Hi,

Here are two variants of positron energy loss distributions as
simulated by GEANT. The results are drastically different.

The DOSP card was used to provide initial kinematics for both
simulations.  Highly polarized muons decayed in rest in the
target. Decay positrons were tracked through the detector and their
kinetic energy at 3 points was stored:

Ek1: abs(z)==4.16 cm  (exit from the central volume)
Ek2: abs(z)==32.16 cm (the middle of UVOL or DVOL)
Ek3: abs(z)==60.16 cm (the end of UVOL/DVOL)

Ek0  is the initial kinetic energy of positron,
cos0 is the cosine of initial angle of positron momentum relative to
     the detector.

The same cuts were applied to select a narrow range of initial energy
and angle.  The cuts are shown at the top of the plots.  (Note, that
0.05283Mev is the kinematic limit on the (full) positron energy.)

File eloss_dflt1.ps was produced using essentially our "default"
ffcard settings. To make eloss_thin1.ps the "STRA 1" card has been
added.  It is the "STRA" card that makes the difference. We don't set
STRA in our standard ffcards, and GEANT defaults it to 0.  STRA=1
turns on the collision sampling method to simulate energy loss in thin
materials. This seems to be more than relevant for our detector.

It looks that with the default settings mean energy loss is
underestimated by one order of magnitude! Also, the energy loss
distribution shows with STRA=1 a discrete structure which is
completely missing for STRA=0.

Simulation takes more CPU time with STRA=1 (digi on):

 **** TIME TO PROCESS ONE EVENT IS = 0.0729 SECONDS  // STRA=0
 **** TIME TO PROCESS ONE EVENT IS = 0.0909 SECONDS  // STRA=1

GEANT documentation says that "one should remember the dependence of
this (STRA=1) method on the photoelectric cross section data".

However, even a non perfect collision sampling is probably better than
a complete lack of it.  So, I suggest to use the STRA=1 setting for
our simulations.  Probably it should be put in the reference
e614.ffcards file. 

This may also be relevant for energy loss in tracking.

Does anyone know how well GEANT describes energy loss with its current
cross section tables? What is the uncertainty in the mean energy loss?
What is the uncertainty in the distribution shape?

Regards,
	Andrei

PS: the number of simulated events with STRA=1 is 1.4 times more than 
that with STRA=0. Still, it doesn't completely explain the difference in 
the number of histogram entries.

Description: GEANT energy loss, default settings , Filename: eloss_dflt1.ps

Description: GEANT energy loss, STRA=1 , Filename: eloss_thin1.ps

Created for the The Center for Subatomic Research E614 Project Projects Page.

Created by The CoCoBoard.