HARRISONBURG - String together 16 personal computers that each run at 450 megahertz and hold 256 megabytes of memory, and what do you get? The answer is a supercomputer - the equivalent of a large machine that cost millions only a decade or so ago.

At James Madison University, through a grant from the National Science Foundation, the mathematics and physics departments have built a supercomputer with 16 PCs, and an add-on 17-gigabyte disk of memory, stacked on a bookshelf.

Other universities have built similar parallel computer systems for graduate student and faculty research. But, at JMU, the powerful networked computer is aimed at teaching - and retaining - undergraduate students in mathematics and physics.

"A single PC now is about the equivalent of the top-of-the-line supercomputer in 1976," said mathematics Associate Professor C. David Pruett.

"And those supercomputers cost $20 million or $30 million, and we bought this for $50,000," added mathematics Associate Professor James Sochacki.

Sochacki and Pruett are co-directors along with physics Associate Professor Dorn Peterson and Professor William Ingham of an NSF-funded project to develop a national model to train undergraduates for the increasingly-in-demand field of computational science.

"We hope to have a curriculum constituted in three or four courses that introduces undergraduate students to computational science - the use of computers to model the physical world," said Peterson. "Prior to this, that's primarily been a graduate school's job, and we think we can do a good introduction at the undergraduate level."

The professors said that when they started developing the program they knew of no such undergraduate programs except in Sweden.

The $163,105 NSF grant, matched by JMU, is also funding a fluid dynamics laboratory and a computer visualization lab for the collaborative program among the departments of mathematics and physics at JMU and North Carolina Central University in Durham.

At the end of the two-year project, they'll offer to other universities course materials, software and laboratory experiments in computational science for undergraduates.

Because computer processors are now so fast, other universities will be able to afford to build supercomputers, which allow researchers to work on numerous calculations at one time.

Today, it's speed that counts. "In the old days, a supercomputer was a computer that was a lot faster than the computer on your desk," Sochacki said. "Nowadays, a supercomputer is more of an environment."

The program's physics component involves fluid dynamics, the study of the behavior of liquids and gases. A fluid dynamics lab is currently being built in JMU's physics department.

The experiments that are being developed for use by the students in the fluids laboratory will be designed with the idea of providing real examples of systems that will then be modeled on computers by the students in their courses. The experiments are also being designed so that they can be inexpensively reproduced at other institutions.

"Fluid mechanics is something that everybody has some familiarity with," said Pruett, "whether you're pouring cream into a cup of coffee and watching what happens, or watching waves on the shore. It's a very difficult area of physics and mathematics, but it's one in which the average person has some intuition about what will happen, and we try to develop that in the lab component."

While fluid dynamics is an area in which JMU's math and physics faculties have expertise, for the third component of the program - scientific visualization - JMU has partnered with North Carolina Central University for their expertise in computer visualization to be used to understand the results generated by the computer models.

Leading the NCCU component is Dr. Alade Tokuta, chairman of the mathematics and computer science department at the Durham, N.C., institution who previously served on JMU's faculty. The two schools will exchange information and students as they develop the materials for the national educational model project.

A driving force behind the project is to attract and to keep talented students in the hard sciences and mathematics by giving them challenging and interesting hands-on research experiences often reserved for graduate students.

"It's really tough to be a science major," said physicist Peterson, "During your first two years of courses, a lot of the time it's just really hard work and you're not really getting into any of the interesting stuff yet. We're hoping that the freshmen and sophomores will see the interesting things that the juniors and seniors are working on and realize early that all of the work is worth it."

Further, they're hoping to attract good students with testimonials from successful graduates with high-paying, high-tech jobs in computational science, which uses computers to study scientific and engineering problems that may be impractical, dangerous or expensive to test using traditional techniques.

Many fields - such as meteorology and weather forecasting, oceanography, geology, oil prospecting, air traffic control and aircraft design, medical imaging, environmental science, and surprisingly, finance - are using computational science.

"The new Boeing 777 was not designed in an air tunnel," said Peterson, by way of example. "It was done totally on a computer."

Industry, including the high-tech businesses in northern Virginia, is looking for math and science graduates who can work the high-speed computers.

"They are looking for skills," Pruett said. "We're providing a special mixture of skills that's rare."

Eventually the computational science program may be tailored to reach students in other disciplines such as biology, chemistry, computer science, geology, finance and media arts.

For more information, contact Dr. James Sochacki, project director, at 540-568-6614.