I have implemented the algorithms I suggested in my earlier message
for three-supermodule and non-adjacent supermodule events, and they
are basically correct. I don't yet know how robust they are, but they
usually give the correct answer and require no new calibrations. For
example, I have successfully reconstructed a N/1/2 event (R7504
E11985), a 1/2/3 event (R7504 E16489) using SM2 to bridge between
mvax1 and mvax2 times, a 2/3/4 event (E7504 E5449), a 3/4/5 event
(R7504 E3403), and a 4/5/6 event (R8678 E6918) using SM5 to bridge
between mvax2 and mvax3 times. The algorithm also handles many (but
not all) non-adjacent pairs. Ironically, it can do 1/2/3 events but
not 1/3 events. However, when I threw 10,000 upgoing muon events with
Colin's event generator, of which about 4100 passed all my cuts, only
14 events (0.3%) were rejected because they involved supermodule
combinations the algorithm could not adjust. I have the algorithm
coded in C++, which I can show to anyone interested.
I am still working on the solution to the problem of bad interERP
hardware infecting interERP calibrations in other parts of the
detector. The things I have tried so far give very little
improvement. Meanwhile, Colin has made some significant progress in S
face timing calibrations.
Bob
-- Why worry? There should be sunshine after rain There should be laughter after pain These things have always been the same So why worry now? -- Dire Straits