[daip] Re: CVEL anomaly in VLA processing.

[Eric Greisen]

Lincoln Greenhill writes:

 > I am having trouble with either CVEL. I think the problem may lie in
 > FILLM's processing of what is a very complicated VLA experiment. 
 > 
 > I'd like to run by you my observations and my hypotheses.  Then you can 
 > tell me if I am barking up the wrong tree (e.g., AIPS cannot do what
 > I suggest it might be doing).

     I know very little about all the modes of the VLA and even less
about what FILLM may do with them.  It is entirely possible for FILLM,
UVCOP, etc to have errors in subtleties such as upper vs lower
sideband.

 > 
 > Because of the proposal deadline Monday, I am in a rush and hope that you 
 > can offer at least an initial assessment.

     Given that you observed 2 years ago in a difficult experiment, I
cannot treat your emergency as my emergency...

 > 
 > Background -
 > 
 > I have loaded VLA spectral line data (AG578, 2001 Jan.) 
 > using FILLM 31DEC002. The frequency setup for the experiment 
 > involved 10 different line and continuum setups.  Each setup 
 > used mode 2AB and the PT-link.  I observed the lines with fixed 
 > frequencies.
 > 
 > FILLM generates 8 output files as it attempts to organize the 
 > data.  I am processing 1 file that has 2 IFs.
 > 
 > I copy scans from this file to 4 output files so that each has only 
 > 1 IF and 1 frequency setup within it.
 > 
 > In three files, the lines in each IF pair are made stationary when I run
 > CVEL. In one file, to make the line stationary, I must change the sign of
 > the channel shift in CVEL.

    Are you saying you ran a modified CVEL.FOR/CVEL.EXE or are you
referring to one of the adverbs?

 > 
 > I hypothesize that the frequency axis of this one IF is backwards.
 > I note that sideband flags, as shown by LISTR (optype=scan) 
 > are +1 in all files.
 > 
 > Can FILLM do that by accident?  I note that FILLM showed another
 > anomaly when I loaded the data.  It divided information between 
 > the FQ and SU tables such that for one IF in one output file, AIPS
 > could not reconstruct the proper LO frequency.  I do not think this is 
 > related, but it demonstrates that FILLM was strained by the organization 
 > of the experiment.

I need a lot more documentation - e.g. the exact observation dates,
the adverbs to FILLM, the background to the remark about LO frequency
(I think AIPS only deals with observing frequency outside such
programs as FILLM that may have to construct it)

Eric Greisen