[daip] ANTEN under CALIB

[Jim Ulvestad]

Hi John,

I never thought about this a hell of a lot, but I think there's a
pretty simple answer.  When you include all the antennas, most of
those antennas have either no or one baseline that falls within
the 0-150 klambda range.  The strength of CALIB and like programs
is the fact that you (in general) have many baselines going into
the solution for each antenna.  I suspect that trying to back out
individual antenna solutions in this case is a very poorly constrained
mathematical problem, because you have many antennas that are
included on only one baseline, and this is why it doesn't work very
well.

Imagine, for example, that you have an antenna at AE4 that has several
baselines under 150 klambda.  But there's an antenna at AE5 whose only
baselines of that length are to AE4 and AE6.  Then the antenna at AE6
has a useful baseline only to AE5.  Trying to bootstrap out to solutions
for AE5 and AE6 antennas is likely to do nasty things to the least
squares fitter.

Probably, someone else could answer this better, but what the
heck, it's Friday evening ...

Cheers,

Jim Ulvestad

John David Monnier wrote:
> 
> Hello,
> 
>  I have an AIPS question.  I was using calib on 3C286 to do (primary) flux
> calibration.  The array was between A and B.  The "Accurate Flux Density
> Bootstrapping" memo on the web indicates that for A (B) array, one should
> use UVRANGE 0,150 and use 3 (all) antennas on each arm.
> 
> So this leads me to 2 questions.
> 
> 1) Why does it matter how many antennas I specify on each arm? If the
> minimum # of antennas is 4 (aparm(0) 4) and uvrange is set, it would
> _seem_ to have little impact, saving minimal computer time.
> 
> 2) Under this theory of how CALIB works (and because the array during my
> observation was between A/B), I tried running calib using uvrange 0,150
> and anten 0 (that is, ALL antennas).. Much to my surprise, I got 0 good
> solutions! ("CALIB2: ERROR: NO VALID SOLUTIONS FOUND").
> 
> HOWEVER, when I change my antenna command to
> anten 2,4,18,22,25,24,21,17,6,7,28,19,14   (which includes basically the
> inner 4 antennas of each arm), then RERUN it -- with uvrange 0,150 --
> then it finds lots of good solutions..
> "CALIB2: Found           28 good solutions
> CALIB2: Failed on        2 solutions
> CALIB2:                  2 solutions had insufficient data"
> 
> I don't understand why it couldn't find this solutions when Anten = 0.
> Does ANTEN OVER-RIDE the uvrange command? or is there some other
> explanation?
> 
> I tried to find my answer looking at the EXPLAIN and HELP pages on CALIB,
> but not luck.. I would appreciate any enlightenment...
> 
> Regards,
> John Monnier