Abstract
Currently, the deepest images of the universe is provided by the Hubble Frontier Fields Program. These multi-band observations ranging from optical to near-infrared wavelengths are capable of detecting distant faint galaxies at z~11, when they are still in their infancy - an era that is yet uncharted by astronomy and is fundamental towards understanding galaxy formation and the nature of dark matter. The extreme depth of HFF images is made possible by not only investing long hours of integration in the Hubble Space Telescope, but also by the magnifying effect thanks to gravitational lensing by the most massive galaxy clusters.
In order to obtain the intrinsic properties of the lensed galaxies, the magnification must be quantified. This in turn requires a robust and accurate mapping of mass distribution in galaxy clusters. Previous studies commonly fitted the observations with mass haloes placed at presumed positions, or assuming that mass being traced by light, which significantly biased the derived lens model. We reconstructed the mass distribution with a regular grid of pixels accounting for the diffuse cluster component plus compact haloes accounting for local perturbations induced by member galaxies. The lens model derived with this 'non-parametric' approach suffers much less bias. Preliminary results of the first HFF cluster, Abell 2744, will be presented.