Computer Muscle:
Changing the Way Science Gets Done

During an extensive safety-testing program for the new Boeing 777, researchers purposely crashed one multimillion-dollar jumbo jet after another. Surprisingly, destroying multiple copies of the prototype aircraft was not astronomically expensive. Why? Because scientists using extremely sophisticated computer models destroyed "virtual" rather than real jets.
Along with aeronautical engineers, researchers in widely divergent fields of science are using the staggering advances in computers and computer-related technology--known as high-performance computing or, informally, as supercomputing--to extend vastly the scope and complexity of their work. The field is known as computational science.
"For hundreds of years scientists have used one of two approaches--experimental or theoretical--to do their work," says associate professor of computer and information science Jan Cuny, a member of the University of Oregon's Computational Science Institute (CSI). "But now computation is emerging as a third fundamental approach. Computational science is an extremely powerful and rapidly evolving tool."
The tool is composed of high-speed network technology, large-volume data servers, high-performance graphics workstations and, at its core, powerful parallel supercomputers, explains Allen Malony, a CSI member and associate professor of computer and information science.
Since its inception in 1995, CSI has put together a formidable computer facility with funding from the UO, the National Science Foundation, the Murdock Charitable Trust, the Defense Advanced Research Projects Agency, and the U.S. Department of Energy. Early this year CSI took a major step forward when it opened a new Visualization Laboratory featuring ten high-end Silicon Graphics Inc. workstations suitable for the most complex scientific visualizations.
CSI computer scientists are also involved in larger collaborative computational science enterprises. In one with the Department of Energy, they are working in software support for large computations such as those that will monitor the status of the U.S. nuclear arsenal by exploding "virtual" bombs rather than real ones--at great savings to both the treasury and the environment.
Another aspect of the work of computer scientists at CSI is creating practical applications of computational science for CSI members in other fields of science. "These researchers are eager to apply computational science to their research," Malony says. "Our goal is to develop applications of this tool tailored to their specific needs in fields such as physics, biology, chemistry, linguistics, mathematics, and geology."

CSI member Doug Toomey, a UO seismologist, uses supercomputing muscle to create three-dimensional "CAT scans" of the earth (see photo), that reveal geological features miles below the ocean floor.
During a continuing series of expeditions to mid-ocean ridges, Toomey has gathered an enormous volume of research data. With the aid of CSI computer scientists, he has been able to sift efficiently through the data and extract useful information in ways previously impossible. Another immensely powerful application of computational science, he says, is combining the data from various research efforts, sometimes conducted years apart. This wrings more results from the resources invested in research.
"Whole new avenues of research have opened up for me--not just because of the hardware but also because of having access to a very sophisticated group of computer scientists," Toomey explains. "With the aid of these scientists, I'm now able to push my research forward more efficiently." For example, one aspect of his work that used to take days is now completed in an hour.
Toomey is certain that scientists will increasingly come to rely on computational science. His students, he says, will be a part of this trend as they mature into tomorrow's working scientists. "My students are gaining experience not only in geology but also in developing algorithms for parallel computers. This is a powerful combination of skills--one future employers will recognize and seek out."

For Jan Cuny and Allen Malony, researchers such as Doug Toomey present challenges that serve as test beds for developing new ways of applying supercomputing power to real-world problems.
"It's a wonderful collaboration," Cuny says. "Together we've created a powerful tool, advancing high-performance computing, accelerating research in seismic tomography, and providing our students the opportunity to learn about the application of sophisticated computational technology to science."

Back to INQUIRY, Fall 1997

©1997 University of Oregon