[daip] UVCON

[Eric Greisen]

Katherine Blundell wrote:
> Hi Eric,
> 
> Btw - we're trying to perform a wide-field simulation which demonstrates 
> W-term source smearing at the edge of the field and we're using UVCON to 
> do this using a DFT and a simple clean component sky model with 
> cross-pattern of sources spanning about 6 degrees.
> 
> Performing this with and without the DO3DIMAG option and then comparing 
> the resulting images does not show what I would expect, i.e. sources 
> smearing out radially at the edges of the field.   Rather what we do see 
> is very slight shifts in source position but in the same direction for 
> each of the sources (i.e. uniform position offset for the whole field).
> 
> Using UVPLT to compare the datasets shows that the DO3DIMAG option does 
> affect the w-values.
> 
> Has anyone used UVCON for this sort of simulation before? Is there 
> something we are missing?  Please let me know if you need copies of our 
> task inputs and/or images.

UVCON really has conceptual problems with what is meant by the model.
It probably should not allow the DO3D option and should view the CC or 
image model as consisting of a single plane even if that is a bad model 
when the W term is considered.

If you have a good model (e.g. suitable facets etc) then UVSUB or OOSUB 
may be used to replace the UVCON model values with values computed using 
the full projective geometries available in AIPS.

If your model consists of a few sources, then UVMOD can take an existing 
uv data set and replace the data with the model data.  That does the 
full 3D stuff correctly as tested with a multi-facet IMAGR (UVCON fails 
that test).  I am going to change UVMOD to allow a text file with up to 
10000 components to be read instead of the 4 components in the adverbs.
Look for that change in a day or 2 in the MNJ.

Leonia will reconsider the conceptual basis of UVCON's modeling but only 
after he returns from a meeting of SKA array design.

Cheers,

Eric