G4 Tracing precision tests in magnetic field

1
G4 Tracing precision tests in magnetic field
The purpose of this study is to verify if the
problems encountered in G4 tracing precision
tests in magnetic field in G4 version 4.4.0 as
described in document 570 (BTeV document DB)
are still present in current G4 version
(4.6.0). Then we would like to find a strategy
to define parameters controlling tracing
precision for our detector simulation
2
In the 9th slide of document 570 was evidenced
that the same particles starting from the same
starting point, with fixed values of dChord,
dIntersection and dStep, do not intersect the
same volume at the same point in different runs.
In Geant 4-06-00 this problem is solved.
3
(No Transcript)
4
We observe that dIntersection is not always a
good parameter it is defined as the accuracy in
computing the intersection with a volume in this
projection of the previous plot we can see that
the accuracy of the simulation is (for some
values of dChord) 1 order of magnitude larger
than dIntersection.
5
In these plots we can observe the behavior of the
simulation accuracy in larger scales of dChord vs
dIntersection. What we note is that there are 3
different areas, with 3 different orders of
accuracy. If we stay in the most precise area the
accuracy is always in the order of magnitude of
dIntersection parameter.
6
Setting parameters at reasonable values for our
simulation (TheBeast) dIntersection between 1
micron and 100 microns (pixel accuracy order of
magnitude) and dChord 1 order of magnitude
greater (to keep computing fast)...
... low dependency on dIntersection (except for
particular values of dChord), accuracy worse
than expected, strong dependency on dChord.
?????
7
(No Transcript)
8

• A possible region for the parameters giving a
  precision in the order of magnitude of one
  micron is
• 110-3 lt dIntersection lt 510-3
• 210-2 lt dChord lt 610-2
• dOneStep any value

9
Conclusions

• We have carefully to choose experimentally the
  magnetic field tracing parameters (dIntersection,
  dChord, dOneStep) to match our precision
  requirements
• Once a set of parameters has been chosen the
  results are reproducible and stable (problem
  solved respect older G4 version)
• It needs interaction with G4 developers to
  clarify strange behavior and discontinuity in
  tracing precision respect the tracing parameters
  (nothing changed respect older G4 version)
• The strange behavior and discontinuity seems to
  be independent from dOneStep parameter
• A possible region for the parameters giving a
  precision of about one micron is
• 110-3 lt dIntersection lt 510-3
• 210-2 lt dChord lt 610-2
• dOneStep any value