problems with Cattery fitting routines

[twillis449]

I'm having some problems getting fitting and calibration up and running
for base-line dependent averaging simullations. The following stuff is
available in my directory 'test_fitting' on Jake.

1. I've created a 5 min long test observation at 0.1 sec sampling rate
   with 99 antennas (just taking every 2nd antenna from a comination of
   meerkat + SKA1). The MS is already created - ska1_orig.MS. Run the
   turbo-sim.py script to generate a noise-free sim of a 1 Jy point source
   at the fieid centre whose phase is perturbed by the ionosphre. The
   parameters are already set in the .tdl.conf so the sim will run correctly.

2. OK - I try running calico-generic.py with the parameters given in
   the .tdl.conf file. Basically do a self-cal for each 0.1 sec time
   interval. (Note that I had to modify the standard supplied
   calico-generic script as the standard things like oms-gridded sky had
   been commented out.) Every thing works find for the first ~2500
   intefrations - phase is perfectly straightened out, but then the
   meqserver just goes 'poof' with no proper explanation except a status
   of -1, so the last 500 integrations are not calibrated! I get the same
   exact result if I use just a complete meercat 64 antenna configuration.
   From looking at the output of 'top', my computer is not running out of memory etc.

3. Since calico-generic doesn't work, I try calico-stefcal.py
   (again modified to read in the oms-gridded-sky models). Here, stefcal
   seems to do some solving as a lot of stuff gets printed out on the
   terminal, but no calibration corrections are applied to the input data,
   and nothing appears in an inspector plot of the stefcal gains. I suspect
   that I'm not setting some flag here as I haven't used stefcal before -
   my current parameters are also in the .tdl.conf file. Could you please
   try a run and tell me what I'm not specifying correctly. Also I note that many of the stefcal parameters have no 'information' section associated with them so I don't have the slightest idea what they do.

So -- I'm stuck with the snail-like CASA gaincal and applycal routines until I can get one of these calibration schemes up and running. Advice needed!

[twillis449]

RE: 2) above - I meant everything works fine for the first ~2500 integrations, and then the meqserver goes 'poof'!!