Extrapolating to the real detector
All our studies have been carried out using Monte Carlo data. We expect differences with
real data. Here we outline some areas of concern.
Event generation
Event generators may not match well with real events or even with each other. In our
benchmark Z->mu mu events Isajet produces much more underlying event than Pythia. Pythia
with one background event is roughly comparable to Isajet with none.
The problem is more serious if our minimum bias bg events are not accurate because
there can be many of these in ine interaction. our CPU time is a sensitive function
of the number of bg events.
Detector response
The response of the real detector may differ from that of our simulation (d0gstar plus
digitization). For example, the CFT simulation assumes more than 10 detect photons per
fiber and a threshold of 0.5 while hardware is now producing 5-6 and a threshold of 2-3.
Detector inefficiencies
Reduced detector response and dead channels will lead to inefficiencies in the detector
elements. Our present track-finding paths make optimistic assumptions about these
efficiencies especially in the CFT. We can expect a significant increase in CPU time
if we allow for these inefficiencies or a significant loss in track-finding efficiency
if we do not.
Detector noise
If the detectors have more noise that expected, then we can expect an increase in the
CPU time. If the noise is severe, then resolution may be degraded.
D0 Global Tracking Status GCAS March 2000 dladams@uta.edu