[Difx-users] Lack of mixed-mode fringes with 2.6.3, then segfault with 2.8.1? {External} {External} {External}

Sun Aug 13 23:45:39 EDT 2023

hi Eskil,

I also reported an issue in 2.8.1 when using zoom bands a couple months back. I didn't get a chance to pursue it due to other pressing matters, and our correlator hardware is currently offline, but it is likely we are seeing the same issue. I've attached the issue I described to the foot of this message.

Regards,
Cormac.

==============================================
[forwarded message]
hello again developers,

in further testing of the newly released DiFX-2.8.1 I appear to have come across another case of missing subbands. This time in cross correlations rather than autos, but it otherwise looks remarkably similar to the problem I was seeing in DiFX-2.8 before Walter's fix - e.g. I only see the problem when using zoom bands and mixed modes. The same data correlate fine under DiFX-2.6.3. Also the input files for DiFX-2.8.1 seem to be correct, suggesting it is again an mpifxcorr issue.

I attach 2 plots to demonstrate the difference. The first (v622-tst2) shows the spectra from AIPS with DiFX-2.6.3 and the second from DiFX-2.8.1 (v622-tst). Half the subbands on the MP-CD baseline are missing in the DiFX-2.8.1 version. All subbands are present on AT-CD and AT-MP baselines.

Unfortunately I'm about to disappear on leave for two weeks so haven't had a chance to pursue this as far as I would like, but I thought to share it with the brains trust now, especially given the similarity with the recently solved bug. The minimum example is available here:

https://cloudstor.aarnet.edu.au/plus/s/iSZLdIDRHsKspB2
[https://cloudstor.aarnet.edu.au/plus/3rdparty-apps/cloudstortheme/core/img/favicon-fb.png]<https://cloudstor.aarnet.edu.au/plus/s/iSZLdIDRHsKspB2>
CloudStor - CloudStor is powered by AARNet<https://cloudstor.aarnet.edu.au/plus/s/iSZLdIDRHsKspB2>
v622a is publicly shared
cloudstor.aarnet.edu.au

that should contain everything required to re-create the problem including the baseband data if you are really keen.

Let me know if other info is useful.

________________________________
From: Difx-users <difx-users-bounces at listmgr.nrao.edu> on behalf of Adam Deller via Difx-users <difx-users at listmgr.nrao.edu>
Sent: Sunday, 13 August 2023 19:02
To: Eskil Varenius <eskil.varenius at chalmers.se>
Cc: difx-users at nrao.edu <difx-users at nrao.edu>
Subject: Re: [Difx-users] Lack of mixed-mode fringes with 2.6.3, then segfault with 2.8.1? {External} {External}

Hey Eskil,

Have you tried vex2difx -v -v and/or valgrind vex2difx to see where the segfault is occurring?  Obviously, it shouldn't be segfaulting, so presumably finding that and fixing it should go a fair way towards solving the problem.

Cheers,
Adam

On Sat, 12 Aug 2023 at 04:22, Eskil Varenius via Difx-users <difx-users at listmgr.nrao.edu<mailto:difx-users at listmgr.nrao.edu>> wrote:

Dear Difx-users,

Background: I have mixed-mode continuum data from 3 stations: Ns(16 MHz VDIF), Ny(16 MHz M5B), Nn 32 MHz VDIF). Ns and Ny observe standard S/X with R-pol only, and Nn (VGOS) observe dual linear X/Y (strictly speaking H/V but I'm using X/Y here since that's the VEX labels). Correlating Ns-Ny baseline with DiFX 2.6.3, is all good. I use zoom windows (to get 16 MHz from all stations) and FreqId selection (to only correlate X-band and skip S-band channels). But when trying to correlate all three stations I run into problems.

Problem 1: With DiFX 2.6.3 I can correlate all three stations, and then run difx2mark4. Fourfit then gives me nice fringes on the Ns-Ny baseline (R-R pol), and the Ns-Nn baseline (RX and RY pol). But while I do get nice XR fringes on the Nn-Ny baseline, I do not get any YR fringes on this baseline. I know the data are good, since I get fringes to these polarisations on another baseline. So I suspect I have fallen into the "Warning: Limited support for mixed-mode..." here, or do you have other suggestions?

I saw there are improvements for mixed-mode in DiFX 2.8.1, so I decided to try it, but:

Problem 2: With DiFX 2.8.1, I get "segfault" when I run ”vex2difx” (this worked with 2.6.3). Any ideas?

I guess hunting the "segfault in 2.8.1" is the most profitable route for the future? If so, please let me know what info you may need to speculate further. I am happy to share output, v2d, vex, and even raw data for one scan for someone to reproduce this if you would be so kind to help.

I am very puzzled by this!

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

--
!=============================================================!
Prof. Adam Deller
Centre for Astrophysics & Supercomputing
Swinburne University of Technology
John St, Hawthorn VIC 3122 Australia
phone: +61 3 9214 5307
fax: +61 3 9214 8797
!=============================================================!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/difx-users/attachments/20230814/28b51839/attachment-0001.html>