Bachelor Thesis With Prof. Krämer in 2021
The classification of gamma-ray sources with Gaussian processes
The Fermi-Large Area Telescope (Fermi-LAT) has detected almost six thousand point-like γ-ray sources. The majority of the observed sources are blazars, i.e. Active Galactic Nuclei with a jet of outflows pointing towards the line of sight. Blazars can be divided further into BL Lacertae objects and Flat Spectrum Radio Quasars based on their spectral properties. In this project you will learn to classify Fermi-LAT blazar candidates using modern machine learning methods. Specifically, you will investigate Gaussian processes, which allow for Bayesian inference and thus provide a reliable estimate of the uncertainty of the classification. The results of a reliable blazar candidate classification are valuable input for astrophysical studies and to guide future observational campaigns.
Quantum machine learning
Quantum machine learning encompasses approaches that use synergies between machine learning and quantum information. While machine learning algorithms are used to analyze large amounts of data, for example in particle or astroparticle physics, quantum machine learning employs qubits and quantum operations to enhance the computational speed. In this project you will learn the theoretical concepts of quantum computing and machine learning, and program a quantum machine learning algorithm for supervised classification: a quantum support vector machine. Combining quantum computation with established machine learning algorithms provides insights into the structure of quantum algorithms and has the potential to dramatically speed up computations.
Searching for new physics at the LHC with machine learning
Current and future searches for new physics at the Large Hadron Collider (LHC) focus on complex signatures and more model-independent methods. Machine learning is a powerful tool for finding and classifying physics beyond the Standard Model. For example, deep neural networks can separate and identify subtle signatures of new physics from the background of Standard Model events. Moreover, methods of unsupervised deep learning allow the detection of anomalous signatures of new physics without having to make specific assumptions regarding the underlying theory. In this Bachelor thesis, you will explore neural network architectures for classification or unsupervised learning and their application to detect exotic signatures from new physics models at the LHC.
Cosmic rays from primordial black hole evaporation
Primordial black holes (PBHs) could be generated in the early Universe if sufficiently large density perturbations in the primordial plasma collapse gravitationally. Depending on their mass and formation mechanism, they could contribute to dark matter. For masses below about 1017g, PBHs are expected to inject sub-GeV electrons and positrons in the Galaxy via Hawking radiation. In this thesis project you will study the production of cosmic electrons and positrons from PBH evaporation and their propagation in our Galaxy.
Synchrotron emission from Galactic cosmic ray electrons and positrons
Cosmic-ray electrons and positrons are produced in our Galaxy through different astrophysical processes, from the spallation of hadronic cosmic-rays to the acceleration in different stages of massive star evolution, such as supernova remnants and pulsars.
No matter their source, after being produced these particles propagate in the interstellar magnetic fields, and produce secondary emissions through the synchrotron process at radio and microwave frequencies. In this thesis project you will simulate the synchrotron emission from cosmic-ray electrons and positrons in our Galaxy using the public code GALPROP, and compare your prediction with the measurements of the radio and microwave sky in order to study the properties of cosmic rays electrons and positrons, and of the interstellar magnetic fields.