We use cookies to ensure that we give you the best experience on our website. By continuing to browse this repository, you give consent for essential cookies to be used. You can read more about our Privacy and Cookie Policy.

Durham e-Theses
You are in:

Mining for cosmological information: Simulation-based methods for Redshift Space Distortions and Galaxy Clustering

CUESTA-LAZARO, CAROLINA (2022) Mining for cosmological information: Simulation-based methods for Redshift Space Distortions and Galaxy Clustering. Doctoral thesis, Durham University.

PDF - Accepted Version


The standard model of cosmology describes the complex large scale structure of the Universe through less than 10 free parameters. However, concordance with observations requires that about 95\% of the energy content of the universe is invisible to us. Most of this energy is postulated to be in the form of a cosmological constant, $\Lambda$, which drives the observed accelerated expansion of the Universe. Its nature is, however, unknown. This mystery forces cosmologists to look for inconsistencies between theory and data, searching for clues. But finding statistically significant contradictions requires extremely accurate measurements of the composition of the Universe, which are at present limited by our inability to extract all the information contained in the data, rather than being limited by the data itself. In this Thesis, we study how we can overcome these limitations by i) modelling how galaxies cluster on small scales with simulation-based methods, where perturbation theory fails to provide accurate predictions, and ii) developing summary statistics of the density field that are capable of extracting more information than the commonly used two-point functions. In the first half, we show how the real to redshift space mapping can be modelled accurately by going beyond the Gaussian approximation for the pairwise velocity distribution. We then show that simulation-based models can accurately predict the full shape of galaxy clustering in real space, increasing the constraining power on some of the cosmological parameters by a factor of 2 compared to perturbation theory methods. In the second half, we measure the information content of density dependent clustering. We show that it can improve the constraints on all cosmological parameters by factors between 3 and 8 over the two-point function. In particular, exploiting the environment dependence can constrain the mass of neutrinos by a factor of 8$ better than the two-point correlation function alone. We hope that the techniques described in this thesis will contribute to extracting all the cosmological information contained in ongoing and upcoming galaxy surveys, and provide insight into the nature of the accelerated expansion of the universe. [math mode missing closing $]

Item Type:Thesis (Doctoral)
Award:Doctor of Philosophy
Keywords:Cosmology, large scale structure, machine learning, simulations
Faculty and Department:Faculty of Science > Physics, Department of
Thesis Date:2022
Copyright:Copyright of this thesis is held by the author
Deposited On:25 Aug 2022 16:23

Social bookmarking: del.icio.usConnoteaBibSonomyCiteULikeFacebookTwitter