[daip] New Staff Reply - [!HIU-291162]: AIPS task UVRFI

[Eric Greisen]

New Staff Reply: AIPS task UVRFI

I have looked at the code a little and asked Leonia
about your questions.  

1. leonia thinks that the leftover time is lost -
   but looking at the code, it may attempt a solution with
   the short interval.  You could look at your output to
   settle that question.

2. The upper limit should be set by the upper limit of
   data averaging that is acceptable for your imaging problem.
   That could be quite short.

3. If Yinc is very small, your fringe rate resolution will be poor
   and the task will have trouble distinguishing signals at the
   source fringe rate (stpped presumably) from those at other rates
   (not stopped).

1. The task was tested with data in BT order and had an error
   that precluded it working in TB order.  I have today fixed
   that error - so if your computer has lots of memory - TB
   should work (after tomorrow's MNJ).  Note that I have not
   tested this but the code is based on the well-tested UVAVG
   which works for both sort orders.  The NX table messages are
   benign except that source-dependent selection and flagging
   are not likely to work.

2. With TB now allowed the calibration adverbs make some sense.
   Leave them set to "don't" with BT order.

3. If you want to apply flags/calibration before UVRFI, run
   SPLAT.  Afterwards, calibrate as normal.

4. The components are equivalent to the Clean components in imaging
   but they have both an amplitude and a phase and so are complex.
   The task is experimental so there is little lore about what might
   work.  It would not take more than say 10.

5. There needs to be an exclusion region around 0 fringe rate to
   protect the desired signal.  This is a region in which no components
   will be found - not a limitation on "subtraction" but on the finding
   of components.

6. Again, this is experimental.  I suspect 3 sigma is low enough for
   this operation (unlike normal image Clean).  Leonia used APARM(8)=0.

7. APARM(9) is the width of the dirty beam actually subtracted from
   the data - any beam outside that is ignored.  Leonia recommends 10
   but I would think it would depend on the shape of that dirty beam.

8. ZINC is the pre-averaging time, which would be the on-line averaging
   if there has been no additional averaging.

If you get UVRFI to work for your data, please let us know.  It would be
good to acquire experience of this sort.

Eric Greisen




Ticket Details
===================
Ticket ID: HIU-291162
Department: AIPS Data Processing
Priority: Default
Status: Open
Link:  https://help.nrao.edu/staff/index.php?_m=tickets&_a=viewticket&ticketid=2393