[alma-config] telecon this Wednesday

[Min Yun]

> 
> I think the near-in sidelobe levels reported by Morita for the
> spiral array are in error -- I believe they are the main beam's
> wings rather than sidelobes -- this may be relevant or irrelevant,
> I don't know.  I suppose it would make more sense to look at the
> deviation from a Gaussian.  While it is true that the near-in sidelobes
> of the spiral configuration will not be reduced by earth rotation,
> I contend that they start out much smaller than the ring or donut
> configurations near in sidelobes.
This raises an interesting issue -- do you really want a PSF that has
a long tail?  I remember trying to combine the VLA C and D array
data in a "dumb" way, and the resulting PSF with a plateau posed a
major problem in deconvolution, not surprisingly.  Even when not
being so dumb, what one really wants is a truncated Gaussian rather than
honest goodness Gaussian, isn't it?

> Similarly, I contend that if Leonia did an optimiazation over a long
> integration, the outer sidelobes would be small due to earth rotation
> averaging down, so the algorithm would concentrate more on reducing the
> inner sidelobes via structural means, ie, by moving antennas.
Leonia's routine in practice only works on the near-in sidelobes.  You can
make the optimization region as large as you want, but this drives
you do the solution space which is not particularly unique.  Again,
this is why my current strawperson donut configurations have only 5% 
sidelobes rather than "bigger than 10%" number routinely thrown around.

> > No one has yet to offer a concrete reason why Leonia's snapshot 
> > optimization is flawed.  While earth rotation clearly has an impact
> > on the imaging quality due to changes in the far-sidelobe levels,
> > I don't know of any telescope TAC that has yet to grant extra observing
> > time so the imaging quantity can be improved by a factor of 2.  
> 
> Not in so many words, but I think long tracks are often granted at the
> VLA, not for sensitivity, but for imaging quality.  But it is done with a
> wave of hands, not by looking in tables or on graphs.
ALL VLA proposals I have been associated with over the last 15 years
are strictly limited by sensitivity and not by imaging quality (incidentally 
several of the projects had to worry about DR in the end, however).  All
of the OVRO proposals I refereed for 4 years justify their time
request on sensitivity -- exceedingly few on imaging quality.  ALMA
will be much closer to the VLA, of course, but it won't change how
people do science.

> 
> OK, there are three issues for hour angle range:
> 	1) fraction of filled cells, which translates to image quality
> 	2) sidelobe level, which translates to image quality
> 	3) noise as a function of elevation angle, which translates
> 		to sensitivity as a function of observing time.
> 
> Memo 201 did not say anything about fraction of occupied cells, it looked
> at hour angle ranges as a function of declination and observing frequency
> (I think...), considering the increase in noise due to atmospheric
> opacity.  So, that would be the relevant MAXIMUM hour angle range for
> very long integrations (set by SENSITIVITY requirements...ie, multiple
> transits).  For the smaller arrays, we may CHOOSE to optimize for
> less than the maximum hour angle range, as we may say that snapshots
> are more important.
I think you have already made a good case on the sensitivity,
and there is no need to revisit the issue.  I hope we will learn 
more about the first and second point by examining the strawperson 
arrays for objects at different declination and different amounts
of HA coverage.  Something to keep in mind during our simulation study.
Optimizing the array design based on what we may learn is something
desirable.  I am not too concerned based on what I have learned thus
far working on the array design and am not too optimistic that we will
have a lot of time or energy to get this done properly, given the project
time frame.




					-- Min