Investigating the role of the baryon-dark matter transition in galaxy-scale gravitational lenses with ramifications for galaxy structure and cosmology

Abstract

Gravitational lensing is a powerful tool to study the structure of galaxies and cosmology, however the constraints from lensing are subject to degeneracies and cannot provide a unique solution. The lens model informs the ultimate choice of solution, and so it is critical that the lens model accurately reflects galaxy structure. Probably the most commonly-used lens model is a power-law ellipse+shear model. We show that this ellipse+shear model is unable to statistically explain the angular distribution of quad image systems. Considering additional complications to the azimuthal structure of a lens, we show that the observed angular distribution cannot be explained by $\Lambda$CDM substructure, but can be explained by a transition region between two mass components representing baryons and dark matter. The combination of offset centers, misaligned position angles, and Fourier components introduces enough asymmetry in a lens to explain the observed population. Because lensing is used to measure $H_0$, it is important to know the potential biasing effects that simplifying assumptions implicit to modeling can create. We therefore study the effect of the radial profile assumption (that the profile is a power law) and the azimuthal shape assumption (that the lens is ellipse+shear) on the recovery of $H_0$. To do so, we create mock lenses which are more complicated than the model, then fit their images with the model. For the radial structure, we find that when two-component lenses are fit with a power law, they return biased values of $H_0$. Worse, the bias does not match the analytical prediction, making it more difficult to account for. Stellar kinematic information, which in practice is used to inform the solution by providing a measure of mass, does not correctly inform the unbiased value of $H_0$ because the power-law model is inaccurate. For the azimuthal structure, different types of shape complications have different effects, but the recovered value of $H_0$ can be biased substantially, especially if the profiles are offset from one another. Finally, we discover that the image distance ratios of observed quads are statistically different from mock quads, indicating additional complications to the structure of lenses which have not yet been accounted for.