[daip] AIPS - DBCON + IMAGR
Patrick P. Murphy
pmurphy at NRAO.EDU
Tue Nov 7 11:15:43 EST 2000
"RNM" == R Niruj Mohan <niruj at rri.res.in> writes:
RNM> We have 15OCT99 version of AIPS running on our machine. I know we
RNM> should upgrade to 31DEC99 but there is a meeting coming up next week
RNM> and there is tremendous pressure to get final images etc out !
I'm appending some of the changes made to DBCON since the 15OCT99 release,
so you can realize there have been many changes made.
RNM> The problem with the 15OCT99 we have is this :
RNM> I have C and D array data on an object and I want to concatenate it
RNM> and make a single image. So I use DBCON to create a C+D uv file.
Can you show us (a) the output of IMHEAD on the separate C and D uv data
catalog entries, and (b) the inputs you used to DBCON, especially the
values of REWEIGHT, DOPOS, FQTOL, and DOARRAY. We need a complete picture
like this before it's possible to comment on the specifics of what went
wrong.
Also, I'll assume the C and D array data are *not* multiple source files.
RNM> When I image it using IMAGR, the MSGSRV indicates the correct number
RNM> of total visibilities it takes, but the image is only of the first uv
RNM> file used in DBCON
What are your IMAGR inputs? In particular, what FREQID did you select?
There are a lot of inputs to IMAGR and a lot of opportunity to have one
not quite right...
RNM> ... I copied the latest (30DEC99) DBCON.FOR and compiled it and used
RNM> it
This is dangerous. The subroutines and libraries in the 31DEC99 version
of AIPS have changed sufficiently that I would expect this sort of thing
to not necessarily produce trustworthy results.
RNM> Am I right ? If so, is there a quick temporary way out, like a patch
We are not releasing patches for 15OCT99 any more, unless they are
extremely serious. We recommend using 31DEC99
RNM> (I noticed that all the patches for the older AIPS have been removed
RNM> from the webpage)
This was an oversight (and perhaps us being a little too zealous about
encouraging use of 31DEC99). I've put a reference to the patch page
back. The very first one is a DBCON patch, FWIW; maybe this will help.
RNM> Or can you tell me which of the new IMAGR-related files of 30DEC99
RNM> version should I copy over and compile and therefore cheat a bit ?
Trust me: it's easier to get the 31DEC99.tar.gz file, the install or
update perl script, and run that way. It would take far longer to figure
out the bits you need, and longer again to squeeze them in. That approach
is just not worth the effort.
- Pat
------------------------- fixes to DBCON in 31DEC99 --------------------
10434. November 22, 1999 DBCON Eric
DBCON forgot that the number of output correlators can exceed
the number of input in 1 of the 2 data sets. It also did not
zero the right array to help cover that. Fixed.
Moved nowhere: should go to patch.
10519. February 29, 2000 DBCON Eric
1. Changed code and help to remove the claim that it did
frequency independent phase shifts unless DOPOS was 2.0. It
did the right phase shift but would refuse if the two files
were not in the same sort order and DOPOS>1.
2. Changed it to examine the FQ tables and renumber the FQs if
needed in output UV data, and output FQ, CL, and FG tables.
Moved nowhere.
10522. March 7, 2000 DBCON Eric
DBCON did not scale the u,v,w's for any frequency difference
between the two data sets! This is okay, so long as they are
not multi-channel.
The error message about frequencies not matching was wrong.
It came out when it should not rather than when it should.
Moved nowhere.
10529. March 16, 2000 DBCON Eric
For 2 single channel data sets the u,v,w are already in
wavelengths at whatever freq they were actually observed. One
does not want to change the u,v,w therefore - making the fix
of March 7 partly wrong. Backed out the extra mutiplication.
Moved nowhere.
10532. March 21, 2000 DBCON Eric
The recent change to handle FQ numbers properly put a road
block into the attempt to combine data with different
bandwidths as if they were the same. DBCON now - rightly -
makes them into 2 FQs. Added FQTOL to adverbs to allow the
task to ignore FQ when told to (-1) or to use FQTOL (kHz) to
decide whether FQs are the same.
Moved nowhere.
10553. April 19, 2000 CookBook Eric
Updated WHATSNEW.HLP and CookBook chapters 4, 6, 7, 8, 9, 10,
and 13 plus the consequent changes to the Index and table of
contents. TVSLICE et al, DBCON, SPFLG, UJOIN, FILLM and
DOCAL=2, ICHANSEL were the causes of the changes.
Moved nowhere.
10560. April 27, 2000 DBCON Eric
DBCON attempted to copy CL and FG tables with translation of
FQ and source numbers and times. It failed if only one of the
two files had a file of this type. It also did not fix the
subarray and did not use the correct time offset. Added a
TABCOP to fix the missing file problem and fixed the calling
subroutine to send in correct time offset and a new subarray
offset and fixed DBCAPP to do the subarray offset.
Moved nowhere.
10561. April 27, 2000 DBCON Eric
DBCON also had an error when combining LL data with an rr/ll
data set and writing the result out in compressed form. The
output (compressed) LREC was used to zero the accumulation
buffer not the LREC prior to compression. This does not
matter unless some input correlators are not present so that
values are not filled in.
Moved nowhere.
10583. June 2, 2000 DBCON Eric
Changed the test on coordinate rotation so that an angle of
1e-6 and 0 would be regarded as close enough to each other
(!) Cleaned up the code some too.
Moved nowhere.
10648. August 2-3, 2000 ZOPEN Eric
When ZOPEN returns an error code 1 (already used in FTAB) this
means that some file is open with that LUN. To use that LUN
for some other purpose at that point allows that other purpose
to overwrite the original file - the users data in FITAB for
example. It is never right to mask this error. Changed
maksing of this error in:
AMERGE EXTINI IMGERR RELPOP TABINI UVPREP UVPROT
ALGSUB ONV1 CONV3 GRDMEM UVTBGD PASS1
AHIST BSROT CLIP CORER CORFQ CPYRT DIFRL DIFUV
FUDGE HISEQ REMAG SDLSF SDMOD TAFFY TBAVG UVBAS
UVCMP UVLSD UVLSF UVMTH UVPOL WTSUM XBASL XGAUS
XMOM XPLOT XSMTH XSUM ACCOR AVER AVSPC BASRM
BLAVG BLOAT BSFIX BSMAP BSMOD CANDY DAYFX DBCON
DECOR DESCM DTCHK FETCH FILLM FILLR FLGIT FXTIM
FXVLA GLENS IMFLT IMLIN IMMOD INDXH M3TAR MANDL
MK3IN MODVF MULTI MWFLT NINER NNLSQ PATGN PCCOR
PHNEG PHSRF RESEQ SBCOR SDTUV SELSD TI2HA UJOIN
USUBA UVAVG UVCOP UVCRS UVDGP UVFIL UVFIX UVGLU
UVLIN UVMOD UVNOU UVSIM VBCAL VBGLU VBMRG VLBIN
WTMOD BLCAL BPASS CALIB CPASS FRING GRIDR LPCAL
PCAL SDIMG APCLN APGS APVC SDCLN STEER UTESS
UVMAP BLANK BLSUM IMVIM IBLED
Moved nowhere - INSTEP4 to be done.
10652. August 10, 2000 DBCON Eric
One of the ZOPEN calls corrected a few days ago had an error
test of IRET (set to 1 at that point) while ZOPEN returned
IERR. The correction to allow only 0 caused DBCON to have a
hard-coded failure. Sigh...
Moved nowhere.
More information about the Daip
mailing list