[daip] [!4940]: aips - JMFIT and -ive zero level

[Michael Bietenholz]

Michael Bietenholz updated #4940
--------------------------------

JMFIT and -ive zero level
-------------------------

           Ticket ID: 4940
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/4940
           Full Name: Michael Bietenholz
               Email: mbieten at yorku.ca
             Creator: User
          Department: AIPS Data Processing
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template Group: Default
             Created: 20 May 2014 03:01 PM
             Updated: 20 May 2014 03:01 PM
                 Due: 22 May 2014 03:01 PM (2d 0h 0m)
      Resolution Due: 29 May 2014 03:01 PM (9d 0h 0m)



JMFIT (31Dec14) seems to have problems when fitting zero levels which are < 0. 

Fitting a region with a +ive point source but a below-zero average level with
CTYPE = 0,2,0; DOMAX = 1,1,0; DOWID=-1,-1,0;  OFFSET 0
GPOS 6,6,0; [position of point source]; GWID 0; GMAX 0;
(attached file test-ive.fits, I just used BLC 0; TRC 0; I saw the same behaviour
with an unrelated bit of a different image that also had a -ive zero level)

Produces this response
  JMFIT1: Component   1-Gaussian
  JMFIT1:   Peak intensity    = 1.2887E-03 +/-  3.62E-04 JY/BEAM
.  ....
  JMFIT1: Component   2-Baseline
  JMFIT1:   Slope=  0.000E+00( 1.349E-03) Ampl/pixl
  JMFIT1:      in angle   0.00(  0.00) Deg

Which is strange because I asked only for CTYPE=2 which doesn't have a slope!
Giving it sensible starting values in GMAX doesn't help

If I reverse the sign of the pixel brightnesses in this region, I get the fit I expect,
with now a -ive point source sitting atop a +ive zero level (attached file test+ive.fits)

  JMFIT1: Component   1-Gaussian
  JMFIT1:   Peak intensity    =-1.2887E-03 +/-  3.62E-04 JY/BEAM
  .....
  JMFIT1: Component   2-Baseline
  JMFIT1:   Zero level=  6.3375E-03( 6.4805E-05)  Ampl. units

Also, setting CTYPE 2,0; NGAU 1 never seem to converge regardless of whether the level is +ive or -ive.  I would have thought this should converge rapidly to the mean level in the fitting window.  Since IMEAN/IMSTAT is an easier way of getting the average brightness in a box I don't imagine people do this very often - I only did it to try and troubleshoot the issue with the -ive zero-level.


------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-ive.fits
Type: image/fits
Size: 20160 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140520/80f639d9/attachment.fits>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test+ive.fits
Type: image/fits
Size: 14400 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140520/80f639d9/attachment-0001.fits>