Thesis BELLE2-UTHESIS-2022-002

Predicting injection backgrounds at SuperKEKB using neural networks.

Lukas Herzberg ; Dr. Benjamin Schwenker

II. Physikalischen Institut Göttingen

Abstract: Every process that has ever been observed in particle physics respects the conservation of baryon and lepton numbers. That means every time a particle is created or destroyed, a corresponding antiparticle is also created or destroyed at the same time. Following this principle, the difference between the number of particles and the number of antiparticles will always be the same, yet if we look at the observable universe it is clear that this difference is not zero. Matter is abundant while antimatter is absent in comparison. It is hypothesised that this imbalance came to be in the early stages of the universe in a process called baryogenesis, but the exact mechanism can not be explained by modern physics. One possible approach to find an explanation is to study the violation of the charge conjugation parity (CP) symmetry. CP-symmetry states that the laws of physics should stay the same upon exchanging a particle with its antiparticle and mirroring its spatial coordinates. In 1964, James Cronin and Val Fitch showed that CP-symmetry is broken in the neutral kaon system for which they earned the Nobel Prize in 1980. This discovery shows that physics is different for matter and antimatter and therefore is a natural starting point to investigate the matter-antimatter discrepancy. For this purpose the KEKB particle collider and Belle detector have been constructed at the High Energy Accelerator Research Organisation known as KEK in Tsukuba, Japan. The experiment ran from 1999 to 2010, where it found evidence for CP violation in B mesons which confirmed the theory established by Cabibbo, Kobayashi and Maskawa of CP violation caused by flavor changing charged current in the quarks sector and led to the Nobel Prize for Kobayashi and Maskawa in 2008. Due to the success of the experiment an upgrade of the accelerator and detector has been commissioned, which has been realized in the form of SuperKEKB and Belle II. The goal of this new experiment is to do precision measurements of rare processes that were limited by the amount of data taken by the Belle experiment, which achieved a total integrated luminosity of 710 fb^{-1}. To do this, Belle II is set to collect around 50 times more data than its predecessor, mostly due to a 40 times increase in luminosity, a quantity that describes the number of collisions in a certain time period, of the KEKB. Increasing the luminosity by such a large margin is not trivial by any means and raises many problems such an increase in undesired effects seen by the detector. These undesired effects are universally referred to as backgrounds. In this thesis a machine learning approach will be introduced to better understand the source of one of these types of backgrounds, so that its impact might be mitigated in the future of Belle II and SuperKEKB.

Note: Presented on 31 03 2022
Note: BSc

The record appears in these collections:
Books, Theses & Reports > Theses > Undergraduate Theses

 Record created 2022-06-13, last modified 2022-06-13

Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)