[evlatests] Time stamps: BDF vs. SDM vs. CASA & AIPS

[David Harland]

On Fri, June 24, 2011 12:51 pm, Michael Rupen wrote:
> ...
>
> * SDM Tables take their times from observation documents which give
>    the requested scan start times.  The OPT produces scans in LST,
>    which unless you work hard are aligned on LST seconds.

An amplification here for those less familiar w/ the OPT, and a question
regarding how the OST (dynamic scheduler) is used in practice:

The OPT allows users to choose from one of 6 types of timings for a scan. 
Two axes:
  1. Time definition: UT vs LST
  2. Time type: scan-stop, on-source, total-duration

In order for all OPT scans to be aligned on integral LST seconds we'd need
to have all scans expressed as LST total-duration (which is, at least
currently, the norm) AND we'd need the model-to-script (m2s) algorithm to
be fed a start time that was an LST w/ no fractional seconds.

Question (for Keith): are the start times that the OST feeds to m2s
necessarily LSTs w/ no fractional seconds?

Even if the above is answered in the affirmative, as we build the script
we're continually converting away from LST to UT in order to put scan
start MJDs in the script.  I'd be quite surprised if we didn't accumulate
rounding error of at least fractions of a millisecond, and possibly more. 
UT is the native time type for most, if not all, of our group's code.  LST
is a convenience provided for the users.


>    ..the requested scan interval will not in general contain an
>    integral number of 1sec UT integrations aligned on UT seconds...

I've had a chat w/ more than one person about forcing scans to be integral
multiples of integration times.  Nothing has been settled in this area
yet.