[daip] Problem with JMFIT (zero level is not correct for CTYPE=2)

[Eric Greisen]

Michael Bietenholz writes:
 > I think there is a problem with JMFIT (31 DEC07 Sep 3; sorry if this has
 > already been fixed).

     The bug has been there a very long time and no one has noticed.
If one changed DOCRT from 1 to -1, the output baseline parameters
would change.  The last of the fit parameters would be left 1 sigma
off for baselines only unless one recomputed one last time to print
the model image.  An extra recomputation was added to force all
parameters to their correct values.

Eric Greisen


 > 
 > JMFIT returns the wrong value for the fit zero level (CTYPE 2) - its too high
 > 
 > It gets a different value than IMFIT and also (for a CTYPE 2, NGAUS 1
 > fit) MSTAT does (the least-squares zero-level should be the same as
 > the mean).  I think JMFIT's zero level is actually the (zero level +
 > rms) which is a bit odd.
 > 
 > The zero-levels returned by IMFIT seems more reasonable and agrees with
 > the mean levels found by IMSTAT.
 > 
 >               michael b
 > 
 > _______________________________________________
 > Daip mailing list
 > Daip at listmgr.cv.nrao.edu
 > http://listmgr.cv.nrao.edu/mailman/listinfo/daip