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

[Stephane Guilloteau]

more ...
----- Original Message -----
From: "mel wright 456" <wright at astron.berkeley.edu>
To: "Stephane Guilloteau" <Stephane.Guilloteau at obs.u-bordeaux1.fr>
Cc: "Mark Holdaway" <mholdawa at tuc.nrao.edu>; <almasci at nrao.edu>; <mmaimcal at nrao.edu>
Sent: Tuesday, April 06, 2004 8:42 PM
Subject: Re: [Almasci] Re: [mmaimcal]Re: Comment in ALMA memo 489


> Hello Stephane,
>
> I have compared 3 approaches for deconvolving with single dish and
> interferometer data  with the same or different primary beams: i) constraining  the total
> flux, ii) using a default image, and iii) joint deconvolution.
>
> The joint deconvolution almost always gives the best image fidelity,
> which one can  understand by watching the convergence; the deconvolutions of different
> sized structures assist each other in iteratively converging to the true source
> distribution.
>
> I used the MEM algoriths in MIRIAD, but the same 3 approaches can also
> be done using CLEAN. Joint deconvolution of single dish (using the single dish primary
> beam(s)),  and interferometer array(s) using their synthesised beam(s) could be
> done using a multiresolution clean algorithm.
>
    I tried that with multi-resolution CLEAN. This is actually the basis for the joint deconvolution of ACA + ALMA + SD.
It does not work so well with ALMA + SD because of the large difference in scale between both data sets. But, as Mark
pointed out, the method was put rather quickly when studies of the ACA were requested by the ASAC. It may well be worth
trying to improve on the ALMA + SD case with this approach.


> MEM gives nice results, but is quite time consuming.  It would be nice
> to have an objective comparison with clean based algorithms.
>
    Yes, and perhaps an hybrid method would even be better.

        Regards,

            Stephane