Abstract:
Muon Ionization Cooling Experiment (MICE) is an experiment of high-energy physics. It is used
to cool muon beams using ionization cooling. The target of MICE is to obtain a muon beam with
reduced phase space. Particle detection inside the cooling chamber is important to gain insight into
the process. It is a challenging task because millions of particles pass through the particle detectors,
and it becomes further complicated due to the presence of multiple scattering. When particles pass
through the detectors, multiple scattering occurs. Multiple scattering distorts the trajectories of
particles. As a result, the position and momentum of particles change, and the non-linearity of the
system increases. Track fitting algorithms are used for particle tracking in MICE. In the past,
Kalman Filters (KFs) and Particle Filters (PFs) and their adaptations have been used for non-linear
tracking in MICE experiments in the absence of multiple scattering. PF has been widely used in
non-linear systems, but it was not used for non-linear non-Gaussian multiple scattering in MICE.
Ignoring the multiple scattering or taking it as a Gaussian approximation, PF outperformed
Unscented Kalman Filter (UKF) and Unscented Particle Filter (UPF) when compared in terms of
accuracy but was more computationally complex than UKF. UPF was more computationally
complex than PF but solved the problem of particle degeneracy and shifted the particles to the
regions of high likelihood. In this thesis, non-linear non-Gaussian multiple scattering is studied by
using the filters, which perform well for non-linear non-Gaussian systems because multiple
scattering is non-Gaussian in nature, whereas most of the filtering techniques are designed for
Gaussian approximations. The effect of multiple scattering, in its true form, on performance the of
MICE has not been studied before. This will be the contribution of this research. In this research,
UKF, PF, and UPF are used to study the behavior of the system in the presence of non-Gaussian
multiple scattering and to compare their effectiveness in terms of accuracy and computational
complexity.
