S. Digel, 4 October 2005
I've documented in Confluence the problems that I'm aware of with the checkout 3 sky model. There's quite enough wrong with the simulation to make it worth regenerating the data in case anyone wants to, say, analyze pulsars, with a simulated sky before DC2.
I've regenerated the data; it takes about 10 CPU hours on one of the noric-new computers.
The new data pass some of the tests - like having SAA passages and having isotropic emission that is truly isotropic. I have not been able to test the pulsar light curves yet because the pulsar tools assume that MJDREF is 54101 (Masa & James are going to fix this shortly or perhaps have already) and I am out of the business of manually changing MJDREFs and event times. Also, GRBobsmanager internally assumes that the times in the source specifications are with respect to January 1, 2007, which causes problems in the case that they are MET. The solution is not hard (I think) but because of the way the random number seed is selected, by shifting to measuring times with respect to the correct reference date, the GRBs will all be different (even though they will correspond to the same absolute times as before).
For some unspecified period of time (depending on when the server needs to come down for further development), Tom can post updated checkout 3 data in the server at the GSSC. I do not know the current state of the data server at SLAC, but at last look it was very much in the prototyping stage and could not yet make FITS files.