Researchers at UCL have won grants totalling around £2m from the Engineering and Physical Sciences Research Council (EPSRC) to advance the capacity and reliability of exascale computing, which allows scientists to perform massive simulations and data analyses.
Simulations and data analyses will soon take place on an unprecedented scale as supercomputers enter what is being called the ’exascale’, where machines are capable of 10¹8 floating point operations per second, comparable with the processing power of the human brain.
One of the UCL projects, referred to as SEAVEA, which stands for Software Environment for Actionable & VVUQ-evaluated Exascale Applications, will provide a software environment to allow researchers to make these simulations reliable and their outcomes actionable. It will ensure they are verified, validated, and their uncertainty quantified.
The results could have a significant impact in a wide range of research fields. SEAVEA principal investigator Professor Peter Coveney (UCL Chemistry) said: "Such exascale simulation will model nuclear fusion, weather, climate, fluid turbulence, drug selection and surgical interventions, materials and molecular modelling, systems engineering and beyond. By making these simulations reliable and actionable, we can trust them, and therefore put them to practical use on workable, and very short, timescales."
A second project, LEXCI, which stands for Learned Exascale Computation Imaging, will provide a new paradigm of computational imaging to unlock the potential of deep learning and modern supercomputers for inverse imaging, where the goal is to recover images from raw, and often huge, data.
Computational imaging is one of the most widely encountered class of problem across science and industry, with applications ranging from astrophysics and geophysics, to medical imaging and computer graphics, and beyond.
Principal investigator Professor Jason McEwen (UCL Space and Climate Physics) said: " We are delighted that the LEXCI project has been funded. It will allow us to not only develop a new paradigm of computational imaging that will have widespread application in many fields, helping to unlock the secrets of the Universe and to image the neuronal pathways in the human brain, but also to develop numerous new computing paradigms to help to usher in the era of exascale computing."
UCL Vice-Provost for Research Professor David Price said: "UCL has a long and proud history of disruptive discoveries. These projects represent cutting edge research at the heart of machine learning as computing enters the exascale age, with all the potential advancements and challenges that will undoubtedly bring both to science and humanity."