Abstract
Ultra light Axions are scalar field particles with a mass in the order of 10−19 eV and lower. In this thesis, we consider an Axion mass range of max ∈ [10−28eV,10−22eV]. The Axions exhibit a non-negligible pressure on scales smaller than the Jeans length of the Axions, which serves to suppress structure formation. The Jeans length describes the scale below which oscillations in the density perturbations occurs. Meanwhile, on scales larger than the Jeans length, the Axions behave similarly to cold dark matter.
Full simulations of Axions are computationally very expensive. In this thesis, we explore an alternate method to predicting the non-linear matter power-spectrum in a mixed Axion dark matter model, that is computationally cheap. This is done by using AXIONCAMB to generate an initial power-spectrum for a given combination of cosmological parameter, Axion mass and Axion fraction. This power-spectrum is then used as an initial condition at high redshift for a fast approximate N-body simulation, which evolves the system and outputs the non-linear matter power-spectrum. The caveat with this approach is that the Axion physics are only present in the initial conditions, as this approach evolves the particles as pure cold dark matter particles.
We generate a data-set using this approach and use machine learning to train an emulator to it. This emulator takes 6 input parameters — ΩCDM (the cold dark matter energy density), log10 As (the amplitude of scalar fluctuations), fax (the fraction of the cold dark matter budget that is Axions), log10 max (the Axion mass), z (redshift) and k (scale) — and makes a prediction for the ratio between the non-linear matter power- spectrum of a mixed Axion dark matter model and ΛCDM. Comparing this emulator to axionHMcode, which calculates the non-linear matter power-spectrum analytically through a modified halo model, shows that the emulator gives similar results, except for some missing features, which can be chalked up to Axion physics being absent in the simulations.
With this emulator, predictions of the matter power-spectrum can be computed very quickly, making it a useful tool for constraining Axion dark matter models using, for example, weak lensing data from current or near future surveys, like Euclid.
Supervisor: Associate Professor Hans Arnold Winther, Institute of Theoretical Astrophysics, UiO (main-supervisor)
Intern. assessor: Professor Øystein Elgarøy, Institute of Theoretical Astrophysics, UiO
Extern. assessor: Associate Professor David Alonso, University of Oxford