Fit a binary position to visibility data
First, we compute the model (Model|Calc), then we plot the visibilities
(Plot|INTERFEROMETRY). You could also use Calibrate|Scans|PLOT for plotting,
but the other choice is more tailored to AMOEBA. Study the selections made
in the two widget, the plot widget and data selection widget. Note that
slice 'ch' is selected, which produces a more readable set of plots with the
data plotted versus channel, one scan at a time. If Options|Model is selected,
a line will indicate the model values (by the way, it is not interpolated).
Obviously, in this example, the fit is not great, but we are about to improve
Bear with me, but several more widgets have appeared on the screen! To fit
a model, start with Model|FIT, which lists (almost)
all parameters for all components
of the model. In our example, we actually select none of those listed, because
the two parameters we are interested in, position angle and separation, are
listed in a separate widget under Fit|INTERFEROMETRY. This is because these
parameters are not really part of the physical model and are therefore dealt
with using an additional fit widget. To start the fit, click on the Fit button
on the Model|FIT widget, NOT in the Fit|INTERFEROMETRY widget. The latter would
start a customized fit for a list of nights, writing the results into a
preformatted LaTex table file.
Note by how much we reduced the ChiSq. The fact that the reduced value is
rather high has more to do with the fact that we have not re-calibrated the
error bars from the scatter of the calibrator visibilities. This can be done
using Calibrate|Scans|SetErrors. Also note that while we have plotted only
VisSqC for one output beam, there are more, and also there is the triple
amplitudes and phases!
In the figure below, the triple phase and amplitude is plotted with the model.
I went back to Calibrate|Scans|CALIBRATE to set the calibrator stars which
I had used for the calibration of FKV0123. This information is needed by
the set_viserrors procedure for the calibration of the errors. The messages
printed out at the bottom of the PV-WAVE window show that, of course, no
valid data was found for channels 11 and higher than 20, which were flagged
previously. Also note, that one has to calculate the estimated visibilities
for that, some thing which AMOEBA does not usually do. In order to do this,
click on Calibrate|Scans|Astrometry|Calc. Otherwise, set_viserrors will
complain that no valid data at all could be found for the error calibration.
Note that the error bars look quite a bit more realistic. The reduced ChiSq
is now 3.3, still not perfect, but OK for now.
Back to tour page
Back to main help page