Varsha's talk is here in pdf This was pre-Blessed already. Comments included: use bunch luminosity instead of instantaneous: luminosity varies notable from bunch-to-bunch. The Delta-X and Nstub and Em and Ehad plots seem to be using at least partly the same data. The Nstub monte carlo is wrong--the noise hit rate was tuned for lower luminosity and so the 2-hit rate is wrong for high luminosity. The monte carlo Ehad Em is about 5% low--maybe a problem with Gflash?
The reconstruction efficiency is 10% lower for monte carlo than for data. It would be useful to look at this as a function of phi, and to dump a few dozen reconstructed monte-carlo events to try to find out where the effect is. It might be due to using alignment constants with opposite signs between reconstruction and monte carlo (which would be a blunder on my part), which ought to show up more dramatically where the misalignment is greater.
Modified 16-January-2007 at 15:09
http://hep.physics.wisc.edu/~jnb/imu/16Jan2007