[daip] 31DEC14 AIPS - Change in behaviour of CLCAL for de-selected SOURCES
Eric Greisen
egreisen at nrao.edu
Thu Jan 16 18:51:09 EST 2014
Michael Bietenholz wrote:
> Happy New Year Eric
>
> I noticed that CLCAL in 31DEC14 writes out CL table records even for
> sources deselected by the SOURCES adverb. (In 31DEC13 it would not
> write out such records).
>
> Was this an intended change? It seems to just pass an unmodified version
> of the input CL table through for those sources. THis behaviour is
> probably
> safe and perhaps even more useful than that of the old version, I'm just
> checking that this was an intended change.
>
> (For complicated CLCAL's I would check the number of CL table records
> as a quick check that I hadn't lost a bunch of data due to missing
> calibration;
> which check no longer works now; one has to be careful in such cases
> that one does not wind up with some unintended mixture of updated and
> un-updated CL records)
>
> cheers, michael
This behavior is controlled by the adverb OPCODE with CALP (for pass)
the new default. Too many CL tables were gradually losing entries for
scans and sources in a way that was not intended. A bug in the
application code (actually many bugs) caused it to apply CL records for
source A to source B and/or to interpolate over vast distances in time.
Since CL tables have source-dependent values (elevation corrections in
phase, gain, position corrections, antenna position corrections, etc.)
one cannot interpolate between sources. I fixed that starting with
changing the OPCODE default Feb 20, 2013. From then into June I
gradually solved all the ramifications of requiring a CL table record to
apply to the current source and even had another fix in October. All
this was in 31DEC13 - there are no changes in 31DEC14 from 31DEC13.
Eric
More information about the Daip
mailing list