[daip] how to printout the closure phase in AIPS?

[Eric Greisen]

Z.-Q. Shen writes:
 > Then, my question is how you estimate the error in case of closure phase.
 > It is simply the summation of the error of visibility phase on each of three
 > baselines, weighted by their previous weights. This is done after the
 > average of phase on each baseline over the averaging time. Is this exactly
 > what is done in AIPS before plotting closure phase with CLPLT?
 > 
       The weights in AIPS are supposed to be 1/sigma**2.  Thus the
uncertainty in the closure phase is the sqrt if the 3 uncertainties
added in quadrature (sqrt (a*a + b*b + c*c) where a,b,c are the 3
uncertainties).  The uncertainty in phase is the ucertainty in
amplitude divided by the amplitude.  All this CLPLT gets right - but
it takes the uncertainty to be the weight when it should take 1 / sqrt
(weight).  Thus the error bars plotted by CLPLT are wrong - I will fix
this.

       Note that weights in AIPS are sometimes rather unrelated to the
real uncertainty and are usually at best proportional to the
uncertainty rather than being exactly the correct value.  This makes
such plotting difficult.

 > Even though, a more general question is how you calculate the error or 
 > scattering
 > in phase. We can do coherent and incoherent average. But I am not sure about
 > this in AIPS. In particular, in the CLPLT, it only asks for the input for the
 > averaging time. Incoherent average is simple. But in case of coherent average,
 > I do want to learn how this is done in AIPS for not only phase also amplitude.

Incoherent averaging produces only amplitude - all averaging to
produce phase must be done coherently by averaging the real and
imaginary parts.  The computation of uncertainties is simple.

Eric Greisen