[alma-config] Re: CONFI
John Conway
jconway at oso.chalmers.se
Mon Mar 6 10:13:17 EST 2000
Hi Leonia,
Let me know how I get the updated CONFI code, I'd like
to try it.
Whateveer the array 'style', rings, spirals, whatever
ultimately it might be interesting to optimise with both
beam metrics and uv metrics (a la Fomalont) or a mixture.
These metrics should also probably also force some degree
of natural uv taper and forcing of short baselines, since from
imaging simulations these both seem to be desirable
properties,
I guess one way of looking at minismising far sidelobes
in CONFI is that it is producing uv coverages which don't have
periodicities (such periodicites are obvioulsy bad
because these then allow into the final image false unseen
distributions which correspond to compact structure). As you alluded
to in the telecon if one uses some over simple uv metric like
fraction of uv cells sampled then you could imagine
getting uv coverages with the same score in the uv metric but
one of the two having much higher dirty beam sidelobes. Making the
uv metric more complex can obviously mitigate this problem
but it seems natural that if one is trying to reduce periodicties
in the uv coverage then looking at the Fourier transform of this
uv coverage (i.e dirty beam) would be a natural thing to do(!).
The other practical advantage of the sidelobe minimisation is
that you have devised an efficent optimisation algorithm.
Whether a dirty beam minimisation is the only/best thing one should
aim for is a different question. I'm sure we will talk a lot about
metrics again in Tuscon, my position is pragmatic, we can write down
half a dozen properties of the uv coverage/beam that are
desirable, some of them hard to have simultaneously, deciding the
balance between these different properties needs some imaging
simulations as a 'reality check'.
As far as running the new CONFI here we don't have a midnight job or
anything like that here (and I understand there will be no more
offcial classic AIPS releases), could you put the .FOR
code on an anoymous site somewhere? In short whats the
best way to get it?
Cheers
John.
P.S Where can I get the ASCII version of the masks?
I had a quick look at Bryans web page but I didn't see
it there yet.
On Fri, 3 Mar 2000, [Leonia Kogan] wrote:
> John,
>
> As you know CONFI (AIPS task) can read the topography information as
> an input file written in ASCII.
> The original file prepared by Bryan was in fits format.
> Recentely he created the files in ASCII format also.
>
> These files are much bigger than the Holdaway's file I have used early.
> So I modifyed CONFI to read the Bryan's ASCII files.
>
> I have changed some defaults of CONFI in accordance of last development
> of ALMA.
>
> I added the new parameters BLC, TRC to allow using a portion of the mask
> for smaller configuration.
>
> The new version of CONFI is in the AIPS system now.
>
> Be ready to see that CONFI works rather slow with the huge file.
>
> Leonia
>
More information about the Alma-config
mailing list