Masterarbeiten in der Astroteilchenphysik und Kosmologie (2019-2020)

 

Project 1: Cosmological Tests of Exotic Physics in the Neutrino Sector

Some non-standard physics in the neutrino sector has been proposed in order to solve a number of tensions and anomalies in cosmological and laboratory data: Hubble tension, low matter fluctuation amplitude, anomalous neutrino oscillations... These models involve small neutrino self-interactions, or light sterile neutrinos interacting with a scalar field, or low-temperature reheating preventing neutrino thermalisation, or large neutrino chemical potentials... After incorporating these models in the Einstein-Boltzmann solver CLASS, we will compare them with more recent and precise data than in the previous literature, and check to which extent they can still address the tensions and anomalies.

Project 2: Cosmological Tests of Dark Matter Interactions

Dark Matter model building usually invokes a new dark sector with several new particles and interactions, that could lead to very small interactions between DM particles and other particles of the visible or dark sector. If such interactions exist, they have a potential to solve a number of tensions and anomalies in cosmological data: Hubble tension, low matter fluctuation amplitude, late thermal history... Most of these models have already been incorporated in the Einstein-Boltzmann solver CLASS and compare to the data using a top-bottom approach, in which one starts from a particle physics model and fits its parameters to the data. Here we will carry on such investigations with a more bottom-up approach, varying several parameters at the same time before identifying the characteristics of the best model to resolve the tensions in the data. We will also consolidate the implementation of dark sector models in CLASS.

Project 3: Cosmological N-body simulations with massive Neutrinos

In the recent years significant progress has been made in the field of relativistic N-body simulations. We have developed a novel method for simulating the impact of massive neutrinos on cosmological structure formation with significant numerical advantages over existing techniques. We have implemented this method in the N-body code Gevolution as a proof-of-concept. The student will learn the methodology, and implement the approach in a state-of-the-art N-body code. The goal is to obtain an efficient implementation that can be used to run the large-scale simulations required to analyse the data of future experiments such as ESA’s Euclid satellite mission.

Project 4: Developing a Likelihood to Interpret the Euclid Satellite Data

ESA’s Euclid satellite will be launched in a few years and will map the large scale Structure of the universe with unprecedented accuracy. Several previous students in the group have worked on a mock likelihood, i.e. a code simulating the analysis of fake data. This is very useful to forecast the sensitivity of the experiment to cosmological parameters (such as the neutrino mass or the dark energy equation of state), and to prepare the implementation of the true likelihood. In this project, we will improve over previously existing mock likelihoods by adding new ingredients (super sample variance, inclusion of additional relativistic effects). We will then perform several tests and comparisons between different approaches (P(k) versus C_l approach, photometric vs spectroscopic surveys, role of the number of redshift bins, neutrino mass modelling, nonlinear cutoff modelling, Limber approximation, line-of-sight integral versus FFT method).

Project 5: Neural Networks in Boltzmann Codes

Last year, a master student has developed a new method to speed up Einstein-Boltzmann Solvers (EBSs) like CLASS or CAMB, by replacing the integration of a big system of differential equation by a neutral network approach, which non-trivial design has been dictated by several physical considerations. A proof-of-concept paper will be published in July. This very promising work is the first step of an ambitious program that will lead the development of an adaptive machine learning network in the community using EBSs. This year we need a second student to carry on this program. The student will compare the performances of various possible forms of the neutral network inside CLASS.

Project 6: Interpolating Cosmological Observables

The comparison of cosmological observations to theoretical model requires a large number of extremely heavy simulations of the evolution of the universe. In this project, we will try to build a very modern and innovative interpolating scheme that will allow us to guess the results of these simulations for free for particular parameter values, given a small set of already computed models. To do that, we will use a combination of non-parametric interpolators and machine-learning techniques. If successful, the result of this project will allow us to test cosmological models against data at a fraction of the computational cost. This is a very hot topic in the present era of data-driven Cosmology.

Project 7 Shared With Prof. Krämer: Polarized Synchrotron Emission From DM Annihilation/Decay

When Dark Matter particles annihilate or decay in the magnetic field of the galaxy, they generate synchrotron emission. The intensity of this signal has been studied a lot, but not its polarisation. The student will try top compute it using some existing public numerical codes (like Galprop or Hammurabi). The signal predicted by the code will then be compared to the Planck polarisation maps (designed to measure CMB polarisation, but in which this synchrotron signal would be a foreground). The goal is to try to extract bounds on the dark matter decay rate and annihilation cross-section.