[daip] index problem in a SPLAT-ed uvdata

[Eric Greisen]

Hu Bo wrote:
> Dear Eric,
> 
> I came across a confusing problem when trying to completely change the 
> source information of a SPLAT-ed uvdata.
> The original dataset has 15 sources with 7 of them are of interests.
> So I used SPLAT to produce a new dataset containning only those 7 
> sources.
> Here is the input of SPLAT:

> AIPS 1: SOURCES    'F036.1638'             Source list
> AIPS 1:            'G036.11+0.55'  'J1855.036'
> AIPS 1:            'J1903.036'  'J1848.036'
> AIPS 1:            'F036.1800'  'F036.0019'
> AIPS 1:            *rest ' '
>
> Then I started calibration with TECOR. But it returned me an error 
> message of:
> 
> MCN4  > TECOR1: Task TECOR  (release of 31DEC14) begins
> MCN4  > TECOR1: Reading from CL table version number    1
> MCN4  > TECOR1: Searching for earliest and latest times in CL table
> MCN4  > TECOR1: Done
> MCN4  > TECOR1: Dispersive delay corrections will not be calculated
> MCN4  > TECOR1: ZTXOP2: using translated file name =
> MCN4  > TECOR1: ZTXOP2: /data/hubo/AIPSRUN/BR149N/epoch5/jplg3650.13i
> MCN4  > TECOR1: Creating CL table version    2
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     2
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   2 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     7
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   7 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     8
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   8 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     7
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   7 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     9
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   9 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     7
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   7 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.    10
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE  10 INFORMATION
> ...
> ...
> ...
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   7 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.    11
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE  11 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.     7
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE   7 INFORMATION
> MCN4  > TECOR1: GETSOU: COULD NOT FIND DATA FOR SOURCE NO.    15
> MCN4  > TECOR1: TECCOR: ERROR  11 READING SOURCE  15 INFORMATION
> MCN4  > TECOR1: Copied  83577 records to the output table
> MCN4  > TECOR1:  83577 were modified
> MCN4  > TECOR1: Appears to have ended successfully
> MCN4  > TECOR1: mcn4 31DEC14 TST: Cpu=      0.6  Real=      1  IO=      
> 113
> 
> These not-found sources are the 8 sources that I discarded in SPLAT.
> I guess this may cause further problems in calibration so I used TABED 
> to correct the SOURCE-ID column in SU table,
> from 1,3,4,5,6,12,14(the original source id in the big 15 sources 
> dataset) to 1,2,3,4,5,6,7. And deleted NX and CL table,
> using INDXR to produce new ones. But the NX and CL are still with the 
> original source ids, like this(the NX table as an example):

It is dangerous to use data editing tasks like puthead, tabput, tabed, 
etc unless you know what you are doing.  The visibility data contain the 
source number and it is dangerous to change the SU table to fail to 
match your data.  In fact TECOR worked fine - the issue is that the 
other source numbers remained in the CL table after SPLAT but not the SU 
table.  TECOR needed SU data to deal with the CL records of the removed 
sources and SPLAT eliminated the sources that no longer had data.

We used to allow CL records from one source to affect another, which is 
why SPLAT kept everything.  That is no longer allowed so I will fix 
SPLAT (and perhaps others) to make a proper matching CL table.  If your 
CL table contained nothing but the TECOR and initial values, then you 
can deleted all CL and NX and run INDXR, then re-run TEXOR - but not 
after you mess up the CL table.  If you really need to change source 
numbers, there is a task DSORC for this purpose.

Eric Greisen