Gravitational Lens Inversion Using the Maximum Entropy Method

Sylvanie Wallington
Harvard-Smithsonian Center for Astrophysics


I present an algorithm for inverting extended gravitational lens images using the Maximum Entropy Method (MEM). Gravitational lensing is the phenomenon whereby multiple, distorted images of a cosmologically distant object are seen due to the lensing effect produced by an intervening mass. In attempting to recover the original source from a gravitationally lensed image, it is the knowledge that a single source produces the multiple images that allows solutions for the unknown source intensity distribution and the unknown lensing potential to be disentangled. The success of MEM in the common astronomical problem of reconstructing noisy and incomplete data produced by atmospheric distortion, detector noise, and diffraction, motivated me to apply MEM techniques to the inversion of gravitational lenses. I find that this method accurately inverts typical ring-type gravitational lens images and is effective at solving for unknown lens parameters. In test cases where the source object is larger than the lens critical radius, however, there are systematic errors in the reconstruction in the form of discontinuities or ``glitches'' at the boundaries separating different image multiplicities. I show that these errors are intrinsic to MEM when applied to the lensing problem. The glitches can be removed by filtering the source to remove unphysical high frequency power. The sources for which these glitches occur have never been observed, so I conclude that MEM is an effective method for lens inversion.
MaxEnt 94 Abstracts /