DescriptionStrong lensing by galaxy clusters offers a unique and powerful tool with which to study the Universe, specifically through the ability to magnify faint and distant sources such as high-redshift galaxies. In order to determine the intrinsic properties of these galaxies, a model of the mass distribution must be made and the lensing quantities estimated. A wealth of data from ground- and space-based telescopes has resulted in lens models that are very precise, but not necessarily accurate.
In this work, we present the first three-dimensional lens model of a galaxy cluster, specifically the field J0850. We find that one of the arcs in this cluster is highly magnified, offering a unique chance to study a distant galaxy at exceptional resolution. We also present the first three-dimensional lens models of the six Hubble Frontier Fields (HFF), along with more traditional two-dimensional models. We then use these to study and quantify systematic errors introduced when line-of-sight galaxies are ignored or when their contributions are only approximated. Next, we gauge the importance of such errors by conducting a study of all submitted models for the six HFF clusters created by research teams worldwide. We find that, while the one-dimensional mass profiles are remarkably well-constrained, there exists a large scatter in magnifications. Further, the statistical error for each model is much smaller than the total error, indicating that there are systematic effects not being accounted for.
Finally, we present a new statistical framework that offers a way to self-consistently study the possible systematic errors impacting cluster lens models. We use this technique to explore effects due to two modeling choices that vary among research teams, specifically how cluster membership for galaxies is determined, and how mass is assigned to galaxies in the model via their luminosity using scaling relations. We examine both cases and measure how they affect model parameters and magnification maps, making recommendations on which aspects should be more carefully considered in the future. While the modeling choices we study here are not exhaustive, it illustrates how our framework can be used as a robust way to study and quantify errors. As statistical errors continue to shrink, systematic errors will become more prevalent, and thus must be better understood if we are to fully exploit the capability of cosmic telescopes.