Radio-Astronomical Imaging with WSClean and Image-Domain Gridding

Modern radio observatories provide astronomers with large data sets. To turn these data sets into science, astronomers require imaging techniques that provide state-of-the-art image quality (e.g. resolution and dynamic range) at high performance (e.g. modest runtime). We integrated an efficient GPU-accelerated implementation of the Image-Domain Gridding (IDG) algorithm into the widely used WSClean imager to fulfill these needs. WSClean features several novel deconvolution techniques, such as auto-masked multi-scale, and parallel cleaning, while IDG is capable of correcting full-polarization direction-dependent effects (DDEs) during imaging. In this paper we give an overview of the current state of our work: we introduce IDG, we present performance of IDG on contemporary accelerator hardware, we show the impact of DDE correction on image quality and we discuss its suitability for the demanding EoR science case as an example. All in all, WSClean+IDG is an excellent imager for current and future radio-telescopes.