[daip] uv subtraction and calibration

[Eric Greisen]

Lorant Sjouwerman wrote:
> Hi Eric,
> 
> I have a line set with a variable continuum source that i want to 
> subtract. As far as I understand uvlin/uvlsf/uvmln/uvbas only subtract a 
> single (time averaged) model from all visibilities. Is there a way, eg 
> with a solint, to calculate the continuum for some time range and so 
> subtract the time var continuum from this time range (and then loop over 
> all time ranges with solint) more accurately to get no residu from the 
> averaging over all time? (I also have constant continuum and several 
> masers in the data for which the residu interferes with the time var of 
> the masers). The flux is bright enough to do this over individual 
> correlator integrations, so I might end up to want to do this for every 
> time stamp (over 1024 channels duAL pol)

You do not understand these tasks correctly.  They fit for a continuum 
and subtrat it on a per visibility basis.  This should be ideal for a 
time variable continuum.  Note that they fit a DC and slope normally 
(although UVLSF can do more) and that works well only if the dominant 
continuum is at the center or can be shifted there temporarily.  UVLSF 
will also write out a continuum UV data set for later, more complex in 
your case, analysis.
> 
> Another question I have is on Calib. I want to use several line sources 
> to create a SN table. No source is in the data twice and no phase 
> calibrators are interspersed (I rely on self cal because there are no 
> nearby high-freq calibrators to be used). I want to use self-cal per 
> source on a source dependent individual channel and write all solutions 
> in the same SN to interpolate over a few sources that are too weak for 
> self cal (they're all near in the sky). So this multi source calib 
> should not produce new SN tables and not produce a new calib file. My 
> understanding is that this can be done on multi sources - only for 
> single source I'd get a new SN table and Calib file, not for a multi 
> source file (I have no data set yet to test this on
> and want to avoid creating 50 new SN/Calib files per hour of observing).

CALIB has an SNVER adverb and can write all answers into a single table.
Note that this requires that all sources reside in a single 
(multi-source) data set.  There is no problem with models in such a data 
set so long as you select a single source carefully.

Eric Greisen