Physical Models and Statistical Methods for Testing Standard Cosmology

Abstract

A wide range of observations have lent support to the Lambda Cold Dark Matter (LCDM) model, the ‘standard model of cosmology’. Though this model’s parameters have been measured to high precision, much of its underlying physics remains unknown. This thesis presents a range of work aimed at testing LCDM. The first part of this thesis is devoted to two specific tests of LCDM. We begin with a test of the assumed non-interaction between dark energy and dark matter. Relaxing this assumption, we perform a data-driven reconstruction of the interaction history at low redshift. The second test investigates the observational basis of the Cosmological Principle: Is the matter dipole seen in radio sources consistent with the expectation from the cosmic microwave background dipole? We estimate the radio dipole from a recent all-sky sample of quasars using a novel Bayesian approach. We compare our results to LCDM predictions with a calculation of the probability distribution of the dipole amplitude. The second part of this thesis is devoted to cosmology with peculiar velocities, a promising probe of gravitational physics beyond the LCDM paradigm. We present a statistical method for extracting cosmological parameters from peculiar velocities derived from the Fundamental Plane. We show how cosmological inference can be performed so as to obviate the need to both assume a fiducial cosmology and construct a catalogue of summary statistics. Next steps towards inclusion of selection effects are discussed. Finally, we explore the theory of peculiar velocities in redshift space. We extend well-known results from galaxy clustering to the velocity field. We show that the analogous effect of redshift-space distortions is a damping that begins on quasilinear scales, and which may be considered a dissipative effect in the fluid description of matter. We offer two models for the velocity power spectrum in redshift space, and discuss challenges towards obtaining more realistic models.