HAIRI: HPC and AI for Regularisation in Radio-Interferometric Imaging

  • Jackson, Adrian (Principal Investigator)
  • Wiaux, Yves (Principal Investigator)

Project Details


Approximately 3 million core hours on the Cirrus Tier 2 HPC system.

The proposed HAIRI (HPC and AI for Regularisation in Radio-Interferometric Imaging) project will contribute to the design transformative methodology to endow advanced imaging instruments with unprecedented visual acuity, impacting imaging science from astronomy to medicine. With a focus on computational imaging, and with AI and Data Science at the heart, the project will develop a new generation of algorithms at the interface of optimisation theory and deep learning to transform data into images and to provide a new regime of precision (i.e. resolution and dynamic range) in comparison to the state of the art, as well as scalability to extreme data volumes. The algorithms, dubbed AIRI (AI for Regularisation in Radio-Interferometric Imaging) will be specifically designed for synthesis imaging by radio interferometry (RI) in astronomy and validated up to 1TB image size on High Performance
Computing (HPC) systems, simply enabling the transformational science planned for the next decades in the field.

A technology transfer at more moderate image size in magnetic resonance (MR) imaging in medicine will be performed as proof of their wider applicability. Algorithmic developments will be prototyped in Python, and a highly scalable parallel C++ software implementation fully exploiting the
algorithmic functionalities will be mapped onto the latest and most efficient HPC hardware technologies for optimised computation and communication performance.
Effective start/end date1/04/2231/03/23


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.