[alma-config] Wholly positive model results

[Mark Holdaway]

My assessment of the Jan 18 simulations:


C array
============================================================
M51:  Spiral is pretty much always superior

Cyg A: two arrays are very similar (simple source?)
The one exception is -23 deg dec simulations; why
does the donut array choke?

SDC-23 Snaphshot: the error image indicates that is some sort of box
in the processing??  Seen to a lesser extent in the other SDC 
sims.
Otherwise, SDC sims are pretty similar for the two arrays.

MPD: imaging works similar for the two arrays for snapshot,
but the spiral array is superior for 4hr tracks

MAR:  spiral is superior; Donut is still missing short spacings,
      and the gap between the two arrays' imaging quality seems to
     be worse when there is no foreshortening (ie, in the snapshot
    imaging). (unfair comparison?)




B array
========================================================================
M51:	+25 S	identical
	+25 4	spiral wins
	-23 S	identical
	-23 4	spiral wins
	-70 S	identical
	-70 4	spiral wins

Q: What is limiting the images for snapshot obs?
I.E., why are the results so similar for the two arrays?

	===> hey, its the NS sidelobes adding up for long track
	observation in the donut array that makes it loose
	This is probably true also for Cyg A and SDC as well.

Cyg A:	+25 S	identical
	+25 4	spiral wins
	-23 S	very similar
	-23 4	spiral wins
	-70 S	similar; spiral slight edge
	-70 4	similar, but spiral wins

	pretty similar; for B array, maybe its is the spiral
	array's improved short spacings that help


	
SDC	+25 S	same
	+25 4	spiral wins
	-23 S	donut wins
	-23 4	similar; donut wins??
	-70 S	similar
	-70 4	spiral wins

As with the C array, there appears to be some box around this source
that is resulting in reconstruction errors.  



	
MPD	spiral array wins in all cases.  Large object, short baseline
	coverage in spiral array wins the day.




MARS	These simulations are affected significantly by the
	lack of short spacings for both arrays, so I would
	say interpretation is meaningless.


====================================================================

IN MY OPINION (not humble, as indicated by the CAPS and lack of the word
"humble"; intended to imply the voice of someone larger than me):

All in all, there is usually not a huge difference in the imaging quality
between the two candidate arrays.  We get examples of the things we
already knew about (N-S sidelobes causing problems for the donut array in
long integrations, short baseline problems with the donut array). The
spiral array seems superior, mainly for the two reasons just mentioned.  

Kogan takes issue with the short baseline point, and I would say it is
very similar, the flip side even, to the issue that the donut array has
higher resolution: if you want higher resolution, add a larger array;  if
you want imaging of more extended structure, add a smaller array. It's not
quite that simple, because lack of short spacings can wreck an observation
while lack of long spacings just blurs it a bit.  BUT, I would say that
neither of those arguments is fundamental, and both should be neglected.  
I would argue that Conway's N-S sidelobes argument IS fundamental, which
favors the spiral array.

However, it seems that the operations considerations, which also seem to
favor the spiral configuration, might be more significant than the
difference in imaging capabilities.

	-Mark