[daip] [!4954]: aips - CALIB amplitude errors in phase-only solution

[Adam T Deller]

Adam T Deller updated #4954
---------------------------

CALIB amplitude errors in phase-only solution
---------------------------------------------

           Ticket ID: 4954
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/4954
           Full Name: Adam T Deller
               Email: deller at astron.nl
             Creator: User
          Department: AIPS Data Processing
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template Group: Default
             Created: 26 May 2014 12:41 PM
             Updated: 26 May 2014 12:41 PM
                 Due: 29 May 2014 12:41 PM (3d 0h 0m)
      Resolution Due: 04 June 2014 12:41 PM (9d 0h 0m)



Hi,

I am processing VLBA + phased VLA data on SgrA* at 8 and 15 GHz (separately).  After performing standard delay calibration, I run CALIB using solmode 'p', providing a model for SgrA* (which is rather resolved at these frequencies).  I use a uv range to prevent data of really low SN contributing to the fit, but by necessity I have to go out to baselines where there is not much flux.

The output SN table contains, on occasion, entries with amplitudes that are not equal to 1.0. Obviously, this should not happen for a phase only selfcal!  I notice that it tends to happen most frequently in solution intervals where one antenna is missing, for example if KP is missing then sometimes the OV solution has an amplitude != 1.  Since in this example KP-OV is the baseline to OV with the highest flux density at that time, what I suspect is going on is that there is a loss of precision somewhere, as the model flux on all the remaining baselines to OV will be low or very low, and so when the visibilities are divided by the model you are probably getting very big numbers and very small weights.  And so when the antenna based solution is calculated somehow it is not being then normalised properly.  Another factoid that points me in this direction is that these bad amplitudes in the SN table seem to be much more common in the 8 GHz data than 15 GHz (and by necessity, I'm going further into the resolved model at 8 GHz).

I have attached some 8 GHz uv data and a model that illustrates the problem.  If you run CALIB with solint 0.2, soltype='L1R', uvrang 0,19000, refant 11, aparm 3,0,0,0,0,2,4.5,0, and the attached model, you should then see the wonky amplitudes on OV.  SN table 2 of the file itself shows what I get when I run CALIB with these parameters.

Looking forward to hearing what the deal is.

Cheers,
Adam

------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: calib-killer_uv.fits
Type: application/octet-stream
Size: 3490560 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140526/4663c6b4/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: calib-model.fits
Type: application/octet-stream
Size: 1736640 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140526/4663c6b4/attachment-0001.obj>