[alma-config]My point of view

[Mark Holdaway]

> 
> effectively, after deconvolution, we *do* divide by the PB, in order to
> estimate *true* source flux density.  this is only in the case of single
> field imaging.  in AIPSese, this is effected by running PBCOR.  you can
> consider this as either lowering the effective signal, or increasing the
> effective noise.  they are equivalent in their effect on SNR of course...

I think there are two relevant facts to remind:
	1) That division is post-deconvolution
           (ie, so we should not weight up the sidelobes out their
            in the optimization process -- if it were a pre-deconvolution
	    deivision, then we would be VERY concerned with the outer
            sidelobes and would have to give them extra weight in the
            optimization).

	2) The compact configuration will be used in mosaicing
           observations like 86.3% of the time.  When mosaicing,
           you don't divide by the PB, but multiply by it,
	   PRE-DECONVOLUTION, thereby reducing the impact that 
           outer sidelobes have on the deconvolution process.


Wind, wind, wind.

	-M