[alma-config] simulations and metrics

[Leonia Kogan]

Mark,

I absolutely agree with you that different algorithms can give different result
for different images. I'd like to add that different people can get different result also. So comparing results we need to be sure that at least parameters 
of the relevant task are selected identically.

Leonia
----- Begin Included Message -----

>From owner-alma-config at kochab.cv.nrao.edu Thu Jan 20 11:22 MST 2000
X-Authentication-Warning: water.tuc.nrao.edu: mholdawa owned process doing -bs
Date: Thu, 20 Jan 2000 11:22:37 -0700 (MST)
From: Mark Holdaway <mholdawa at cv3.cv.nrao.edu>
X-Sender: mholdawa at water.tuc.nrao.edu
To: Adrian Webster <asw at roe.ac.uk>
cc: "Alma Config (E-mail)" <alma-config at cv3.cv.nrao.edu>
Subject: [alma-config] simulations and metrics
In-Reply-To: <Pine.LNX.4.10.10001200831400.8586-100000 at water.tuc.nrao.edu>
MIME-Version: 1.0
Sender: owner-alma-config at kochab.cv.nrao.edu
Precedence: bulk
X-Lines: 23
Status: RO


So far, the simulations people have been doing have just centered
on the standard CLEAN algorithm.  Be aware that different imaging
algorithms will have drastically different results for different
sorts of images.  For example, a maximum emptiness or maximum entropy
algorithm will have much smaller on-source errors for the reconstruction
of an extended source, but much larger off-source errors, than the
standard CLEAN.  A similar statement may be true about multi-scale CLEAN.

We probably don't want to design an array configuration around a single
algorithm.  SO, just as we want a diversity of images in the simulation
library, we also want a diversity of imaging algorithms and a diversity of
image metrics to judge the success of the simulations.

Of course, out of this diversity, we want to be able to make simple
and clear statements about the simulation results and the quality of
the configurations.  That is our challenge.

Take care,

	-Mark




----- End Included Message -----