[daip] [!4536]: aips - SNSMO in VLMB mode sometimes breaks the relationship between phase(per IF) and delay

[Michael Bietenholz]

Michael Bietenholz updated #4536
--------------------------------

SNSMO in VLMB mode sometimes breaks the relationship between phase(per IF) and delay
------------------------------------------------------------------------------------

           Ticket ID: 4536
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/4536
           Full Name: Michael Bietenholz
               Email: mbieten at yorku.ca
             Creator: User
          Department: AIPS Data Processing
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template Group: Default
             Created: 19 February 2014 06:58 PM
             Updated: 19 February 2014 06:58 PM
                 Due: 24 February 2014 06:58 PM (5d 0h 0m)
      Resolution Due: 28 February 2014 06:58 PM (9d 0h 0m)



In AIPS the multi-band delay in SN tables is encoded in the IF phases, or more precisely in the differences in SN-table phase between one IF and the others.  So if for example, FRING is run so as to produce a combined solution over multiple IFs, there should be only one solution over all the for the phase, and the IF-to-IF differences in the SN-table phases should be functionally related to the solution for the delay, i.e., 
dphas(IF1 - IF2) = 2*pi*delay*(freq[IF1] - freq[IF2]), where phase is in radians.   SNSMO with mode = 'VLMB' is supposed to preserve this relationship.  In some cases, however, it seems not to.  

Attached are two pngs, pre-SNSMO.png and post-SNSMO.png.  My apologies - they're a bit hard to read as the overlaying was not exact.
The x-axis is time.  They show corresponding plots of an SN table before and after SNSMO.  The yellow points are the differences in SN-table phase between IF1 and IF 8, scale = -180 to 180 degs (OPTY='PHAS'; OPCO='IFDF'; BIF 1; EIF 8).  IFs 1 and 8 in this data-set are 56 MHz apart, so a difference in phase between the two IFs of 180degs should correspond to 8.9 nano-secs in delay.  The yellow points are the corresponding delays, with a scale of +/-9 nsecs.   If all is good, and the SN-table phases across the IFs differ only by an amount due to the multi-band delay, then the green and yellow points should be mirror-images of one another (there is a sign flip).  This is true before running SNSMO - as can be seen in pre-SNSMO.png.  However, after SNSMO, this is no longer true, and there is an excursion in the IF1-IF8 phase difference not mirrored in the smoothed delay around the middle of the plot.

I'm not sure how often this happens or what condition triggers it.

SNSMO was run in a reasonably standard way, to smooth the rates and delays, in VLMB mode:

AIPS 1: SNSMO     Task which smooths and filters Solution(SN) tables.
AIPS 1: Adverbs     Values                 Comments
AIPS 1: ----------------------------------------------------------------
AIPS 1: INNAME     'V253CB'                Input UV file name (name)
AIPS 1: INCLASS    'TASAV'                 Input UV file name (class)
AIPS 1: INSEQ         4                    Input UV file name (seq. #)
AIPS 1: INDISK        8                    Input UV file disk unit #
AIPS 1: SOURCES    *all ' '                Source list ' '=>all.
AIPS 1: DOBTWEEN     -1                    > 0 -> smooth all sources
AIPS 1:                                    together; else separate them
AIPS 1: SELBAND      -1                    Bandwidth to select (kHz)
AIPS 1: SELFREQ      -1                    Frequency to select (MHz)
AIPS 1: FREQID       -1                    Freq. ID to select, 0=>all
AIPS 1: BIF           0                    Lowest IF number 0=>all
AIPS 1: EIF           0                    Highest IF number 0=>all
AIPS 1: TIMERANG   *all 0                  Time range to use.
AIPS 1: ANTENNAS   *all 0                  Antennas to correct.
AIPS 1: SUBARRAY      0                    Subarray; 0 => all.
AIPS 1: SAMPTYPE   'GAUS'                  Smoothing function
AIPS 1: BPARM         0           0        Smoothing parameters
AIPS 1:               1.5         1.5         1.5         0
AIPS 1:               0           0.5         0.5         0.5
AIPS 1: CUTOFF        0                    Cutoff for functional forms
AIPS 1: DOBLANK       0                    Blanked value interpolation
AIPS 1: SMOTYPE    'VLMB'                  Data to smooth
AIPS 1: CPARM      *all 0                  Range of allowed delays and
AIPS 1:                                    rates.
AIPS 1: NORMALIZ      0                    > 0 => apply a global
AIPS 1:                                    normaliztion - see HELP
AIPS 1: INVERS        7                    Input SN table; 0=>highest
AIPS 1: OUTVERS       0                    Output SN table; 0=>new
AIPS 1: REFANT        7                    Reference antenna 0=>pick.
AIPS 1: BADDISK    *all 0                  Disks to avoid for scratch

I've also attached a TASAV which has the relevant SN and FQ tables attached, although I've stripped the more voluminous other tables
so the file isn't too big.


------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pre-SNSMO.png
Type: image/png
Size: 23242 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140219/a12838fa/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: post-SNSMO.png
Type: image/png
Size: 22767 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140219/a12838fa/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TEST.TASAV.gz
Type: application/x-gzip
Size: 1332664 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/daip/attachments/20140219/a12838fa/attachment.gz>