RADIO ASTRONOMY IMAGING

Moore's Law is also driving digitization much closer to the individual antenna receptors in the data aquisition stream from modern radio telescopes as it is more cost-effective and scientifically flexibile to do so than it has been in the past. This increases the associated computational challenges and software construction costs for modern radio interferometric arrays and moves them closer to large-scale sensor array architectures. Image formation in radio interferometry is a complex, multi-step numerical procedure, as it requires the simultaneous solution of an inverse imaging problem and the determination of an unknown instrumental model. These calibration and imaging algorithms need to be deployed on these new array architectures in the petascale regime.

The increasing complexity of modern radio astronomy instruments also drives a corresponding need for more advanced, automated data reduction capabilities. At present, the inverse imaging process requires specialist knowledge, and is most often a time-consuming, interactive process with many intermediate computational steps. The required specialist technical knowledge of image formation methods poses a steep learning curve for new users of radio-astronomical telescopes, and often limits scientific access to a small subset of the potential US astronomical user community. It is a particular impediment to smaller institutions that may not be able to maintain a local investment in the expertise required to assist new graduate students and other users in this field. Exponential increases in data rates are rapidly rendering this custom, interactive data reduction model obsolete. Current data from modern telescopes demand high-performance automated pipeline reduction, encoding the best community practices in data reduction, as well as distributed access to large, federated astronomical databases to allow archive science.  Modern advances in computational science and engineering and high-performance computing can enable breakthrough scientific advances in radio astronomy by addressing these challenges.

The RAI group is well-positioned at NCSA and UIUC to focus on these key interdisciplinary research questions. The group goals are to:

1. Enable the radio astronomy community to do breakthrough science using the high-end computing resources required to handle the data rates and computational demands of current and future radio interfereometers;
2. Develop new scalable imaging and calibration algorithms required for this new petscale regime;
3. Provide automated workflows to improve community access and increase scientific productivity; and
4. Contribute to the development, definition and deployment of a shared cyber-infrastructure (CI) for high-performance computing.

The National Science Foundation, the state of Illinois, the University of Illinois, industrial partners, and other federal agencies support NCSA.

Photo credits: CARMA site (R. Plambeck, UCB)