[daip] bpass smoothing

[Eric Greisen]

Lynn D. Matthews writes:
 > In the help file for BPASS it states:
 > 
 > "Note that any use of SMOOTH in BPASS will require that
 > the same SMOOTH values be used when accessing any data to
 > be calibrated by the BP table produced by BPASS."
 > 
 > Could you elaborate on what are the consequences of ignoring this advice;
 > i.e., if one applies smoothing when running BPASS to increase S/N and then
 > applies this BP table to unsmoothed data, what undesirable effects (if
 > any) might this produce?
 > 
 > Tasks such as SPLIT or IMAGR don't seem to complain.
 > 

A smoothed BPASS solution will have left and right edges that rise
more slowly than the data themselves.  Furthermore, any real wiggles
in the BP function will be diminished by the BP smoothing.  If you
then apply the smoothed correction to unsmoothed you will get a
difference function between the real smoothed and unsmoothed BP
functions applied to your visibilities.  Thus, at the outermost
channels you will apply to little correction and so they result will
be too low.  At the next bunch of channels inside the outermost you
will apply to much correction and the result will be above the correct
values.

This is a subtlety that is explained at some length in the help files.
It was not realized until a few years ago and at that time I added
options to SMOOTH so that the unsmoothed BP table could be applied to
unsmoothed data and then a SMOOTH done.  However, if the BP table is
smoothed then the data must be too and before the BP table
application.

Eric Greisen