Feature - Let scientists focus on the science

David Abramson (at right) and colleagues show how Nimrod can be applied to scientific problems. Image courtesy of Monash University.

David Abramson, Professor of Computer Science at Monash University in Australia, and Director of the Monash eScience and Grid Engineering (MeSSaGE) Lab, contributed this article.

When scientists find the complexity of dealing with high performance computers too high, they miss out on the advantages it has to offer. An understandable view — their expertise is science, not coercing a computer into doing what they want.  Over the years, our group has helped many scientists and engineers, in disciplines ranging from quantum chemistry to public health policy, to embrace new computing technologies that will advance their research.

This started about 15 years ago. We suggested to a group of physicist colleagues that they run their code in parallel on a cluster. Their problem separated in a straightforward manner into independent, parallel tracks, and appeared a perfect match for software tools (called Nimrod) that we were developing. They declined, saying they didn't have the time and expertise to convert their codes.  We offered to adapt their code, and in a very short time it was able to deliver new results with profound conclusions for their experimental science — a dramatic result.

This pattern has continued. Recently, for example, we helped climate science colleagues answer a complex question for which an actual experiment is effectively impossible: “If you burn the savannah in northern Australia, can this affect the weather?” Using Nimrod, we helped them simulate the weather while varying four key parameters — the fire intensity, the area burned, the timing and the regrowth period. The combinations generated 90 independent simulations that ran on distributed supercomputers continuously for six months.  The experiment shipped some 1.6 TB of data across national and international networks for analysis by collaborators. The team was able to conclude that under the right circumstances the onset of the monsoon can be varied.

UCSD bioengineering undergraduate student Randy Lee explains how he and fellow students have used Nimrod to support cardiac modelling work. Image courtesy of Monash University.

“Nimrod was a great boon to us, because it allowed us to focus on our scientific questions without worrying about clusters, networks and international partnerships”, said Monash Professor Amanda Lynch.

Unlike many other parallel computing environments Nimrod focuses on the science — how to help researchers express their problem in the simplest possible way. While it only solves a narrow range of parallel programming problems (so called "embarrassingly" or "pleasingly" parallel), it does this using a very simple declarative language, supported by a Web interface. Users can easily explore what happens when input parameters change and can access a range of search methods from complete enumeration to algorithms that provide optimization functions.

The powerful, automatic optimization features of Nimrod’s search capability help scientists “tune” their models to better match reality, and save time and compute power by avoiding fruitless parameter combinations. For example, Professor Andrew McCulloch’s bioengineering lab at the University of California, San Diego used Nimrod to run complex cardiac simulations, and to tune the model parameters to minimize the difference between simulated output and experimental data.   

What’s more, McCulloch’s undergraduate students have learned to use Nimrod. Under the Pacific Rim Undergraduate Experiences program (PRIME), some of his students travelled to Australia to perform collaborative research with the Monash-based MeSsAGE Lab.

“These undergraduate students have been able to master Nimrod and get it to perform new science in only a few short weeks. The results have been outstanding,” said UCSD researcher Anushka Michailova.

—David Abramson, Monash University