[mmaimcal] Re: [alma-sw-ssr] mosaicing

[Mark Holdaway]

> 
> Here's another detail which is bothering me. Perhaps someone can
> resolve for me.
> 
> For Nyquist sampling of pointing and uv-data, the phase varies by
> pi for a source at the edge of the field of view; the amplitude
> of that source is attenuated by 2/pi on the Nyquist sampled long
> baseline, by less on the short baselines. In the mosaicing
> algorithms, shouldn't the 'primary' beam correction also include
> a correction for this attenuation ?  Is this one reason why I see
> an increase in image fidelity when I oversample the data ?
> 
> Melvyn.
> 

OK, I believe "Nyquist sampling" is being used in two ways here:
Nyquist sampling of pointings on the sky, and Nyquist sampling 
in the Fourier plane (ie, in integration time such that the long
baseline's phase is not lost).

There are two answers:

1) Use a shorter integration time so that you are oversampled in the (u,v)
or time domain and do not lose much amplitude to time-averaging smearing.

2) AIPS++ treats model components as equal citizens to images
in the synthesis imaging package.  Hence, the sky brightness can be
described as several model components and several images.  While
the effect of bandwidth smearing and time-averaging smearing 
on an image is somewhat involved, it is much simpler to include
these effects when calculating the model visibility from a model
component (code which does not yet exist, but which is quite
easy to see where it goes and how to do it).  AIPS++
synthesis code is sort of like an inverse-problem
machine, and this is a great example of its utility:  you put the
bright sources which would be a problem into model components and
gradually improve your model for them; you calculate the model
visibilities using them, applying the effects of bandwidth and time
averaging in the calculation of the model vis on a per baseline
basis.  The offset between the data and model visibilities tell you
how to tweak your model (ie, components and images) in the next iteration.


    -Mark