[daip] Re: AIPS Multiscale Question

[Eric Greisen]

Jason Peterson writes:

 > I have been working with the multiscale parameters in IMAGR in order to
 > try to extract some of the extended flux from the negative bowl
 > surrounding some edge-on galaxies.  I have done this successfully with
 > some archival data of NGC 4631 and am currently trying this with NGC
 > 891.  Curiously, I've found that the standard Clark clean seems to work
 > better with NGC 891 while the multiscale seems to leave more of the
 > negative "halo" around the galaxy.  I know it's not very probable to be
 > able to diagnose what's going on without further details, so if I took
 > some time up at the AOC would you or someone else familiar with this
 > aspect of IMAGR be available for (extensive) questioning?
 > 

If you come by, of course we will try to help.  The level of knowledge
about this algorithm is however somewhat lacking.  In fact, a user had
to recently point out that it was not working due to a typo I made.
The IMAGRPRM(11) function, which is critical, was essentially out of
action for a year.

I recently worked on the algorithm and had considerable success in
removing the bowl if I cleaned very slowly with several resolutions
each a factor of sqrt(10) greater than the previous.  But when I made
the Clean faster the bowl was more evident.  I suspect that the source
structure itself will affect the performance of the algorithm - since
we are trying to cover up for data we did not take.

There is also aips++ in which this algorithm is implemented and you
might get different results with their version.  I would recommend
coming here to get help trying that.

I am going to be away May 17-June 12.

Eric Greisen