seal  Purdue News
____

From the summer 2003 edition of Purdue Engineering Extrapolations

Interdisciplinary Education

Computationally speaking

Chuck Wright, a master’s candidate in the School of Aeronautics and Astronautics, says he’s always been interested in things that move. "Mostly I like things that move fast–race cars, military equipment, airplanes," he says.

Chuck Wright
Download photo

To pursue his major interest, computational aerodynamics, he uses computers to calculate flow fields through jet engines and nozzles and around aircraft wings, plotting where the air goes, what the pressures are, and what the properties are. "But," he says, "if you want to solve for the full turbulent flow around an entire aircraft, with all the relevant length and time scales, there’s probably not enough power to do it with all the computers in the world and all the time between now and when the world ends."

Getting the most out of what computers do have to offer him is why Wright enrolled in Purdue’s Computational Science and Engineering (CS&E) program, a graduate program offering the opportunity to study a specific science or engineering discipline along with computing in a multidisciplinary environment. In place since 1995, the program aims to teach a student how to integrate computing with another scientific or engineering discipline and make original contributions in both. With more than 80 enrolled, CS&E draws students from aeronautics and astronautics, electrical and computer engineering, mechanical engineering, and nuclear engineering, as well as a number of non-engineering disciplines, including psychological sciences, earth and atmospheric sciences, and chemistry.

"CS&E has opened my eyes to things I wouldn’t ever have thought about," says Wright. "I’ve studied methods used for numerical analysis–how the methods relate to the computer systems themselves. I’ve also learned that when a code blows up, it may not be because the physical equations are wrong but because the numerics are. I have had experience dividing by 0, when the denominator is finite! I’m careful now about how to formulate my equations so that doesn’t happen. I’ve also learned to look out for potential pitfalls such as subtracting two numbers of near equal value. That can produce an answer that’s smaller than the numerical error of the problem. The result is equivalent to a random number!"

Wright plans to enroll for doctoral studies at Purdue after finishing his master’s, so he faces a future of increasingly complex computations. Thanks to CS&E, he’ll be well equipped to take them on.

Writer: Lisa Hunt Tally

Photo: Photographic Services