[mmaimcal]Re: [Almasci] Comment in ALMA memo 489

[Mark Holdaway]

Jerome,

> 
> Dear Mark,
> 
> I fully support Koh-Ichiro's and Tak's remarks. Here are a few precisions:
> 
> 1. The different acquisition schemes of SD data between ALMA+SD and
>    ALMA+ACA+SD come from the fact that the SD have the same diameter than
>    ALMA antennas while a different diameter compared to ACA antennas.
> 
>    The processing of the SD data is to try to compute the visibilities an
>    interferometer like ALMA would observe at spacings smaller than 12m in a
>    thought world. When the SD and ALMA antenna are identical, I claim that
>    this is possible only for the zero-spacing (for other spacings, SD is
>    not giving enough information). Thus there is no need to make an OTF
>    map. Indeed, making an OTF map in such a case would dilute SNR over
>    large area when exactly only the same small amount of information would
>    be used.

Any time you take multiple Single Dish pointings on the sky, you
are obtaining information on spacings besides the zero spacing
on part of the source.  For an extended source, you can
get the zero-spacing information only by combining all the pointings --
but by combining all the pointings, you are also getting,
at least implicitly, the information on other baselines as well.

I don't agree with your claim that you can get only the zero-spacing
information when doing homogeneous array mosaicing.

>    Now, this use of the SD may not be the cleverest one. From what I
>    understand of MEM, people are using the ALMA dirty map and the SD map as
>    boundary conditions to find the best solution according to some
>    criterion. In this case, the SD information (from 0 to 12m) is maybe
>    better used. In principle, CLEAN could use a similar mechanism, ie try
>    to find CLEAN components both on a SD image and the ALMA dirty
>    map. However, I do not know about such an algorithm but I know that
>    similar algorithms are very difficult to stabilize (see the algorithm we
>    use for deconvolving ALMA+ACA+SD).
> 
>    In other term, we priviliged a very robust algorithm compared to an
>    hypothetical, still to be demonstrated better use of the SD time. We
>    must remember that we want a solution which is reliable (ie we do not
>    want have any doubt that in some cases our algorithms may fail).

I suggest that you have an effective algorithm for dealing with the +ACA
data case, and that you are dealing with the homogeneous array (ALMA+SD)
case less effectively.  I am calling for a fair comparison.