Cluster strong lensing with hierarchical inference

Lensing by galaxy clusters is a versatile probe of cosmology and extragalactic astrophysics, but the accuracy some its predictions limited simplified models adopted to reduce (otherwise untractable) number degrees freedom. We aim at cluster lensing where parameters all cluster-member galaxies are free vary around common scaling relations with non-zero scatter, deviate significantly from them if only data require it. have devised Bayesian hierarchical inference framework, which enables determination scaling-relation hyperparameters, including intrinsic constraints (if given) stellar kinematic measurements. achieve this through BayesLens, purpose-built wrapper parametric codes for likelihood that can sample posterior on we release paper. run functional tests our code against simple mock datasets realistic uncertainties. The hyperparameters recovered within their 68% credibility ranges, positions "observed" multiple images accurately reproduced BayeLens best-fit model, without overfitting. shown an accurate description member attainable, despite large freedom, fast tractable inference. This extends beyond state-of-the-art current models. precise impact studies cosmography, evolution high-redshift populations then be quantified real clusters. While other sources systematics exist may significant in clusters, results show contribution scatter now controlled.