Cosmic Shear and the Intrinsic Alignment of Galaxies

Abstract

Cosmology has recently entered an era of increasingly rich observational data sets, all being in agreement with a cosmological standard model that features only a small number of free parameters. One of the most powerful techniques to constrain these parameters and test the accuracy of the concordance model is the weak gravitational lensing of distant galaxies by the large-scale structure, or cosmic shear. This thesis investigates the optimisation of present and future cosmic shear surveys with respect to the extraction of cosmological information and deals with the characterisation and control of the intrinsic alignment of galaxies, a major systematic in cosmic shear data.<br /> A detailed derivation of the covariance of the weak lensing convergence bispectrum is presented, clarifying the relation between existing formalisms, providing illustration, and simplifying the practical computation. The results are then applied to forecasts on cosmological constraints by cosmic shear two- and three-point statistics with the proposed Euclid satellite. Besides, a novel method to assess the impact of unknown systematics on cosmological parameter constraints is summarised, and several aspects concerning the weak lensing analysis of the Hubble Space Telescope COSMOS survey are highlighted.<br /> A synopsis of the current state of knowledge about the intrinsic alignment of galaxies is given, including its physical origin, modelling attempts, simulation results, and existing observations. Possible corrections to the prevailing model of intrinsic alignments are suggested, before presenting new observational constraints on matter-intrinsic shear correlations using several galaxy samples from the Sloan Digital Sky Survey. For the first time a data set with only photometric redshift information is included, after developing the formalism for correlation function models that take photometric redshift scatter into account. The intrinsic alignment signal of early-type galaxies is found to increase with galaxy luminosity and to be inconsistent with the default redshift evolution of a widely used model, both with high confidence.<br /> Moreover the nulling technique is developed, a method to remove gravitational shear-intrinsic ellipticity correlations from cosmic shear data by solely relying on the well-known redshift dependence of the signals, and its performance on realistically modelled cosmic shear two-point statistics is investigated. Subsequently, the principle of intrinsic alignment boosting, an inverse and likewise geometrical approach capable of extracting the intrinsic alignment signal from cosmic shear data, is derived. Both techniques are shown to robustly remove or isolate the intrinsic alignment signal, but are subject to a significant loss of statistical power caused by the similarity between the redshift dependence of the lensing signal and shear-intrinsic correlations in combination with strict model independence.<br /> As an alternative ansatz, the joint analysis of various probes available from cosmic shear surveys is considered, including cosmic shear, galaxy clustering, lensing magnification effects, and cross-correlations between galaxy number densities and shapes. The self-calibration capabilities of intrinsic alignments and the galaxy bias in the combined data are found to be excellent for realistic survey parameters, recovering the constraints on cosmological parameters for a pure cosmic shear signal in presence of flexible parametrisations of intrinsic alignments and galaxy bias with about a hundred nuisance parameters in total.