[daip] Re: DBCON question

[Eric Greisen]

Marc Verheijen writes:
 > I have a question about how to combine two data sets with DBCON. This
 > is not a bug-report so that's why I didn't send it to daip. It's more
 > a conceptual question, I guess.

     I have been away on vacation but am now back.

 > 
 > I have two deep continuum data sets created from an L-band line data
 > cube, observed around 1195 MHz. One set was taken in 2000 (8 epochs),
 > the other in 2001 (13 epochs) and they both have the same pointing
 > center. However, they were taken in different VLA configurations and
 > with slightly different frequencies. Below are summaries from imhead
 > after extensive selfcal iterations on both sets separately. Now I want
 > to combine the two data sets and continue the selfcal process based on
 > the combined and deeper model image. My question is: how should I
 > DBCON these two datasets and take advantage of the slightly different
 > frequencies and configurations to get a better filled UV plane?
 > 
 > As you can see, the 2000 set has 8 AN tables, one for each epoch. One
 > question is; do I first need to consolidate these 8 AN tables into a
 > single AN table before I run DBCON? If so, how do I do that?

     This should not matter.

 > 
 > Another question is: should I first turn them into multi-source files? 

     You want to use tham as single-source - it is better to stay in
that mode.  If the AN tables (all 8 plus the 1 from the 2nd set)
differ then multi-source is not an allowed mode (reading the inputs
info re DOARRAY).

 > 
 > Within DBCON, I have played with the DOARRAY and FQTOL parameters but
 > I have a hard time evaluating the structure of the resulting dataset.
 > I have also tried to DBCON multi-source versions of the datasets but
 > I'm not sure that was the right thing to do.

     You want the final data to have only 1 FQ number so it is best to
set FQTOL=-1 and live with the warning message below.

 > One of the things that DBCON tells me is:
 >  espres> DBCON1: FQIDS HAVE NOT BEEN RESET AS THEY SHOULD DUE TO FQTOL

 > After I DBCON them as multi-source versions and run INDXR, I get the
 > following message which seems to indicate that AN1 got overwritten:
 >  espres> INDXR1: Task INDXR  (release of 31DEC01) begins
 >  espres> INDXR1: ANSAME: AN TABLE   2 DOES NOT MATCH AN TABLE 1 AT ANTENNA  0
 >  espres> INDXR1: A CL table already exists: a new one will not be created
 >  espres> INDXR1: Appears to have ended successfully
 >  espres> INDXR1: espresso     31DEC01 TST: Cpu=       7.4  Real= 8

     Use single-source.

 > When I try to run IMAGR on the DBCON-ed single-source file I get the
 > following message:
 >  espres> IMAGR1: Task IMAGR  (release of 31DEC01) begins
 >  espres> IMAGR1: CHNDAT: FREQID =  -1 INVALID ON READ
 >  espres> IMAGR1: GAININ: CAN NOT READ CH/FQ TABLE
 >  espres> IMAGR1: OUVOPN: ERROR OPENING Input UVdata
 >  espres> IMAGR1: Purports to die of UNNATURAL causes
 >  espres> IMAGR1: espresso     31DEC01 TST: Cpu=       0.0  Real=       0

    I think I need to see the contents of the FQ tables of the inputs
and output.  This message implies that you have more than FQID in the
data set - a no no to IMAGR which can handle only one at a time.

Eric