[mmaimcal]Y+ simulation results

[Mark Holdaway]

I have some results on the Y+ configuration simulations
for the Debris disk.

I used a 1024 square representation of the Debris disk and
munged the cell size (made it smaller, farther away) so that
it would fit with the 12 mas resolution that the Y+ array
was getting.

In order to get a good image, I needed to add more short baselines!
To do this, I used 60 seconds of data from John's 3rd most compact
spiral array and 600 seconds of data from John'2 most extended spiral
array, adding this data to the RING or Y+ array data (which had an
Hour Angle range of +/- 0.2, 0.5, 1.0, 2.0, or 4.0 hours).
 
Furthermore, I used a kludged multi-resolution scheme so I
wouldn't have to rely on CLEAN to clean all the emission:
I imaged the low resolution data with MEM, regridded to the
full resolution image size, subtracted that image from the
FULL VIS SET (all 3 configs) in the Fourier plane, imaged the
residual visibilities with CLEAN, and the added the regridded
MEM and Clean component images, convolved with the restoring beam,
and finally added in the residuals.

I have then calculated the image plane fidelity, much as we did
with the ACA simulations, with different pixel flux cut offs.
The Debris Disk is dominated by the fairly bright and compact inner
structure, which was imaged pretty well.  In order to see more
differences, I went down to the 0.1% level for the graphs I've included,
though the trends are also similar for the 0.3%, 1%, and  3% of
the peak pixel value cutoffs.  The caveat is that what you see in
these graphs is dominated by the imaging of the weak, extended 
stucture (all cases had the same short configuration data added in,
though).


Configurations simulated:

The Loose Y (presented last month on web page)

The Tight Y (presented last month on web page)

Angel's STRICT Y (it actually lies on three straight lines)

John's Ring


John found some fault with my Y configs, as they left holes
in the coverage, so I ran Boone's software again, this time permitting
the algorithm to place 36 antennas anywhere within the LOOSE or
TIGHT masks in ONE GO (called FullLoose or FullTight).  THESE configs
should be better optimized, I would have thought, but NO!

The first plot, LOOSE_TIGHT.PS, indicates:
	* Tight is better than Loose  (surprise)
	* The end point of the incremental configs is better
		than the Full optimized configs
		(ie, the algorithm DOESN'T know best)

The second plot, Y_RING.PS, indicates:

	* Loose is better than FullLoose (as above)
	* The RING array is better for the shortest snapshots,
		but quickly falls behind the other arrays.

THIS WAS THE POINT OF THESE SIMULATIONS, IN MY MIND, aAND THE
RING ARRAY'S ADVANTAGES SEEM TO BE INSIGNIFICANT


The thrid plot, Y_STRICT, compares the Loose, Tight, and STRICT
Y configs, and shows:

	* A straight line Y config sucks bad
	* Tight is paradoxically better than Loose
	  (but not by much)



This seems to be good news, especially if I can understand WHY
the results come out this way.  (MORE WORK TO DO)
The TIGHT Y will be significantly cheaper to build, I think,
than the loose Y.



	-Mark


-------------- next part --------------
A non-text attachment was scrubbed...
Name: LOOSE_TIGHT.PS
Type: application/postscript
Size: 19973 bytes
Desc: 
URL: <http://listmgr.nrao.edu/pipermail/mmaimcal/attachments/20020709/105cd9cd/attachment.ai>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Y_RING.PS
Type: application/postscript
Size: 17800 bytes
Desc: 
URL: <http://listmgr.nrao.edu/pipermail/mmaimcal/attachments/20020709/105cd9cd/attachment-0001.ai>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Y_STRICT.PS
Type: application/postscript
Size: 17213 bytes
Desc: 
URL: <http://listmgr.nrao.edu/pipermail/mmaimcal/attachments/20020709/105cd9cd/attachment-0002.ai>