[daip] FLATN - artifacts

[Eric Greisen]

Hans-Rainer Kloeckner wrote:
> 
> Hi Eric,
> 
>     Thank you ! Somehow this option slipped my radar.
> 
> 
> I have two more question and hope you have a few minutes.
> 
> CC - component model: Is CALIB able to use the CC table produced via SAD 
> (having extended sources). If so where can I find
> the definition of the "extended" CC-table (e.g. the switch if a source 
> is extended).
> 
> 
> I'm calibrating GMRT data (as you might know I'm developing an automated 
> pipeline to handle these dataset)
> and checking my selfcalibrated datasets with the model of my clean 
> components  (using VPLOT). Sometimes, a lot of the baselines with very 
> large negative delta  z components show very little coincidence with the 
> UVdatset. So far I always suspect there is a problem with the GMRT data 
> (I think there still is), but maybe there is also an additional switch 
> in aips to take special care about the W-term in VPLOT (I'm aware of the 
> 3dimaging switch)?
> Or is there a more obvious explanation and I don't get it  (most of the 
> antennas are on the NORTH-SOUTH arm)?

CALIB in 'GRID' mode cannot handle a CC table with different CC sizes. 
The DFT mode might work for this - I am actually not sure.  CALIB can 
even do GRID for the multi-scale data in which each CC has a fixed 
component size even though the different facets have differing sizes.

When using VPLOT do not have it compute a smooth function (NMAPS < 0) 
for the most accurate results - use NMAPS > 0.  The routine that is used 
for NMAPS < 0 has to compute u,v,w ab initio from the AN table and does 
not do the elaborate job which is actually required (as done in UVFIX).

Otherwise, VPLOT should do a decent job of plotting the model including 
the full do3d options etc.  It does have a finite limit on the total 
number of components so be careful of that.

Eric Greisen

Was this new option the cause of your FLATN differences?