Speaker
Mr
Laurent Desorgher Desorgher
(Paul Scherrer Institut)
Description
Monte Carlo code as Geant4 in their forward unbiased simulation mode are the most precise for
computing the radiation effects in space components of payloads and spacecraft.
However it takes most of the time a huge amount of computing time
to achieve a good statistical precision of the simulation results,
as the components where the dose is computed are very small
compared to the entire size of the geometry.
To decrease this computing time efficiently without loosing in the correctness of the
simulation results different biasing techniques may be used. The biasing technique that adapts probably the best to
the case of a tiny sensitive component in a large geometry and within an extensive external source of particles,
is the reverse (also called adjoint) Monte Carlo technique.
In this simulation mode adjoint particles (e-, gamma, protons,..) are generated on the sensitive component
and are tracked backward till the external source. During this backward tracking, physical
processes are also applied but in a reverse mode where
the projectile and products exchange their role. These reverse processes keep the same differential cross sections
but the integral cross section is obtained by integrating over the energy of the product.
In order to keep the same statistical significance of a reverse track compared to
an equivalent forward track, weight correction needs to be applied at the occurrence of any reverse processes
as well as along the transport of the particles.
The Reverse Monte Carlo technique is available in geant4 since the release 4.9.3 for most of electromagnetic processes.
While it has been shown to work well for the examples provided inside Geant4,
recently in the frame of the preparation of the Juice mission significant discrepancies between the reverse
and forward mode have been reported. Moreover for some materials the occurrence of statistical peak
in the computed dose lead to a significant increase of the simulation time.
A review of the Geant4 reverse Monte Carlo mode inside Geant4 has been performed
in order to identify the reason for these problems.
In this paper we report on these tests, present some modifications that have been already brought to the code and discuss
further possible modifications.
Primary author
Mr
Laurent Desorgher Desorgher
(Paul Scherrer Institut)
