[daip] Hi
Eric Greisen
egreisen at nrao.edu
Tue May 7 15:33:09 EDT 2013
Nirupam Roy wrote:
> Hi Eric,
>
> How are you doing? I need your help regarding a question and a curiosity
> about two AIPS tasks (currently I am using 31DEC13 version).
>
> The question is about bandpass calibration. In BPASS, if I understand it
> correctly, the effect of spectral index and curvature terms are taken
> care of very consistently provided, if and only if, I am running
> bandpass with a single bandpass calibrator. It does not work if I have
> multiple bandpass calibrator sources. I am wondering, is there any way
> out to do this consistently for a data set with multiple bandpass
> calibrators? Suppose if I run BPASS for 3C 48 and 3C 286 separately, but
> both the time use output BP version 1, and then finally apply BP version
> 1 with DOBAND = 3 in SPLIT, will it consistently correct for the
> spectral index/curvature terms? Or is it that running multiple BPASS
> with same version of output BP table specified just overwrite the
> earlier bandpass solutions, and retain only the solution from the last run?
>
> The curiosity is about the task CALIB. What exactly DOAPPLY do in CALIB?
> There is no help or explain entry for DOAPPLY under CALIB. Help for
> DOAPPLY only mention the task RLDIF. While running CALIB for a single
> source file, I found it always produce a .CALIB file whether I keep
> DOAPPLY 0 or 1.
BPASS: The spectral index correction is done to the values that are fit
so that the solution as it is stored in the BP table comes out as if it
had been done on a source with no spectral index. Either you enter a
spectral index or the task knows about a few of the standard sources.
But BPASS can only do this for one source at a time. So run BPASS with
only one source at a time and with a fixed BP version out. The multiple
solutions appear separately in the BP table and may be averaged
(DOBAND=1) or time interpolated (DOBAND=3) or other methods as well (see
help doband). The tasks can do this since the BP solutions have been
made into a spectral index = 0 form before they are written in the table.
The DOAPPLY adverb is relatively new in CALIB and I guess the help file
is missing some things. The inputs section says most of what is needed
and the 1st paragraph of the help section says about all that one would
write later in the help section. Some users doing self-cal like to have
new corrected UV files as the start of the next step. Others like to
save the SN table but always go back to the original UV file for each
step. With modern large data sets, it is expensive to write the
self-cal'd output if one does not want it. Hence, DOAPPLY > 0 is needed
to get the self-cal'd output from CALIB. Of course, SPLIT could always
do it later if one forgot and wanted that output.
Cheers
Eric Greisen
More information about the Daip
mailing list