[daip] SYMBOL?

[Eric Greisen]

Ian Pattison writes:

 > I wish to concatenate A,B & C configuration VLA data (5 GHz). The B and C 
 > data have already been concatenated in DBCON, given to me by a 
 > collaborator. I wish to combine this with my A config' data, which was 
 > taken in '4 mode', i.e. 8 x 3.125 MHz per IF. The first thing I did was to 
 > SPLIT the BC data and run UVFIX on each pointing, to convert B1950-J2000. 
 > I then DBCON'ed these back together. I then shifted the centre of my A 
 > data using UVFIX, so that it was the same as the BC data.
 > 
 > The next step was to SPLIT my A data into 14 channels (7 per IF) and then
 > DBCON these back together. My overall concern regards choosing the
 > uniform, natural or robust weighting. I wish to retain the high resolution
 > of A configuration, but also map the extended emission (there is emission
 > up to ~10 asec scale). Now, when I DBCON these 14 together, I can REWEIGHT
 > each relative to the previous. So, assuming I want uniform weighting in my
 > final ABC map, should I run IMAGR (using Uniform weighting) on each to
 > obtain the sum of gridding weights? Always relative to the weight of the
 > newly created .DBCON data. If so then REWEIGHT x y always increases by 2,
 > i.e. 1,1  2,1  4,1 etc..

I do not understand this remark.  Assuming that the 1st data set is
the already concatenated data so far then REWEIGHT(1) = 1.
REWEIGHT(2) will vary but only with which array you are including
being the same for each of the A array channels.

 > 
 > If this is correct then I should also obtain the 'sum of gridding weights'
 > of the BC data, before I DBCON with my A data. This gives a REWEIGHT 
 > of (70'000 , 1) for BC , A. This seems awefully high.
 > 
 > Have I made any mistakes in what I've done?
 
The issue is what sort of weights do you have in these data?  Are they
"correct" weights approx = 1/sigma**2 in 1/Jy**2?  This is what you
want and then REWEIGHT=1 is correct.  Or are the weights simply the
number of 10-second intervals averaged (e.g. 1,2,3,4)?  It is hard to
combine data with mixed methods of weighting.  The task FIXWT may be
useful although it is somewhat problematic.

Note that IMAGR will produce the same weighting whether the A array
data are spectral line or split up the way you have done (and must do
for your data).

Eric Greisen