The Square Kilometre Array will be the largest sensor ever built by humankind and will acquire unprecedented amounts of data in radio astronomy. Recently, convex optimisation algorithms coupled with sparsity priors have been shown to outperform traditional radio interferometric imaging methods such as CLEAN. Thus, it is of paramount importance to extend these techniques to handle large-scale data sets efficiently. In this article, we propose two convex optimisation algorithms for solving the radio interferometric imaging problem. They are tailored for big-data and employ parallel and distributed computations to achieve scalability, in terms of memory and computational requirements. One of them also exploits randomisation, over data blocks at each iteration, offering further flexibility. We present simulation results showing the feasibility of the proposed methods as well as their advantages compared to state-of-the-art reconstruction algorithms. The code is available online.
↧