I read the ERP patch of GMACRO in the airport on my way home from
Mexico (I love my laptop!). I see no evidence of any intentional
inefficiency in the code. Probably the problem is with the geometry
-- the simulated positions of all the attico are near the actual
positions (most within about 3 cm), but I think it will be best to
analyze simulated data with a simulated geometry database that exactly
matches the simulated geometry. I will work on this and get back to
you.
By the way, I did discover one thing I'm not very happy with while
reading the code. If photoelectron fluctuations are turned on, I
think the way it decides if events near trigger threshold trigger or
not is not the most physical possible. However, I think in practice
our uncertainty in real data in setting the ERP energy threshold is
greater than any effect caused by this suboptimal algorithm. I'll
think more about it. If anyone else wants to think about it, here is
what GMACRO does: using the actual energy deposited in the tank
(according to GEANT), it calculates the ideal number of photoelectrons
seen at each end (this is called PEx) and then picks an integer from a
poisson distribution with mean PEx (the integer is called PEr). Then
it calculates the energy seen by each end independently, that is the
energy that would be calculated from PEr on each end, assuming the
position is known. If the average energy of the two ends is above
threshold, the box is considered to trigger. The real ERP trigger
processor does not know the position, so it uses the ratio of PEr on
each end to estimate the position and then comes up with one (not two
independent) estimates of energy.
Maybe that tequila last night didn't kill too many brain cells....
Bob
------------------------------------------------------------------------
Date: Tue, 26 Mar 1996 14:38:33 +0100 (CET)
From: BERNARDINI@axple3.le.infn.it
Cc: BERNARDINI@axple3.le.infn.it, SURDO@axple3.le.infn.it,
"VAXLNF::RONGA"@axple3.le.infn.it
Ciao. I'm working on upward going muon analysis, with
Antonio Surdo. Analyzing simulated data we found an
unexpected inefficiency of scintillator counters, mainly
for attico lateral boxes ( < 80 %), but also for
horizontal and lower MACRO boxes. Two questions :
1) Do you know if E. Diehl inserted some kind of
inefficiency in scintillator simulation ?
2) The SIMULATED geometry of attico scintillators has
been checked in order to verify its compatibility
with streamer and lower MACRO geometry ?
THANK YOU
Paolo