GPU-Accelerated Gravitational Lensing & Dynamical (GLaD) Modeling for Cosmology and Galaxies

Abstract

Time-delay distance measurements from strongly lensed quasars provide a robust, independent method for determining the Hubble constant ($H_0$). This approach cross-checks $H_0$ estimates from the distance ladder in the late universe and the cosmic microwave background in the early universe. However, the mass-sheet degeneracy in lensing models introduces systematic uncertainty, limiting precision. Dynamical modeling complements strong lensing by constraining the mass distribution with independent observational data. We develop a methodology and software framework for joint modeling of stellar kinematics and lensing data. Using simulated data for the lensed quasar RXJ1131$-$1131, we demonstrate that high-quality kinematic data can achieve $\sim$4% precision on $H_0$. Through extensive modeling, we examine the impact of the presence of a supermassive black hole in the lens galaxy and potential systematic biases in kinematic data on $H_0$ measurements. Our results show that imposing priors on black hole mass and orbital anisotropy, or excluding central kinematic bins, mitigates biases in $H_0$ estimates. By testing on mock kinematic data with systematic biases, we highlight the need for sub-percent control of kinematic systematics, which is achievable with current technology. Additionally, we leverage GPU parallelization to accelerate Bayesian inference, reducing a previously month-long process by an order of magnitude. This pipeline offers significant potential for advancing cosmological and galaxy evolution studies with large datasets.