The Weak Lensing Frontier: Overcoming Challenges for Precision Cosmology

Abstract

Weak gravitational lensing by the large-scale structure, so-called cosmic shear, is a powerful cosmological probe. It causes coherent distortions in the observed shapes of distant galaxies and has been instrumental in placing tight constraints on cosmological models. While analysis methods have advanced considerably since the first detection of cosmic shear over two decades ago, significant challenges remain, particularly in the context of upcoming precision surveys. The first part of this thesis addresses the dominant source of statistical noise in weak lensing, so-called “shape noise”. Shape noise describes the fact that the intrinsic orientation of galaxy shapes is unknown to the observer. This intrinsic variance (σWL ≈0.26) far exceeds the percent-level lensing signal, which consequently needs to be extracted by averaging over a large ensemble of galaxies. We introduce a new technique, kinematic lensing (KL), which combines photometric shape data with resolved spectroscopic velocity fields to infer intrinsic galaxy shapes and directly estimate the gravitational shear.We describe the KL inference pipeline, demonstrate unbiased shear recovery on simulated data, and present a pilot measurement showing that KL can suppress shape noise by an order of magnitude. The second part of the thesis focuses on modeling uncertainties on nonlinear scales that are driven by complex baryonic processes tied to galaxy formation and feedback. These effects remain poorly understood and can bias cosmological inference. We present a comprehensive comparison of different modeling approaches, evaluating both their flexibility and physical interpretability. Using a suite of hydrodynamical simulations, we also explore how baryonic feedback depends on cosmology and quantify the resulting bias in simulated parameter constraints for future surveys.