[daip] questions about using and getting help with AIPS

[Eric Greisen]

Ray Lucas writes:

 > This email is addressed to both daip at nrao.edu and to Dr. Lorent
 > Sjouwerman since I am not sure which one is the most appropriate
 > one to receive these questions. (My last question has to do with
 > a VLA pipeline.) If it is more appropriate for the other to reply,
 > my apologies to the one I should not have bothered! In any case,
 > I am not sure who is the "designated AIP" at the moment as per
 > the web page at http://www.aoc.nrao.edu/aips/d_aip.html, but I
 > wanted to make sure that both are aware in case a response has
 > to be coordinated, or in case the daip address is no longer used,
 > etc. (Its last update seems to have been in 2004.) Anyway, my
 > apologies if this email should have been directed elsewhere!
 > ;-)

    daip goes to everyone in the extende3d group including Lorant.
There is no specific daip any  more except for me and I do triage and
answer most e-mails.

 > I have a NVO program to reduce some old VLA data that I took
 > in 1990 (yes, that long ago - but its a long story - basically
 > life intervened!) and incorporate it into a new NVO-compliant
 > database of multiwavelength images and parameters, etc.
 > 
 > The program was a continuum snapshot program of about 70+
 > Texas interferometer steep-spectrum sources and several
 > standard VLA calibrator sources mixed in, and the data were
 > taken in A-array configuration. My VLA program in 1990 was
 > AL218, and I was the PI. I recently requested the data (I
 > don't remember whether I requested it in the "original" or
 > newer "AIPS Friendly" style since it has taken me longer to
 > get back to this in recent weeks than I intended) and
 > received it via ftp, but I have not yet tried the task FILLM
 > or done anything with it. As I understand it, the new format
 > is supposed to help more with keeping track of the files more
 > simply by project name, etc. Perhaps I should re-retrieve
 > the data soon, being sure to request the new "AIPS-friendly"
 > version just in case I requested the original style before,
 > though I suspect I would have requested the new style, anyway.
 > In fact, while typing in this email, I have just re-requested
 > it in the AIPS-friendly format, so now I'll know that I have
 > the new versions.

      The friendly version is just 
           some-name_n
      for n = 1 through N for N files.  This lets aips have an
internal do loop over predictable names.  You can name the files by
hand as well...

 > 
 > I've asked our computer folks to install a recent version of
 > AIPS here, and they've taken their shot at it. I think it is
 > one of the last two versions or so, but not absolutely sure
 > without checking some more with them or searching for the
 > details myself, or trying to start it up and seeing if I'm
 > getting the proper new version or an older one that was in
 > place before it. There is someone here who may have some
 > experience with using AIPS, but I'm not sure how much or how
 > recent that experience may be.
 > 
 > So, I've been looking through some documentation - the AIPS
 > Cookbook and some various pages on task descriptions, etc.
 > and the pages on "who to call or email - the "daips" page"
 > and VLA Contacts page - and I have a few questions initially.
 > 
 > 1)
 > If I decide I need some direct help, i.e. need to go to one
 > of the NRAO facilities and work directly and closely with
 > someone who knows what they're doing and can give me expert
 > and frequent assistance or step through the process with me,
 > it sounds as if I could go to Charlottesville, which is much
 > closer for me, but that there is no real local support there
 > other than making sure that the software is working, and that
 > the system is really geared towards having me come out to
 > Socorro for that kind of more labor-intensive or maybe even
 > collaborative help. Is that still correct? (I have no problem
 > with Socorro - I enjoyed my visit there when I came to take
 > the data back in 1990 - its just that Charlottesville is much
 > closer and I can drive there in about 3 hours for the cost of
 > a tank of gas in my car...!)

        I would hope that collaborative help is not really needed.
There are scientists at both Socorro and Charlottesville who might be
pursuaded into that but everyone is busy as usual.  The daip at nrao.edu
will certainly be able to anser some questions within reason.


 > 2)
 > In looking at various tasks and reading about them, I see
 > that the task SCMAP seems to be the preferred one for making
 > the uv maps, and that MAPIT is described as being inferior
 > to SCMAP in general, but that it is particularly suited to
 > snapshot observations. So, I'm not quite clear on this. Is
 > SCMAP or MAPIT the preferred option for reducing/making the
 > uv maps from the snapshot observations if one wants to make
 > the data as good as possible (or at least as good as is
 > "reasonably" possible!) I would like to make these uv maps be
 > as high quality as I can, but don't want to go too far past
 > the point of diminishing returns to the point that I'm putting
 > in a lot of extra effort for very minimal increase in quality,
 > unless there is something in particular that seems especially
 > worth it, perhaps even to be done in a separate and later
 > reduction after the first basic one is done, if that seems an
 > advantageous way to do it, though if possible, I'd also like
 > to just do the best job all at once and be done with it.
 > Anyway, I'd appreciate an expert's opinion and advice on
 > the best thing to do here in terms of which would be the best
 > way to proceed, and also on which would be the best task to
 > use in this kind of case, which may have some higher-than-normal
 > profile given that it may be used in a kind of NVO prototype.)
 > 
    Self-cal is usually done with a sequence of CALIB and IMAGR but
SCMAP or SCIMG do incorporate some extra intelligence to do the
process inside one task.  MAPIT is an old, no-longer actively
maintained procedure that uses an older, less desirable imaging task
named MX.

 > 3)
 > Is there a sort of detailed (but still somewhat high-level)
 > step-wise procedure documented for making uv maps from snapshots
 > in the preferred manner (i.e. that best combination of quality
 > and reasonable level of effort that I'm talking about)? It may
 > just be that I've somehow overlooked it, but I haven't managed
 > to find something like that yet. (The AIPS Cookbook Table of
 > Contents provides clues somewhat like this, I guess, but its
 > still not the thing in exactly the format I'm really wanting to
 > find.)

    The CookBook is an attempt at it - chapters 4 and 5 mainly for
calibration and imaging.

 > 
 > 4)
 > I also see from the "Contact Info for VLA" web page that there
 > is a VLA pipeline. I had only seen something about VLBA pipeline
 > before... Does this mean that there may be a standard pipeline
 > for processing my data now, producing output products that are
 > calibrated uv maps of my sources from the snapshot observations?
 > If so, where can I find out more about this? And may I use it?
 > If like some other standard data pipelines, I can understand that
 > it might have some limitations which, if one wants the very best
 > output and end data products, might necessitate re-running with
 > more tweaks or updates or more manual intervention, but it would
 > still be an interesting/nice idea for at least a starting point...
 > (For this reason, I am cc'ing Lorant Sjouwerman at the VLA AOC
 > as well as daips at nrao.org, as per the VLA Contact Info Web Page.)

There is a run file called VLARUN which has very recently been updated
(only a current 31DEC06 version of aips would have it) to be quite
powerful.  I do not know - it depends a lot on the data - whether the
results will be publ;ishable quality (although they will be as good as
much of what is on the VO).  For this you really have to have the
latest aips - installation is pretty easy - we now have binary
versions available for Mac and Linux and Solaris and will soon have
Mac Intel.

Eric Greisen