Constrained imaging for radio astronomy

We show that the imaging problem for radio astronomy is not only bounded below (the image is non-negative), but there is also an upper bound. We show that the tightest upper bound is the MVDR dirty image. We propose using active-set methods to solve the imaging problem and show that this algorithm is strongly related to sequential source removal techniques like the CLEAN. Also recent studies show the relation between non-negativity and sparsity which means that the proposed optimization also benefits from sparsity in an automatic way.