[daip] problems with CCMOD, UVSUB, IMAGR

[Lawrence Rudnick]

Dear Gurus:

I am getting what look like fundamental problems in some
combination of these three tasks.  It is either user
error (which is certainly possible, and I apologize for
any wild goose chases I send you on), or there are one
or more serious bugs in these programs. I am using
31Dec02 on linux.

I have performed two tests as suggested by Bill Cotton
on Friday.  These will be described further below.
I have put some of the results on a web site, for your
perusal.
http://www.astro.umn.edu/~larry/LSS/atest/

My suggestion is that to eliminate user error, that someone
repeat the experiments with my test data sets.
I have put  three uv data sets into my anonymous ftp area for
this purpose.  They are

ftp.astro.umn.edu
cd pub/users/larry/atest
   TEST1 - original data
   TEST2 - data after uvsub (add) of 3 ccmod 1000 Jy sources
   TEST3 - data after uvsub (subtract) 2500 clean components


Data description:  these test data are one pointing, D configuration
P band, approximately 5-7 short cuts.  They have gone through all
the standard BPASS, CALIB, lots of editing, etc, etc, although
most of that is irrelevant to the tests.

+++++++++++++++++++++++++++++++++++++

1st test - where does ccmod put its sources?

    I ran SETFC to set boxes (37 fields, 40" cells, 128 imsize,
covering a radius of approx 3deg)
    Using the boxfile from above, ran IMAGR do3d turned on
    Ran CCMOD 3 times, using first three fields of the 37.  Each
time, created a new CC file (vers. 2) with one component in it,
a 1000 Jy sources at pixxy 0,0
    Ran UVSUB with factor=-1 and nmap=3 to insert these 3 sources
(uv dataset TEST2)
    Ran IMAGR using same parameters as one above.  The three
sources do NOT come out in the center of the three fields.
You can see where they come out by going to the web page. The
three sources show up in the SW (same relative position) of
fields 2, 4, and 9, instead of 1,2, and 3

+++++++++++++++++++++++++++++++++++++++++

2nd test - uvsub vs. imagr

     I started over with the original data set (TEST1)
and ran a 2500 iteration IMAGR  with bmaj, bmin =-1
     The resulting image is on the web site.

     I then ran IMAGR again, with niter 2500 and
bmaj, bmin =180
     Then I ran a UVSUB to subtract all components
(uv dataset TEST3)
      IMAGR again, niter=0
     Resulting image is on web site.

      RESULT:  The above two methods do NOT lead to
the same result.

      Then I took the difference between the two above
images, and also did an imvim in a place that had a
lot of mismatches. These are on the web site.

++++++++++++++++++++++++++++++++++++++

Thanks - i look forward to hear what you make of this,
and I apologize again if this is a stupid user error.

lr