[Difx-users] Missing scans when using DiFX 2.6.1

[Jan Wagner]

Hi Matt, Geoff,

did the S&X correlation use multiple datastreams per antenna?

There was a segfault issue in difx2mark4 for EHT 2018 processing,
iirc, for multiple datastreams that I fixed in trunk (rev 8845) but
looking back also patched into 2.5.3 (rev8852) on 2019-04-16.

Thus if your 2.5.3 is from prior to 2019-04-16, and if you are using
multiple datastreams per antenna, then perhaps the indexing/segfault
fix might have triggered some issue further downstream. However for
non-VGOS IVS sessions it seems unlikely that you would have received
telescope data with multiple recorded datastreams. So perhaps this
segfault fix is not the culprit.

Other options, difx2mark4 in 2.6.x and 2.5.3 do differ in their
features. Roger Cappallo had introduced in trunk/2.6 the ability to
handle mixed bandwidth correlations and variable size channels
easily. And as far as I understood his explanation from 2019-04-19
(private exchange) this ability came at the cost of a cap on the
maximum number of visibilities that difx2mark4 can handle. While I
do not fully recall the details of the particular difx2mark4 issue I
had run into at that time and that prompted the exchange, roughly,
there was some issue with visibilities past a hard coded cap of
4080000(?) visibility records not convertering to Mark4. But this
was for a special case with many spectral channels and short
integrations. In R4910 there might be much fewer visibility records
than that cap.

Do you perhaps have available the DiFX files of one R4910 scan that
produces fewer than expected fringes?

Whoever is the active maintainer of difx2mark4 might then be able to
look into this?

cheers,
Jan

Am 30.09.2019 um 21:03 schrieb Hardin, Matthew R CTR NAVOBSY, N3EO
via Difx-users:
> Hi all,
>
> I've noticed that when I compare the output of DiFX 2.5.3 with 2.6.1, a significant number of fringes are missing from the results of 2.6.1 (486 fringes out of 7738 total). I'm using the IVS session R4910 (S/X, 24-hour) for the comparisons. There appears to be one random baseline missing from every scan (both S & X).
>
> However, I don't have the issue if I correlate with 2.6.1, but use the version of difx2mark4 that came with 2.5.3.
>
> Please let me know if you need more information.
>
> Thanks!
>
> Matt
>
>
> ------------------------------------
> Matthew Hardin
> VLBI Scientific Analyst
> Earth Orientation Department
> U.S. Naval Observatory
> 3450 Massachusetts Ave. NW
> Washington DC 20392, USA
> email: matthew.r.hardin.ctr at navy.mil
> alt: mhardin at usra.edu
> phone: 202-762-0059
>
> _______________________________________________
> Difx-users mailing list
> Difx-users at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/difx-users