We present a non-parametric technique to infer the projected-massdistribution of a gravitational lens system with multiple strong-lensed images.The technique involves a dynamic grid in the lens plane on which the massdistribution of the lens is approximated by a sum of basis functions, one pergrid cell. We used the projected mass densities of Plummer spheres as basisfunctions. A genetic algorithm then determines the mass distribution of thelens by forcing images of a single source, projected back onto the sourceplane, to coincide as well as possible. Averaging several tens of solutionsremoves the random fluctuations that are introduced by the reproduction processof genomes in the genetic algorithm and highlights those features common to allsolutions. Given the positions of the images and the redshifts of the sourcesand the lens, we show that the mass of a gravitational lens can be retrievedwith an accuracy of a few percent and that, if the sources sufficiently coverthe caustics, the mass distribution of the gravitational lens can also bereliably retrieved. A major advantage of the algorithm is that it makes fulluse of the information contained in the radial images, unlike methods thatminimise the residuals of the lens equation, and is thus able to accuratelyreconstruct also the inner parts of the lens.
展开▼
