Bayesian large-scale structure inference with cosmological velocities and fast radio bursts

Abstract

This thesis is concerned with extracting information about cosmology from peculiar velocities, the motion of objects induced by cosmic matter field inhomogeneities, and fast radio bursts, short-lived radio signals of unknown origin. The former provide information about the dark matter peculiar velocity field, while the latter inform about the integrated dark matter density contrast. I develop upon the Bayesian hierarchical model, BORG in order to constrain the 3D matter distribution and dynamics by including non-linear physical modelling into the reconstruction of the local peculiar field for the first time. I demonstrate this method as a proof-of-concept with simulated data, and apply it to the Second Amendment catalogue of type-1a supernovae. I study the statistics of the dark matter velocity dispersion tensor, which is an avenue to study the behaviour of objects undergoing gravitational collapse. In particular, I investigate how the statistics of the velocity dispersion vary as a function of simulation and cosmological parameters. I further explore the behaviour of the distributions of the velocity dispersion conditional on density. These provide a method to include velocity dispersion information into velocity field reconstructions. Lastly, I explore what cosmological information can be extracted from existing fast radio burst datasets. For the first time, these bursts are modelled as existing within an inhomogeneous universe. The method of Bayesian inference is used to constrain cosmological parameters from a set of localised and a set of unlocalised fast radio bursts. Furthermore, I demonstrate a proof-of-concept method for constraining the dark matter initial conditions using these bursts, finding that we will need at least three orders of magnitude more data than currently exists. I also investigate using these bursts to augment conventional datasets used in initial conditions inference, demonstrating that, with sufficient bursts, their addition can improve our knowledge of the initial conditions.