[daip] [!11333]: AIPS - AIPS UVSUB doesn't work with multiscale CLEAN images and CMETH='GRID'

[Michael Bietenholz]

Michael Bietenholz updated #11333
---------------------------------

AIPS UVSUB doesn't work with multiscale CLEAN images and CMETH='GRID'
---------------------------------------------------------------------

           Ticket ID: 11333
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/11333
                Name: Michael Bietenholz
       Email address: mbieten at yorku.ca
             Creator: User
          Department: AIPS Data Reduction
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template group: Default
             Created: 05 December 2017 01:02 PM
             Updated: 05 December 2017 01:02 PM
           Reply due: 07 December 2017 01:02 PM (2d 0h 0m)
      Resolution due: 31 August 2020 12:00 AM (999d 10h 58m)

AIPS UVSUB (31DEC17) seems to get the wrong answer when asked to subtract multiscale clean images in 'GRID' mode, but produces a reasonable answer in 'DFT' mode.  I attach a png showing UV plots which were made from UVSUB OPCO='MODL' output, with the only difference being yellow was CMETH='DFT' and green was 'GRID'.  Details below.  Now you wouldn't expect the two to be *exactly* the same, but they should be pretty similar, and as you can see they are completely different.  I think in the 'GRID' case UVSUB only uses the last of the input imagesI ran an IMAGR with NGAUS=3 and WGAUS=0,30,90 (normal beam is around 15" for this particular data).  IMAGR produces three images (.ICL001, .ICL002, ICL003) of same patch of sky but at three different resolutions (with first one having clean comps from all three restored).  There are >30 clean comps in each resolution.I tried running UVSUB with OPCO='MODL' to replace my vis. with the model values from all three images, and got strange results (model had much lower total flux density than was in the clean comps)As a test I ran UVSUB w/ NMAP=3; NCOMP = 30,30,1.With CMETH 'DFT' results are reasonable (yellow points in attached plot).  With 'GRID' (green) I think its ignoring the first two images, and only taking the single requested component from the last image, which is a single Gaussian.I note that UVSUB claims to need 'XY' sorted data for 'GRID'.  I did sort my data to XY, but 'GRID' still gets it wrong.

------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20171205/c075e610/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: uvsub_uvplot.png
Type: image/png
Size: 5166 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20171205/c075e610/attachment-0001.png>