[daip] CALIB with model

[Eric Greisen]

Olaf Wucknitz writes:

 > But the results do not change if I SETJY any value in the SU table for
 > the primary calibrator! This does work for point sources but not if I
 > use the model. I had a look at CALIB.FOR to understand what is done
 > there and found the subroutine SETFAC which seems to be responsible
 > for rescaling the model. If I understand this correctly (and I am
 > anything but an expert in reading AIPS code!), the routine looks for
 > the header key SUMFLUX and if it finds it, it does the rescaling with
 > the SU entry.
 > But my model files do not have this SUMFLUX! 

    I have dropped this SUNFLUX which was based on a bad idea of how
modeling should be done.

 > 
 > A typical CALIB call produces this:
 > 
 > alf.as> CALIB1: Task CALIB  (release of 31DEC02) begins
 > alf.as> CALIB1: CALIB USING 20030601     . U BAND .   1 DISK=  1 USID=3062
 > alf.as> CALIB1: Selecting, editing and calibrating the data
 > alf.as> CALIB1: Doing self-cal mode with CC model
 > alf.as> CALIB1: SETGDS: imaging done with reprojected tangent point(s)
 > alf.as> CALIB1: VISDFT: Begin DFT component division
 > alf.as> CALIB1: VISDFT: Model components of type Point
 > alf.as> CALIB1: I Polarization model processed
 > alf.as> CALIB1: Model computation is    90 percent complete
 > alf.as> CALIB1: Field    1 used     178 CCs
 > alf.as> CALIB1: Determining solutions using amp-scalar averaging
 > alf.as> CALIB1: Writing SN table    1
 > alf.as> CALIB1: Found          996 good solutions
 > alf.as> CALIB1: Appears to have ended successfully
 > 
 > There are no messages from SETFAC here, so I believe that no rescaling
 > is done. 
 > 
 > Could this missing SUMFLUX header be the reason of my confusion?
 > If so, is there an easy way to create this key without calculating the
 > sum myself?
 > 
 > 
 > And I have another simple question: Is it correct that the SN tables
 > hold something like "sqrt (expected flux/measured flux)"? I am
 > surprised about the sqrt.
 > 
 > 
 > Cheers,
 > Olaf
 > 

I have put back a change to the modeling routines so that they all
scale the model by
   1) The beam area in the case of Clean image models (with a stern
      warning) 
   2) The total flux / sum of the CC flux in the case of Clean
      component models
It will be available for a MNJ in about 4-5 hours.

Eric Greisen