[daip] aips calib

[Eric Greisen]

Cao Hongmin wrote:
> Hi Eric
> 
> So, if the source is very strong, and there are enough antennas for 
> amplitude self-calibration, it will not make much difference (to set 
> aparm (1) =3 or 4), is it?
> 
> if the input model/image of the calibrator has (big) amplitude 
> calibration errors, whether it will have big influences on the total 
> flux density detection of the weak (~1 mJy) target sources, after 
> applying the solution got from "calib" to the target sources ? and how 
> about the strong source (eg. also apply the same solution to the 
> calibrator)?

This is not quite what I said.  If you have a bad model for the 
calibration source and set APARM(1)=3 you run the risk of freezing in 
that bad model in your self-cal solutions.  You would be better off with 
APARM(1)=30 if that could work (it won't because you probably do not 
have that many antennas)
because that would mean that the antenna-based calibration dominates the 
solution rather than the (not good) model.

The amplitude errors are less important than the phase errors in finding 
weak sources.  If the phases are poor, then the visibilities will cancel 
each other.  If they are good the visibilities will add up well and the 
amplitude errors will to some extent offset each other.

Eric Greisen


> 
>  > -----ԭʼÓʼþ-----
>  > ·¢¼þÈË: "Eric Greisen" <egreisen at nrao.edu>
>  > ·¢ËÍʱ¼ä: 2013Äê2ÔÂ26ÈÕ ÐÇÆÚ¶þ
>  > ÊÕ¼þÈË: "Cao Hongmin" <hmcao at shao.ac.cn>, daip at nrao.edu
>  > ³­ËÍ: 
>  > Ö÷Ìâ: Re: [daip] aips  calib
>  > 
>  > Cao Hongmin wrote:
>  > > Hi Eric
>  > > 
>  > > when we do amplitude+phase self-calibration using calib with an input 
>  > > image,
>  > > 
>  > > will it make much difference if we set aparm(1) =3 and aparm(1)=4 ?
>  > > 
>  > > thanks for the helps
>  > > 
>  > > best regards
>  > > 
>  > >      hongmin
>  > 
>  > When there are only 3 antennas, the phase will exactly match
>  > your model.  For antenna-based calibration such as CALIB it
>  > is probably best to insist on 4 antennas at least.  In normal
>  > cases, there should not be much difference.
>  > 
>  > Eric Greisen
> 
> 
>