A robust gravitational lens model for the massive merging galaxy cluster Abell 2744

Abstract

To elucidate the formation and early evolution of galaxies, it is necessary to search for and study galaxies in their infancy at redshifts approaching z~10. The ongoing Hubble Frontier Fields (HFF) program images six massive galaxy clusters that gravitationally magnify background galaxies to reach depths unrivalled by standard deep field imaging. In order to derive the intrinsic luminosities, sizes, and space densities of the lensed background galaxies, the magnification provided by the clusters must be quantified. Determining the magnification as a function of position requires constructing accurate cluster mass models. Constructing accurate mass models for the HFF clusters is particularly challenging because these clusters are in the midst of major mergers and therefore far from relaxed.

In my thesis, I model the mass distribution of the first cluster completed in the HFF program, Abell 2744, using a free-form approach that makes minimal assumptions about the cluster-scale distribution of dark matter. Specifically, the mass distribution is modeled with a uniform grid on the cluster scale, and with NFW-parameterized components on scales of individual cluster galaxies. We find that the reconstructed mass distribution on the cluster scale not only smoothly traces the overall distribution of cluster galaxies, but also exhibits structures that coincide with bright peaks in the X-ray emitting intracluster gas. To assess the robustness of the lens model, I show that the centroids of multiply lensed images can be generally reproduced to within 1" - a testament to the internal consistency of the model. I also show that the lens model generally reproduces internal structures seen in the lensed images with the correct distortion and orientations. Most importantly, I show that the predicted relative magnifications of multiple images agree very well with the observed relative fluxes to within ~0.25 mag (25%), the first time that such a test has been applied to any cluster lens model. The predicted absolute magnification at a single position in the cluster, however, is in slight discrepancy (1σ) with the magnification inferred recently for a lensed Type Ia supernova discovered following the publication of our work (Lam et al. 2014).

The minor inconsistency between the predicted and inferred magnification of the Type Ia supernova motivated a number of refinements that I then made to the modeling method. The first modification reduces the arbitrariness in the NFW-parameterized components by replacing it with the stellar light profile. The second modification improves the quality of the constraints by imposing stricter selection criteria for the lensed images. The refined lens model has a similar image plane dispersion, but predicts an absolute magnification that is in agreement within the uncertainties with the supernova-inferred value. These results demonstrate the significant progress I have made in reliably deriving the magnification of galaxy clusters, even though the cluster modeled is far from relaxed. To make further strides forward, it essential to model simulated cluster lenses to better assess the systematic errors, strengths and weaknesses of the method I have used, in the hope of identifying better approaches to remedy the weaknesses.