Pawsey staff is returning from a busy week of activities with researchers from across the nation. The Centre’s strong presence at this year Collaborative Conference on Computational and Data Intensive Science (C3DIS) conference included an exhibition booth, workshops and special activities for currents and potential users. The event took place at the Melbourne Convention Centre, and it is now organised by Australasia research Organisation (AeRO).
Overall, the conference theme focused on data, specifically – collection, curation and the federation of data. The week’s conference concluded this topic’s importance is in making information and data available across different organisations, thus breaking down silos across teams to increase efficiencies, knowledge and decrease collection times for the same information.
The week of events began with pre-conference workshops. During the day Pawsey run a joint session with Australia’s other Tier-1 High-Performance Computing facility, National Computational Infrastructure (NCI), on Introduction to High-Performance Computing (HPC).
The Introduction to HPC workshop included an overview of Pawsey’s use of machine learning and containerisation technology to make use of HPC and parallel systems even more accessible for researchers across a range of technologies. Container technology encourage a more straightforward workflow between systems. Machine learning and containers played centre stage during this year conference.
The C3DIS conference itself begun with a keynote address by Dr Eng Lim Goh, the Vice President and Chief Technology Officer of HPC and Artificial Intelligence (AI) from HPE, Singapore. Prediction: use science or history? was Dr Goh keynote presentation.
Majority of interactions between conference attendees and Pawsey included commendations of the fascinating talk given by Dr Goh. Staff at Pawsey have had the opportunity to personally meet Dr Goh again at the Centre in Perth on Friday, where he presented similar findings on his work.
HPC: an indispensable complementary computational laboratory for Australia’s researchers
The week was not complete without a showcase of some of Pawsey’s respected researchers in Melbourne. Monash University’s Prof Julio Soria and Dr Simon Campbell highlighted their research and how Pawsey supercomputers and services enable their scientific outcomes.
Dr Simon Campbell shared his work on 3D fluid dynamic models of the interior of stars using the Magnus supercomputer. Because of the complexity of and massive amounts of data produced by stars, Dr Campbell predominately only visualise stars in 1D stellar models rather than 3D. 1D models can be very successful in reproducing many features of stars, their parameters and calibration.
Though this gives valuable insights into the makeup, life and death of stars, a 3D simulation can show so much more. 3D simulations offer stellar interior convection zones, with a view to understand the fundamental physics that is needed to improve 1D models. Through the example of two different stars, Dr Campbell discussed how he used Magnus and Zeus to visualise and understand stars at the Pawsey Supercomputing Centre.
Professor Soria explained how he discovers the mysteries and dynamical complexities of fluid turbulence using direct numerical simulations with HPC. As humanity progress into the 21st century and deplete non-renewable sources of energy such as crude oil, increase climate change and ecosystems, new methods of technology must be developed to sustain economies and populations.
Scientists such as Professor Soria are dedicating their life’s work to solve this problem.
The continuous advancement in the processing power of HPC with large memory allows the high-fidelity direct numerical simulation of turbulent flows, which isn’t possible to replicate in a laboratory. These simulations are the building blocks of turbulence and its dynamics.
The use of HPC is becoming an indispensable complementary computational laboratory, which integrates the experimental apparatus, the instrumentation and experimental measurement methodology into one. This provides time-resolved, spatially fully resolved, three-dimensional measurements of all fluid dynamic variables allowing researchers to probe the mysteries and complexities of fluid turbulence with unprecedent detail.