Measuring the distribution of stars and dark matter in galaxy-scale strong gravitational lenses: lessons from an automated approach

Abstract

In this thesis, I constrain the distributions of mass and light of the largest sample of galaxy-scale strong lens systems modelled using an automated approach to this date. New surveys will soon observe hundreds of thousands of galaxy lenses requiring reliable automated methods to be exploited. With this in mind, I present not only the successful model fit results, but also the reasons why some lenses failed initially and the strategies we adopted to ultimately fit all the galaxies with minimal intervention. I discuss what we can learn from this process that could benefit future large scale studies.

I propose and evaluate a likelihood cap method to avoid the underestimation of errors due to noisy likelihood evaluations that appear in pixel-based source reconstruction methods. I test the method on a large sample of mock galaxies which significantly improves the coverage probabilities on all of the model parameters. With this approach to error estimation I find that the Einstein radius is typically constrained to and the errors on the model parameters including density profile slope do not degrade with redshift. This will be beneficial for studies of galaxy evolution.

I measure an average mass density slope of with little intrinsic scatter for the (typically) early-type galaxies acting as lenses in our sample. This is consistent with those measured using an independent lensing and dynamics approach. More generally, this result supports the empirical coincidence that the total mass profiles of early-type galaxies can be well described by a single power law -- known as the `bulge-halo conspiracy'. However, I reveal a disagreement between our measurements of the coefficient describing the relationship between slope and surface mass density and that inferred for the slopes measured using the independent method. This can potentially be explained by finer structure in the mass density profile. Finally, I demonstrate that the `external shear' parameters commonly adopted in strong lens models can not be assumed to represent perturbations only external to the mass model. Instead, they highlight the inability of the power law to fit the distribution of mass in these types of galaxies. Future strong lensing studies will require more complex mass models to appropriately describe the lens galaxy, and avoid biases on high precision measurements of galaxy masses, cosmological parameters , and dark matter substructures.