[alma-config] Re:

[Ed Fomalont]

Hello all,

    I have just read Mark's and Woody's comments on the use of the PSF
(with some primary beam correction factor) as a metric of the
configuration.  The problem is that the PSF is too 'point' source
oriented (by definition).  If the field contains a relatively small
number of compact sources, then we know that cleaning (or whatever
deconvolution you use) works very well, and you can often get down to
a level which is <10% of the peak sidelobe level if the SNR and the
phase calibration permit this.  So, I think using the peak sidelobe as
a limit is too pessimistic.  Also, I wouldn't multiply the PSF with
the Primary beam, because the strongest object could be near the
half-power point.

    A more major concern is that we may really want to optimize the
configuration for extended sources.  As you know the sidelobes
associated with an extended source are much worse than indicated by
the PSF.  I'm guessing here, but I think the worst sidelobes with an
extended source are not from highest sidelobe peaks in the PSF (which
tend to be small in angular size), but from the somewhat lower level,
but more extended rolling sidelobes that are associated with missing
short spacings or missing wedges in the u-v plane.  In order to
determine the sidelobe pattern for an extended source, you need only
taper the u-v coverage (or convolve the PSR) with a function which
mimics the extended source.  As an example you might use a taper which
drops off to ~1/4 power at about the half-way point of the u-v
coverage to simulate something which is big, but not too big.  Then,
the properties of this 'resolved' PSF as well as the true PSF may give
a better indication of the configuration behavior for extended as well
as point sources.

    Cheers,  Ed