[daip] AIPS calibration of VLA data

[Eric Greisen]

Data Analysts writes:
 > 
 > ----- Begin Included Message -----
 > 
 > >From fich at astro.uwaterloo.ca Thu Aug 24 14:41 MDT 2000
 > Date: Thu, 24 Aug 2000 16:37:20 -0400 (EDT)
 > From: Michel Fich  <fich at astro.uwaterloo.ca>
 > To: analysts at cv3.cv.nrao.edu
 > Subject: AIPS calibration of VLA data
 > MIME-Version: 1.0
 > X-Lines: 39
 > Status: RO
 > 
 > HI,  I must be forgetting even more than I thought.... I am having a lot
 > of trouble calibrating what should be easy data.  ...
 > 
 > I have two programs - one really old (AF141 - 23Apr87,  D array,
 > snapshots in C and L band, about 35 objects in each. don't need
 > polarization info) and one relatively new (AF346 - Jul and Dec 98, B
 > array, snapshots in C and L - same objects.  to be combined with
 > the old data to improve resolution).
 > 
 > I am running 15APR98 version of AIPS (I tried to upgrade to 31DEC99
 > but had problems - and couldn't get a response from AIPS people...
 > well to be fair I got a response to my first email, but then my reply to
 > their reply was ignored...)

     We do try to respond - if you do not get a response in a few days
you should try again.  We assume that if you go away and stay away
that you have solved your problem or no longer care.  Many people do
solve their own problems and negelect to tell us - our assumption is
reasonable.

I delete messages received by daip if and only if a reply that should
have been acceptible was sent.  In some cases - as below, a simple
request for more information.  I find one old message from you on 2
February, 2000.   We have sent you I am sure an adequate reply to that
one but if not I will repeat it.  The shared memory id failure arises
from one of two causes.  The first is that the program was trying to
share memory between two separate computers.  We catch this under
normal circumstances and prevent the failure, but ssh can fool us by
claiming that the remote machine is <local_machine>:10.0 or some such.
If you are sitting at the console of the computer on which you are
computing, that will not be the problem.  The second reason - on Sun
computers - is that the default maximum shared memory segment is too
small for a single display plane.  Your display plane is 1270x924
pixels = 1173480 bytes or 1.119 Mbytes.  The default maximum is
usually 1.0 on suns.  The file that contains this info is /etc/system
in which lines such as

* From Dave Brown 1997.04.22 reduced to 4MB 1998.04.21 by dbrown
*                            increased to 10MB
set shmsys:shminfo_shmmax=10485760

should be inserted.

 > 
 > I have been following to most recent cookbook I can find (15OCT99)

      See http://www.cv.nrao.edu/aips/

and follow a path to the documentation and cookbook ToC from which you
mat select chapters.  Appendices A and B have not needed update in
some time.

 > but I don't seem to make any progress ....  I have been through the
 > simplified version of the calibration process described in the
 > appendix, but when I finish and do a UVPLT on particular calibrators
 > I get what appears to be complete garbage...
 > 
 > I know the data is reasonable since I calibrated (and published)
 > the 1987 data long ago - but my old tapes are unreadable so I
 > have the data fresh out of the archives again.
 > 
 > I know I haven't given you enough to go with yet, but I am kind of stuck
 > on even where to start at this point.
 > 

    The analysts are not supposed to know much about aips and cannot
be expected to help you.  On the other hand, with this level of info,
neither can I.  What do the data look like in UVPLT without applying
the calibration you determined?  What does SNPLT on the final CL table
used show?  Etc Etc  Being told that you have done a long and messy
process and the result is garbage does not help without assertions
that the initial data - as now loaded on your disk - are ok and that
the calibrations were ok (messages from CALIB and CLCAL are helpful),
and so on.  The aips group has dwindled mostly just to me and that
means I have little time to extract this sort of information from
people who do not provide it...

Eric Greisen