difx-users

--
!=============================================================!
A/Prof. Adam Deller
ARC Future Fellow
Centre for Astrophysics & Supercomputing
Swinburne University of Technology
John St, Hawthorn VIC 3122 Australia
phone: +61 3 9214 5307
fax: +61 3 9214 8797

office days (usually): Mon-Thu
!=============================================================!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/difx-users/attachments/20210204/c09771ac/attachment.html>