Experimental Flavor Physics
print


Breadcrumb Navigation


Content

Bachelor Thesis Topics

b2display

Detection of anomalies in the SuperKEKB operation data

At the electron positron collider SuperKEKB luminosities should be reached that are more than an order of magnitude higher than at its predecessor. This is a very challenging task and often instabilities are observed that lead to the loss of beams. For a stable operation it is important to understand under which conditions this happens so that counter measures can be taken. The task of this project is to analyze the data from a large set of beam monitoring devices with machine learning methods to identify anomalous conditions.

Neural network for the classification of Y(4S) events (topic not available any more)

B factory experiments such as Belle II are well suited to measure branching fractions of B mesons because the number of Y(4S) -> B Bbar events can be determined well. However, the fractions of Y(4S) decays to pairs of neutral and charged B mesons are needed as well and their ratio may depend on the center of mass energy. For a measurement of this energy dependency a multivariate classifier should be developed that can distinguish between pairs of neutral and charged B mesons.

Analysis of semileptonic B meson decays with multiple pions (topic not available any more)

Semileptonic B meson decays, where the B meson decays to a lepton, a neutrino, and one or more hadrons, usually including a charm meson, are used for many measurements at Belle II. However, decays into a charm meson and multiple pions are still poorly constrained. Further measurements are needed to better understand the dynamics of semileptonic B meson decays and to find out if the multi pion modes could close the gap between the inclusive decay rate and the sum of exclusive modes.

Search for B0 -> J/psi K0 pi0 decays

B mesons can decay in many different ways. A good understanding of decay modes and rates is important for many analyses. A decay that was not measured so far is B0 -> J/psi K0 pi0. The task of this topic is the development of an analysis to either observe the decay or set a limit in its branching fraction.

Improvement of the eta selection with AI

Eta mesons are light uncharged mesons which are used in many analysis at Belle II. Their reconstruction is challenging as it is dominated by a large amount of background. The idea of this task is to improve the eta meson selection by employing artificial intelligence algorithms to reduce the amount of background.

Exploration of AI methods for software optimization (topic not available any more)

In recent years new developments in artificial intelligence had a large impact on many areas. In this project, the potential of AI methods for software development should be evaluated. On the one hand there are tools like GitHub Copilot that assist developers. On the other hand tools like AlphaDev may help to discover new algorithms. Knowledge of C++ is required for this project.

Break point detection using Kalman filters

Particles can be scattered or decay during their traversal of the detector. These cases show up as break points in the reconstructed tracks. The aim of the offered task is to develop a break point detection algorithm using Kalman filtering for the Belle II track finding following the method described here. The start point would be the already existing Kalman algorithm implemented in the Belle II software. Knowledge of C++ is required for this project.

Analysis Grand Challenge benchmarks

In modern particle physics experiments such as ATLAS or Belle II, the datasets to be analyzed are usually very large (O(10-100 TB)). In order to process them efficiently, highly specialized software such as ROOT or the SciKit HEP software stack in Python is required. Physicists need to be able to develop their analysis in rapid iterations. This in particular will be a major challenge in the future, as ever-increasing amounts of data also mean ever-longer analysis runtimes. The Analysis Grand Challenge is a project that deals with this topic and wants to show how modern software tools can be used to process large amounts of data in a user-friendly way. The aim of this bachelor thesis is to test and compare two different implementations of the Analysis Grand Challenge on the computing infrastructure of the LMU. It is a rather technical thesis, basic knowledge of Python and a strong interest in programming, data analysis and linux are required.

Further topics may be available.