Enabling non-parametric strong lensing models to derive reliable cluster mass distributions - wslap

In the strong lensing regime, non-parametric models struggle to achieve sufficient angular resolution for a meaningful derivation of the central cluster mass distribution. Cluster members perturb lensed images and generate additional images, requiring high-resolution modelling. In practice, the required resolution for a fully non-parametric mass map is not achievable because the separation between lensed images is several times larger than the deflection angles by member galaxies. Here we bypass this limitation by incorporating a simple physical prior for member galaxies, using their observed positions and their luminosity scaled masses. This high-frequency contribution is added to a relatively coarse Gaussian pixel grid used to model the more smoothly varying cluster mass distribution, extending our established wslap code (Diego et al.). We test this new code (wslap+) with an empirical simulation based on A1689, using all the pixels belonging to multiply lensed images and the observed member galaxies. Dealing with the cluster members this way leads to stable convergent solutions, without resorting to regularization, reproducing well smooth input cluster distributions and substructures. We highlight the ability of this method to recover 'dark' subcomponents and other differences between the distributions of cluster mass and member galaxies. Such anomalies can provide clues to the nature of invisible dark matter, but are difficult to discover using parametrized models where substructures are modelled on the basis of the visible data. With our increased resolution and stability, we show that non-parametric models can be made sufficiently precise to locate multiply lensed systems, thereby achieving fully self-consistent solutions without reliance on input systems from less objective means.