International computing experts augment NeSI team

New Zealand’s researchers will benefit from the addition of nine global computational science experts to the team supporting NeSI, the nation’s high performance computing platform, says its director.

“NeSI is the most human-faced supercomputing centre in the world,” commented a researcher at the AMN-6 (Sixth Advanced Materials and Nanotechnology) Conference in Auckland recently. NeSI is the country’s national high performance computing (HPC) platform and has supported many successful research projects since its commissioning in early 2012. Such projects not only demand the swift solving of multiple, complex computational problems, they also need a scalable resource available to multiple researchers simultaneously. In spite of this digital emphasis, researchers also benefit from a level of human interaction and understanding not generally available from the large, international HPC providers; hence the comment from the AMN-6 researcher.

Now, the addition of nine internationally renowned computational science experts to the NeSI team is allowing it to offer science researchers even more opportunities to improve their productivity. NeSI director Nick Jones says their combined expertise in the application of computational methods to science results in the largest ever group of computing specialists to support New Zealand research. “With this team in place, as a sector we can now approach larger challenges in our economy, environment and society, such as those defined in the National Science Challenges,” says Jones.

Global experience

NeSI’s enhanced computational science team is led by Mark Cheeseman, who brings experience to the New Zealand research sector from a background in international HPC centres in Switzerland, Saudi Arabia, Canada, the US and the UK. Cheeseman is joined by team members with scientific programming experience, knowledge of numerical and statistical methods, optimisation techniques for scaling computations and a working knowledge of specialised HPC computing systems.

NeSI’s HPC services now include this more focused computational science capability, which international scientists say is invaluable to researchers. Professor Peter Hunter, a globally recognised leader in computational physiology, anticipates significant benefits to the NZ science and engineering research and development community as a result. “I’m absolutely delighted to see this initiative from NeSI on computational science – which is a critical discipline for an innovation-led economy based on fundamental scientific research,” says Hunter. “For example, our ability to predict the properties of the complex, composite materials used in the manufacture of many new products depends very much on multi-scale modelling and high performance computing.”

One example provided by Hunter is tissue engineering for implanted medical devices, where computational design is used to generate 3D printed scaffolds into which cells can be seeded. The creation of replacement tissues in bone and skin benefits from the so-called multiscale modelling of mechanical properties. “The computational science expertise made available to the NZ science and engineering R&D community through this initiative will be invaluable,” Hunter predicts.

Solving complexity

But as software to support complex computing becomes more commonly available, other challenges remain. Professor Ian Foster is an expatriate Kiwi and a top international computer scientist, currently Director of the Computation Institute at the University of Chicago and Senior Scientist at Agronne National Labs. In a Science magazine article in April 2012, ‘Shining Light into Black Boxes’, Foster and his co-authors note that, as every domain of science becomes more reliant on software, the publishing of research data also becomes more important.

Foster welcomes the addition of computational science skills to the NeSI team and highlights the significance of software in scientific exchange. “As science becomes increasingly computational, software emerges as both essential tool and new medium of scientific exchange,” says Foster. “The new indigenous capability provided by this team will both give New Zealand researchers access to the software they need to succeed and spur the development of a culture of free exchange of open computer software.”

Meanwhile, the ability of Cheeseman and the NeSI team to analyse and to optimise the performance of specific research tasks contributes to their understanding of the specialised computing algorithms and other methods used by researchers and scientists.

“Programming to scale across large numbers of CPUs is hard, yet is quickly becoming required across many fields of research.” says Cheeseman. “Programming with distributed memory is harder. Understanding the challenges within each separate research community requires years of exposure and collaboration. Why would we bother doing this? Because solving this complexity enables us to tackle problems with increased accuracy and significance.”

Currently the NeSI computational science team is reviewing all the projects that have been run successfully on its platforms. The team has already identified several strategic research community needs, problems and related software codes for early collaboration – a proactive approach to identifying the research sector’s capability needs that’s a first for New Zealand. Foster at the University of Chicago predicts the establishment of the computational science team will accelerate the impact of the NeSI platform on the research community. “Professionally optimised scientific software will typically run faster, be available sooner, and be more capable than un-optimised software,” he says.