Mapping Dark Matter in Galaxy Clusters with Gravitational Lensing

Abstract

This thesis uses gravitational lensing to map the distribution of dark matter around galaxy clusters, and to infer their formation history. Galaxy clusters are the oldest and most massive gravitationally-bound objects in the Universe, exploited in the most discriminating tests of cosmology. It is therefore essential to understand the astrophysics of their formation. Indeed, clusters grow through filamentary connections with surrounding large-scale structures - and to chart their history is to trace the evolution and trajectory of the Universe itself.

Gravitational lensing is the apparent distortion in the shapes of distant galaxies due to foreground mass, such as a galaxy cluster. Many software algorithms have been developed to measure gravitational lensing and to reconstruct the distribution of foreground mass. In this thesis, we assess the performance of two mass-mapping techniques, using mock images of the BAHAMAS simulation, where the true distribution of mass is known. We find the methods suitable for different applications: MRLens suppresses noise without bias, while Lenstool suppresses noise further, but at a cost of over-estimating the mass in cluster outskirts (R>1Mpc) by up to a factor 2. We also develop a filter to search for large-scale filaments connected to galaxy clusters. We then use these calibrated techniques, and the largest ever mosaic of Hubble Space Telescope imaging, to study galaxy cluster MS 0451-03 (z=0.54). We map the distribution of its dark matter, and discover six group-scale substructures, linked to the cluster halo by three possible filaments. By comparing lensing results with analyses of X-ray emission and optical spectroscopy, we conclude that the cluster collided with another 2--7 Gyr ago. Its star formation was quenched and its gas was heated; its gas has still not yet relaxed, and the dark matter halos are approaching second apocentre.

In the next decade, space-based telescopes will reveal this richness of detail about tens of thousands of galaxy clusters. If these observations are properly calibrated, via studies like this thesis, they will bring a new era of precision cosmology. As a final step towards this future, we present preliminary results from two ongoing projects: using deep learning to further suppress noise in lensing mass reconstruction, and the first successful measurement of gravitational lensing from a balloon-borne telescope at the edge of space.