[daip] aips imagr

[Eric Greisen]

Cao Hongmin wrote:
> Hi Eric
> 
> if we accidentally put clean boxes on some spurious structures which are 
> caused by such as sidelobes or errors of phase/amplitude, does it will 
> prevent the image thermal noise (rms) from decreasing ?
> (it seems that it does help to invent some structures.)

If one cleans erroneously, then there will be errors that will cause the
process to converge very much more slowly.  In principle, sidelobes 
taken as sources should eventually be canceled out, but the process of 
canceling takes many iterations.  It also can fail with unfortunate 
problems in the dirty beam (as with vlbi with a small number of antennas).

It is best to Clean only obvious sources at each step - the autoboxing 
algorithm in IMAGR offers that option.

Bad calibration can also give apparent sources that are not real.  If at 
all possible avoid them since they will not have the beam pattern of the 
dirty beam and so cannot be deconvolved without spreading the error 
around the rest of the field.

Eric Greisen