[daip] scaled peeling

[Eric Greisen]

Adam Deller wrote:
> Hi Eric,
> 
> I have a data reduction problem that I thought you might be interested 
> in.  There is a faint pulsar target separated by 3" from a very bright 
> source.  The observation is VLBA plus phased EVLA, and the phased EVLA 
> synthesized beam is < 3".  So the bright source is in the primary beam 
> for the VLBA, and in the (messy) sidelobes for the EVLA.
> 
> No problem - we'll just peel the bright source, right?  Only thing is, I 
> have a dataset where pulsar gating was using to boost S/N on the target. 
>  The flip side is, in this dataset, S/N is correspondingly reduced on 
> the bright source.  The correlator is blanked for 96% of the time due to 
> the gating, so the S/N change is a factor of 5.  So if I just run a peel 
> on the gated dataset, the S/N on the bright source is not good.
> 
> However, I also have an ungated dataset (so full S/N on the bright 
> source, reduced S/N on the faint target).  So what I'd like to do is 
> *derive* the peeling corrections on the ungated dataset, but *apply* 
> them (and peel out the bright source) on the gated dataset.
> 
> Since the difference between gated and ungated is simply a constant 
> amplitude scaling factor as far as the bright source is concerned, and I 
> know this factor because I know the width of the pulsar gate, this 
> should be easy enough in principle.  Take my model for the bright 
> source, run a peel on the ungated dataset, divide the flux in the bright 
> source model by 1/0.04 = 25, and run peel again on the gated dataset 
> using this model *and the solutions already derived from the ungated 
> dataset*.  That last bit is the bit that is missing, I think.  Is there 
> an easy way to provide PEEL with precooked solution tables?  Or if not, 
> might it be easy to make this an option?

Peeling in AIPS is done with a procedure which could easily be modified.

However, what it does is find the self-cal for the peeled source and 
then restore that source to the data set.  I gather you just want it gone.

I do not understand the "scaling".  The source in X Jy independent of 
whether it is observed 1 msec out of 100 or 100 msec out of 100.

The PEELER script takes an image as input and that could certainly come 
from the full time data set.  If you did it by hand, then one could put 
together something that also used an SN table developed from the one 
data set but used in the other.

I attach the PEELER script which is where we would start...  Take a look.

BTW - I think you were interested in the task that corrects for pointing 
and phase stopping positions being offset.  That is called CLVLB and is 
available in the 31DEC13 AIPS.  Enno and I did not have the happeist 
results with ot in his test data set.  He got what he claimed were 
better results with a handling of the geometry that I think is not right 
  (and he sort of agrees).  The task has both methods plus debug print 
options.

Cheers

Eric

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: PEELR.001
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20130626/ac650be6/attachment.ksh>