[mmaimcal] Re: subarrays

[Michael Rupen]

Just a quick response on the subarrays question...I think there's quite
a good case for having many more subarrays on ALMA than we're used to thinking
about.  I agree with Larry that we need to weigh those scientific/practical
benefits against the real engineering costs, but I don't think two or three
is anywhere near the number such a careful analysis would come up with.  We
summarized some of the reasons for subarraying in the infamous correlator
"white paper", _Astronomical Requirements for the MMA Corraltor_, Rupen,
Shepherd, & Wright 1998, section 3.7 .  Very briefly, one wants subarrays
for a wide variety of purposes:

* Simultaneous multi-freq. capability, useful for
   - solar flares
   - atmospheric studies
   - cases (as at the VLA currently for 7mm) where only some of the antennas
     are outfitted with certain types of receivers or bands. I suspect this
     will be much more common for an array as huge as ALMA.

* Calibration.  If we're at all thinking about simultaneous calibration using
  part of the array, to monitor opacity and/or gain on short timescales
  through tipping scans or other measurements at one or more wavelengths, that
  suggests subarrays to do so.

* Mapping large areas.  I know, I know, no-one but me & some other crazy 
  Americans will ever want to do this :)  but with M subarrays you do shorten
  the time to map a given area by a factor M, if sensitivity is not an issue.
  This is useful e.g. for finding interesting "weather" on the Sun and other 
  solar system objects (this can change VERY fast), or simply for doing big
  surveys...though as Larry will point out the latter does not absolutely
  _require_ subarraying.

* Fast-response observations.  If something nifty goes off you may want to
  quickly dash over to have a look, without stealing every antenna in the
  array.  Similarly ALMA will be so absurdly sensitive that monitoring obs.
  even of faint sources might easily be done with a subset of the antennas.

* Lower data rates.  Experiments which require lots of channels or very high
  time resolution will probably have to cut down on the number of baselines;
   this will undoubtedly be true initially.

* VLBI.  I don't imagine people will want to give up all or even half of ALMA
  on a regular basis for VLBI, but it will certainly be in enormous demand
  for high-freq. experiments.

* Inhomogeneous arrays.  If ALMA does go this route (cf. Wright's memos),
  there are probably reasons to split into homog. subarrays every now and then.

* Single-dish observations.  ALMA will probably do its own single dish work,
  but it's not clear whether one wants (or will be able) to use the entire
  array for this all at once.  So we'll want to split off bunches of single
  dishes, which you may or may not want to call subarrays.

* Some folks have suggested faking big dish diameters by phasing up lots
  of little (well, moderate-sized) antennas (cf. MMA Memo 165).

So I think there are some good reasons for a fair number of subarrays. 
Exactly how many depends on how you think the instrument will be used, and
how much it costs in $$ and complexity to build them in.  I'd be very surprised
if we wanted fewer than 6 or so myself.

-- Michael
   -