We present a gravitational lensing study of the massive galaxy cluster A2219 (redshift 0.22). This investigation is based on multicolour images from U through H, which allows photometric redshifts to be estimated for the background sources. The redshifts provide useful extra information for the lensing models: we show how they can be used to identify a new multiple-image system (and rule out an old one), how this information can be used to anchor the mass model for the cluster, and how the redshifts can be used to construct optimal samples of background galaxies for a weak lensing analysis. Combining all results, we obtain the mass distribution in this cluster from the inner, strong lensing region, out to a radius of 1.5 Mpc. The mass profile is consistent with a singular isothermal model over this radius range. Parametric and non-parametric reconstructions of the mass distribution in the cluster are compared. The main features (elongation, sub-clumps, radial mass profile) are in good agreement.

The lensing information provided by multiple images, arclets, and statistical distortions can all be formulated as linear constraints on the arrival-time surface, and hence on the mass distribution. This reduces cluster lens reconstruction from combined strong and weak lensing to a standard type of inversion problem. Moreover, the mass sheet degeneracy is broken if there are sources at different redshifts.

We apply a new non-parametric technique to reconstruct, with uncertainties, the projected mass distribution of the inner region of Abell 2218, using combined strong and weak lensing constraints from multiple-image systems and arclets with known redshifts. The reconstructed mass map broadly resembles previous, less detailed, parametric models, but when examined in detail shows several sub-structures, not necessarily associated with light but strongly required by the lensing data. In particular, the highest mass peak is offset by about 30 kpc/h50 from the main light peak, and projected mass-to-light in the directions of different cluster galaxies varies by at least a factor of 10. On comparing with mass estimates from models of the X-ray emitting gas, we find that the X-ray models under-predict the enclosed mass profile by, at least a factor of 2.5; the discrepancy gets worse if we assume that mass traces light to the extent allowed by the lensing constraints.

We describe a new non-parametric technique for reconstructing the mass distribution in galaxy clusters with strong lensing, i.e., from multiple images of background galaxies. The observed positions and redshifts of the images are considered as rigid constraints and through the lens (ray-trace) equation they provide us with linear constraint equations. These constraints confine the mass distribution to some allowed region, which is then found by linear programming. Within this allowed region we study in detail the mass distribution with minimum mass-to-light variation; also some others, such as the smoothest mass distribution. The method is applied to the extensively studied cluster Abell 370, which hosts a giant luminous arc and several other multiply imaged background galaxies. Our mass maps are constrained by the observed positions and redshifts (spectroscopic or model-inferred by previous authors) of the giant arc and multiple image systems. The reconstructed maps obtained for \a370 reveal a detailed mass distribution, with substructure quite different from the light distribution. The method predicts the bimodal nature of the cluster and that the projected mass distribution is indeed elongated along the axis defined by the two dominant cD galaxies. But the peaks in the mass distribution appear to be offset from the centres of the cDs. We also present an estimate for the total mass of the central region of the cluster. This is in good agreement with previous mass determinations. The total mass of the central region is M=(2.0-2.7) 10^14 Msun/h50, depending on the solution chosen.

We propose a new technique to reconstruct non-parametrically the projected mass distribution of galaxy clusters from their gravitational lens effect on background galaxies. The beauty of our technique, is that it combines information from multiple imaging (strong lensing) and shear (weak lensing) as linear constraints on the projected mass distribution. Moreover, our technique overcomes all the drawbacks of non-linear methods. The method is applied to the first cluster-lens A370 and to the cluster-lens A2218 which is exceptionally rich in multiple images and arc(let)s. The reconstructed maps for both cases revealed unprecedented levels of detail.