[alma-config] Re: SIMs + Mels comments

[John Conway]

On Thu, 7 Sep 2000, Steven wrote:

> The B and C arrays for all declinations and A array for +25 degrees have
> been CLEANed to 10000 iterations. The other declinations for A will take16
> hours per declination, so this should be achieved by the weekend. If there
> is a consensus on what kind of difference images etc. to calculate, Ican
> start to implement that now and have loads of results with you verysoon.
>
> Cheers,
>     Steven
>

Hi Steve et al 

 Good to hear about progress. I think you should go ahead and produce
difference images based on the subtraction of CLEAN image - model
convolved with the same beam. This will give some quick-look estimate of
which array has the largest reconstruction errors etc and what form these
reconstraution errors are (ripples, missing short spacings etc).
Furthermore  we need the difference to later compute the more complex
fractional error maps. Having the intermediate step of difference
images to peruse as quickly as  possible will be useful.

The more complex fractional error estimates are definitely 
required for a  more quanititative estimate of Fidelity 
indices/dynamic ranges etc. As discussed in earlier emails and telecon
I think we are all agreed we need  a measure of both the fractional errors
on-source and off-source.  I guess we should try to come to a consensus
soon on which of the measures that have been proposed to adopt in
practice. Seeing the pure error images will help in making a sensible
choice.

   Cheers
      JoHn


P.S Mels points of about a week ago. Its clear that we going to have 
to do simulations of mosaiced observation before finalising the array
design. The present simulations all use  test images which are smaller
than the primary beam (i.e. Cygnus A has been scaled in size to be 
much smaller than the 2' the real source extends on the sky) and
are  single pointing simulations.

As an uneducted  guess I'd imagine that virtually all of E,D  and perhaps
50% of C array  observations will be mosaiced. However most of A and B and
50% of C probably will not be. I therefore believe that the single
pointing  simaulations we are doing are a useful first stage. I understand
that  the experience on BIMA and OVRO has shown the importance of short
spacings and mosiacing especially for galactic work. Such observations
will also be important for ALMA. On the other hand
ALMA is going to have a lot more sensitivity than present mm arrays
and so I believe  higher resolution observations which are not useful on
present mm-arrays because of brightness sensitivity limitations will be
more important in the ALMA era. I guess  looking at circumnuclear gas in
AGNs or at  protoplanetary disks  are two important science drivers that
will primarily use A, B and C arrays in single pointing mode.

I think the simulations that we are doing wil tell us something
about the dynmamic range limits just due to pure uv coverage 
effects in single pointing mode which then can be compared to other
sources of errors  such as pointing or phase-calibration; they will also
tell us  whether (as I suspect) both strawperson designs could benifit
from  more short spacing coverage. I believe its better to do some
simulations covering part of the observing modes of ALMA as soon
as posible and we can do the others as soon as the software for 
running them is ready. However I suspect that the difference between 
the two strawperson designs for moscaiced observations 
will bot be signifcant; after all the share identical E-array 
designs. So I doubt whether they be decisive for the issue of fixed
array versus zoom (although maybe I'm wrong about this, I'm not that
familar with mosaiced observing though I've read the papers). 
Obviously though that E array design should  be tested in
mosiaced observations to see that in itself it good one.



 


On Thu, 7 Sep 2000, Steven wrote:

> The B and C arrays for all declinations and A array for +25 degrees have
> been CLEANed to 10000 iterations. The other declinations for A will take 16
> hours per declination, so this should be achieved by the weekend. If there
> is a consensus on what kind of difference images etc. to calculate, I can
> start to implement that now and have loads of results with you very soon.
> 
> Cheers,
>     Steven
> i