Weak gravitational lensing as a probe of large-scale structure and galaxy formation

The study of the relation between galaxies and their surrounding haloes of dark matter is a fundamental component to understand the evolution of the large-scale structure of the Universe. In this thesis, we investigate how galaxy-galaxy lensing, the distortion of the shapes of background galaxies around selected foreground lens galaxies, can help to elucidate this interplay together with complementary galaxy clustering measurements. Galaxy-galaxy lensing, on one hand, provides a direct estimate of the mass inside a given aperture and also its distribution, which allows for connecting certain classes of galaxies, chosen according to properties such as stellar mass and colour, to the mass and shape of the encompassing dark host structures. Galaxy clustering, on the other hand, describes the spatial distribution of galaxies and the combination of the two probes can be used to jointly constrain the cosmological parameters for the matter fraction Ωm and the amplitude of the matter fluctuations σ8. This thesis is split into three parts addressing three outstanding challenges each for small-scale galaxy-galaxy lensing to act as a competitive probe for current and future large-scale structure surveys. In Chapter 3 (Renneby et al., 2018), we investigate how a cosmological rescaling algorithm, which fast and cost-efficiently maps particle and halo distributions from one N-body simulation to another one with a different set of cosmological parameters, can be adapted to accurately predict galaxy-galaxy lensing profiles and quantify the induced errors. The subsequent Chapter 4 (Renneby et al., prepa) deals with verifying that both lensing and clustering probes yield consistent predictions in semi-analytical models of galaxy formation (SAMs) and hydrodynamical simulations. To conclude in Chapter 5 (Renneby et al., prepb), we examine the main systematic effect on lensing profiles, namely the imprint of baryonic processes, using a range of hydrodynamical simulations. The major findings are the following: In Chapter 3 we establish that an N-body simulation with a set of parameters (Ωm, σ8) can be used to emulate the lensing profiles for central galaxies with no further restriction in a different background cosmology with two principal biases in halo concentrations Δc and the positions of the halo splashback radii Δrsp. These biases can be predicted well with the concentration-mass-redshift relations presented in Ludlow et al. (2016) and the splashback radius-mass-redshift relations from Diemer et al. (2017). To continue, we discover that lensing and clustering observations in Chapter 4 point towards a consistent picture for the feedback prescriptions. The hydrodynamical IllustrisTNG simulation suite is in agreement with current constraints from the KiDS+GAMA surveys for stellar mass only selected samples as well as locally brightest galaxies (LBGs) in SDSS. For the Munich SAM L-Galaxies, constraints from LBG lensing and general clustering demand a weaker radio-mode AGN feedback and shorter dynamical friction merger time than the default setup in the latest model from Henriques et al. (2015). Still, this comparison also highlights difficulties in the two modelling frameworks to accurately predict the signal for intermediate mass red galaxies below < 10^11 Msun, where the observations suggest lower host halo masses for quenched satellite galaxies. This calls for improved environmental quenching and merging mechanisms in galaxy groups and clusters. Finally, we retrieve a similar baryonic imprint as previously established in the literature for specific lens samples (e.g. Leauthaud et al., 2017) with suppressions of 10 − 20% for 0.1 < r h−1 Mpc < 1 and show that it is generalisable to a large range of stellar masses and for central galaxies in groups. Despite their different galaxy formation recipes, the Eagle and IllustrisTNG simulations produce similar lensing profile descriptions consistent with observations. The considerable gas ejection of the AGN feedback implementation in the Illustris simulation puts it at the extreme end in terms of the extent of the suppression up to r ~ 5 − 6 h−1 Mpc whereas its successor IllustrisTNG achieves mass convergence at r ~ 1 − 2 h−1 Mpc. These radii are largely independent of the stellar mass of the samples, with a slightly larger impact for group class haloes where the AGN feedback is most efficient, and there is little redshift evolution to z = 1. We attempt to parameterise the effect using the baryonic correction model of Schneider & Teyssier (2015) for group and cluster-size haloes in the TNG300 simulation. We find that the model captures the main deformation features but that further work is required for it to properly adjust the gravity-only mass profiles