Across the disciplines
Program trains graduate students from many fields in emerging approach to research
By Steven Schultz
Princeton NJ -- It may not have seemed so unusual for the first day of classes -- a capacity crowd in the 50-seat computer science auditorium for a new course that was generating a lot of excitement.
What was not so ordinary was that it was a 500-level graduate course, the kind of numerals that often draw fewer than a dozen students. The popularity of the class, "Computational Methods and Their Applications Across Disciplines," illustrates what its organizers say is a deep need for a new kind of training among students in nearly every scientific discipline.
The course is part of an extensive series of new classes, workshops, invited lectures, lunchtime seminars and graduate fellowships designed to train students in an emerging type of research known as computational science. This new approach involves recreating anything from the early universe to the human immune system on computers. Such simulations allow researchers to study and manipulate complex systems in ways that would be impossible with conventional experiments or pencil and paper calculations.
Computational science also helps researchers sort through the staggering quantities of data that computers and experiments generate. Scientists who analyze genome data or invent better Internet engines increasingly rely on this emerging area of research.
"Computation has really changed science," said Jim Stone, professor of astrophysics, who uses computer simulations to study the birth of stars. "Now computation has become a subdiscipline of its own, and scientists of the future need to know what it is all about."
That is one of the goals behind Princeton's Program in Integrative Information, Computer and Application Sciences, dubbed "PICASso." The program brings together faculty members from computer science, applied mathematics and many other areas of science and engineering to train graduate students in techniques that would be hard to acquire within the confines of a single discipline. Researchers in diverse fields often face similar computational challenges and benefit from understanding the solutions others develop, said Jaswinder Pal Singh, professor of computer science and director of PICASso.
A particular focus of the PICASso program is to foster an interplay between researchers who engage in computational science and the computer scientists who develop methods and conceive new computer systems and software. Both the designers and the users can do better work if they understand each other's increasingly complex needs and constraints, Singh said. An area of emphasis is the concept of "scalable computing," which means building and programming a computer system so that it can be expanded indefinitely simply by adding more machines.
"This is a great new set of activities that are really coming together at Princeton," said Singh. "And it has been wonderful to see how many people from so many disciplines are attracted to this effort."
Among the initiatives PICASso has undertaken are:
Four new graduate courses in the departments of computer science, chemical engineering, mechanical and aerospace engineering and the Program in Applied and Computational Mathematics.
A series of three-day mini-courses focused on teaching practical methods. Computer science professor Brian Kernighan will lead one in March on good programming practices. Future courses will address visualization of scientific data and techniques for parallel programming (using multiple computers to tackle the same problem).
A lunchtime seminar series addressing general topics in computational science. The seminar takes place every Wednesday at 12:30 p.m. in 402 Computer Science. The seminars usually will be led by graduate students. Periodically, they will feature invited "Successes of Computational Science" talks by established researchers, who will talk about major advances in their disciplines.
A second seminar series, every Monday at 12:30 p.m. in 302 Computer Science, will focus on computation and data analysis in biology. The inaugural seminar on Feb. 16 will feature a "Successes in Computational Science" talk by David Botstein, director of the Lewis-Sigler Institute for Integrative Genomics. The seminar will be in a special location, the Friend Center convocation room, for Botstein's talk.
Interest is strong
Initial interest in these events has been very strong, said Singh. The opening lecture of the computational methods course on Feb. 2 was particularly well attended with 75 students because it featured a guest talk by astrophysicist David Spergel on how computational techniques provided the foundation for cosmological discoveries that were heralded as the "breakthrough of the year" by Science magazine. At the second meeting, 45 students signed up to attend for the full semester.
Interest among faculty members also has been high, said Singh. The computational methods course will be co-taught in one- or two-week rotations by eight faculty members from five departments. "They are all teaching this course without getting teaching credit in their own departments," said Singh. "But it was easy to get them to do it because they all recognize it is a great thing to do and many students will benefit."
Faculty members also are benefiting by having closer contact with each other's work. Maria Pino Martin, assistant professor of mechanical and aerospace engineering, said she was inspired to investigate new methods for analyzing data after hearing Spergel's introductory lecture. Her work involves simulating turbulent gas flow at hypersonic speeds; her findings may help NASA plan the entry of a spacecraft into the atmosphere of Neptune.
The PICASso projects have accelerated this year with the participation of faculty members and the hiring of researcher Steven Kleinstein as program coordinator. Kleinstein, who received a Ph.D. in computer science from Princeton in 2002, uses computer simulations to investigate the workings of the human immune system. His experience gives him a unique perspective on helping develop and run new programs. "It keeps me current and understanding what the challenges are," Kleinstein said. "Many of the ideas I come up with for the program are things that I would like to have for myself."
The PICASso program is funded by an interdisciplinary graduate research grant from the National Science Foundation. More information on the program and its initiatives are available at <www.cs.princeton.edu/picasso/>.