[alma-config] the comparison of configurations

[John Conway]

On Tue, 18 Jan 2000, [Leonia Kogan] wrote:

> All,
> 
> Of course comparison with a ring is not correct because a ring can not
> be consider as a good configuration since the beginning.
> 
> In July, August 1999 Me and John Conway carried out simulation of two
> donoughts two circle and a spiral configuration to compare the quality
> of the restored images. We used two source models given by John: M51
> and CYGN.
>
> The donoughts and two circles configurations showed better identity
> with the model and more clean image in comparison with the spiral.
> 
> But I would not like to say forever that a spiral is worse, because the
> parameters of the spiral can be modify and anothe spiral can give
> different result.
>

Right, the difference for the version of the spiral tested in August was
that had too few short spacings compared to the doughnut. This
is corrected  in the latest spiral array by including some baselines down
to 15m (maybe overkill for a 3km configuration!) 
 
In the simulations performed in August, the doughnut therefore gave
ripple-like errors and the spiral had bowl-like errors due to lack of 
short spacings. The rms of errors for the donut was in this case somewhat
lower in those simulations, because the bowl errors  were more significnt.

I havn't yet performed smulations comparing the donut with the new
spiral for the  Cygnus A and M51 images, but I will do so. The reason
incidently I didn't do these simulations for the web page initially with
the 64 element two ring donut Leonia sent me in August was is 
that I didn't know whether that was public yet since it had not 
appeared in any memo. I didn't want to step on any toes with 
testing an array design which wasn't considered by its author 
'final; in some sense. That why I tested it against a simple
ring which belongs to nobody, but it appear I have attacked 
a paper tiger!

 
    John

 
> The last John's simmulation tells us again that the simulation have to
> be done on the planned basis with fixed sources as models, with time
> planned, with wide discussing of the having planned simmulations and
> ....
> 
> We need to have a plan with WHO, WHEN, WHAT collumns.
> 
> Leonia
> 

Well I think the final decision between array designs should be made by
testing them against a range of images we have agreed in. 

No  definite conclusions can be made by single authors making simulations,
since it will always be the case that certain arrays will be better
with certain images. However I don't think this should stop single
authors doing their own simulations to illustrate points of imaging
(like extrapolation vs interpolation needs of arrays) 
that they think should be helpful. The main thing is to get started! 
If we wait for a final organisation plan to be in place we will
have to wait forever.