Speaker
Description
Strong-lensing models of galaxy clusters are traditionally constrained by the positions of multiply imaged sources, leaving much of the information encoded in the resolved surface brightness of extended arcs unused. This talk presents a JAX-based modelling framework that combines multiple-image constraints with pixelated source reconstruction and GPU-accelerated Bayesian inference. Three modelling strategies are applied to the Carousel Lens: position-only modelling of five source systems, joint modelling of one pixelated extended source with independent image-position constraints, and simultaneous pixelated reconstruction of five sources at four distinct redshifts. The comparison demonstrates how extended surface brightness provides complementary constraints on the cluster mass distribution and improves the reconstruction of lensed source morphology. The talk also discusses the current limitations and prospects for large-scale pixelated modelling of cluster strong lenses.