TDCOSMO XXII: Triaxiality and projection effects in time-delay cosmography

Abstract

Constraining the mass-sheet degeneracy (MSD) is crucial for improving the precision and accuracy of time-delay cosmography. Joint analyses of lensing and stellar kinematics are widely adopted to break the MSD. A 3D mass and stellar tracer population is required to accurately interpret the kinematics data. Our forward-modeling procedure aims at evaluating the projection effects of strong lensing and kinematics observables and to determine an optimal model assumption for the stellar kinematics analysis leading to an unbiased MSD and $H_0$. We numerically simulate the projection and selection effects for both a triaxial ETG sample from the IllustrisTNG simulation and an axisymmetric sample that matches the properties of slow-rotator galaxies representative of the strong lens galaxy population. Using the axisymmetric sample, we generate mock kinematics observables with axisymmetric Jeans Anisotropic Modeling (JAM) and assess kinematic recovery under different model assumptions. Using the triaxial sample, we quantify the random uncertainty introduced by modeling triaxial galaxies with axisymmetric JAM. We show that spherical JAM analysis of spatially unresolved kinematic data introduces a bias of up to 2%-4% (depending on the intrinsic shape of the lens) in the inferred MSD. Our model largely corrects this bias, resulting in a residual random uncertainty in the range of 0-2.2% in the stellar velocity dispersion (0-4.4% in $H_0$) depending on the projected ellipticity and the anisotropy of the stellar orbits. This residual uncertainty can be further mitigated using spatially resolved kinematic data which constrain the intrinsic shape. We also show that the random uncertainty in the velocity dispersion recovery using axisymmetric JAM for axisymmetric galaxies is at the level of < 0.24%, and the uncertainty using axisymmetric JAM for triaxial galaxies is at the level of < 0.17%.