[daip] JMFIT - OFFSET

[Eric Greisen]

Ian Pattison writes:

 > I have a query about the use of OFFSET in JMFIT. I wish to fit a gaussian
 > component to a circular source ('looks' point-like, but the integral flux
 > is ~3 times the peak/beam) surrounded by some extended emission. I dont
 > want this extended emission to affect the fit, so I have been setting
 > OFFSET to a suitably (by eye!) high level. I would just like to know if
 > this method is acceptable - finding info on this subject has been
 > surprisingly tricky!

What information do you expect?  OFFSET simply says exclude (actually
blank) those image pixels less than a certain amount.  All other
pixels are left alone.  The situation you describe is probably very
much more like a core-halo 2-component source model which can be fit
as two Gaussians or as a Gaussian core and a fit "dc" level (with
slopes, curvatures, etc.).  Component Type 5 allows you to specify the
DC bias, etc and to fit or hold fixed all 6 parameters of that
component type.  There is rather less than enough information about
component types 1-5 than there should be, probably because people do
not use them much.  The initial guess for types 2-4 is forced to zero
and the parameters are fit as specified (type 2 zero level, type 3
zero and slope (mag, angle), 4 zero, slope, curvature).  OFFSET is
jsut used to remove weak pixels form the fit not to subtract a DC
bias. You must use Type 5 for that and set DOPOS, DOMAX, DOWIDTH
carefully.

Eric Greisen