[daip] UVMOD/UVFIX question

[Eric Greisen]

Hanno Spreeuw wrote:
> Hello Eric, I have posted a UVMOD test on my homepage, please have a look.
> http://staff.science.uva.nl/~hspreeuw/
> Cheers,
> Hanno.
> 
> Eric Greisen wrote:
>> Hanno Spreeuw wrote:
>>
>>> Thank you,
>>>
>>> It seems that inserted sources with FPOS<>0 do not appear in the 
>>> images exactly where they should. The errors increase away from the 
>>> tangent point. This confirms earlier results by James Miller-Jones 
>>> (CC). However, with FPOS=0, the source always appears at the tangent 
>>> point.
>>> We are testing new source extraction code. In order to check our 
>>> deblending algorithm, we are inserting many sources in a small patch 
>>> of the sky. With these artificial sources one always knows the true 
>>> flux and position, so that is good for testing.
>>> I am considering the workaround with UVFIX for these tests.
>>
>> If what you say is true, then it is unlikely for any of AIPS to work 
>> correctly.  UVMOD uses the same phasing routines used elsewhere in AIPS 
>> including but not limited to UVFIX and IMAGR.  What telescope has 
>> produced your data and what are you doing?


I have loaded your data and - sort of - followed your procedure.  I 
worried that the shift of 1000 arcsec might be rather large.  SETFC does 
say that one can make this image - but I am surprised that it does.
Anyway - what I did was to do 2 fields with RASH=0; DECSH=0,1000 and 
with DO3D 1; OVERL 2.  In the first run, I restricted the clean box to 
the second shifted field.  Afterwards, IMFITs gave the desired answers 
to great accuracy in each field.  Then I reran IMAGR, restricting the 
Cleaning to field one.  Afterwards, IMFITs gave very similar and shifted 
results.  This is due to the geometry and various 3D effects which are 
much worse at low elevations, esp low declinations.  This is illustrated 
in some of the lectures in the NRAO Synthesis Workshop volume(s).  One 
should also note that, compressing the CCs in the first case led to only 
3 pixels with CCs - a nice clean source although my window allowed for 
some negatives etc to appear.  The second clean compressed down to 24 
CCs - a lot of smeared flux.  In fact the flux will be smeared along an 
ellipse which will have a maximum where the traverse of the ellipse is 
slowest (at the bottom) and then curves upwards on both sides.  You can 
see it in the images.  I attach the IMFIT print outs.

Thus, the problem lies in your use of AIPS IMAGR, not in UVMOD.

Thanks you for pursuing this so that the correct answer could be found.

Eric Greisen
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Snake.prt
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20080612/cd054d5d/attachment.ksh>