[alma-config] CLEAN and difference results for B arrays

[Bryan Butler]

>A disk model can easily be subtracted analytically from the uv-data
>removing any doubts about pixelation. The goal of deconvolution is
>to subtract the most reasonable models (consistent with our a-priori
>information) which leave a residual image which is consistent with
>the noise.

i don't want to go into too much depth on a relatively minor point,
but, the goal in the initial "model subtraction" step (whether it
is u-v or image based) should be to do as well as you can with the
model (to get the residual image as close to consistent with the
noise as one can).  the fact is that uniform disk models simply 
don't do that great a job for the planets.  limb-darkened ones are
better, but the problem with that is finding a limb-darkening model
in image space that has an analytically solvable FT (e.g., i use
cos^n(theta), which transforms to a Lambda function of order
1+n/2).  but the ones that are analytically solvable are also not
particularly great representations of what you expect (at least
the ones i know about).  so, you are stuck with either doing what
you know is an imperfect u-v based subtraction, or going to a
clean component representation which does a better job of
representing what you expect on the sky (but has some problems
with pixellation).  either way, you suffer problems, but it is
probably an oversimplification to say that one is "better" than
the other (IMHO)...

	-bryan