Paul Swarztrauber looks back on 41 years at NCAR
One afternoon in 1962, Paul Swarztrauber almost missed a job interview. He was looking for a relatively new atmospheric research center in Boulder that had an opening for a computer programmer, but he couldn't spot the building.
Tempted to give up, he instead drove around the block once more, and this time he spied a small sign pressed into the lawn. He had found the National Center for Atmospheric Research, housed in an unimposing two-story building at the corner of Marine and 30th streets.
The 26-year-old Paul did well in the interview and, as fate would have it, he never had to look for another job again. After a 41-year career at NCAR, the highly regarded computational mathematician retired this year as one of NCAR's most senior staffers.
"I can't believe how lucky I've been to work here," Paul says.
In a recent interview with Staff Notes Monthly, he reminisced about his tenure at NCAR and how the organization has changed. This article is part of a series on the early days of NCAR, which was founded just two years before Paul's interview.
Punch cards and pink monoliths
Paul's decision to come to NCAR was anything but predestined. Recently discharged from the Air Force, he had already been offered a job with the Rand Corporation in California. But a friend showed him an ad in the Denver Post for a computer programmer at NCAR, and he decided to apply because he had been advised to "always give yourself the opportunity to say no."
Glenn Lewis, the first head of computing at NCAR, interviewed him for hours--an experience Paul recalls as almost as rigorous as the oral test for his doctorate.
"He really grilled me," Paul says. "It was a fledging organization, but the plans sounded terrific, and my sense was I could both learn and contribute. If it didn't pan out, I could go
to Rand or another lab or university."
NCAR was just in its third year when Paul began work on January 9, 1963.
Staffers in NCAR's early days created punch cards to run computer programs.
His job consisted of creating punch cards for computer programs that enabled scientists to analyze their findings. A team of operators, working in the makeshift computer room, inserted the cards into readers to load the data into NCAR's computers. Although advanced for that day, the storage of an entire NCAR mainframe computer in the '60s "wouldn't take up much space on a DVD," Paul says.
When Paul began his job, NCAR staffers were housed in rented office space. But they were buzzing about the new NCAR laboratory that would be built on a mesa just west of Boulder. Paul can still recall his first visit to the mesa site, which he reached by hiking up with his wife, Suzanne, while carrying their new daughter, Karleen, in a pack. Although the Mesa Lab was not yet built, several pink concrete monoliths had been erected to make sure that the weathered color of the laboratory would blend in with the natural surroundings.
Once construction was finished, Paul commuted to the lab from his Boulder home by foot. The mathematician estimates he has made the climb about 8,000 times.
Contributions to mathematics
Thanks to NCAR, Paul took math classes at CU. He received his doctorate in applied mathematics in 1970 and embarked on a distinguished mathematical career. He has published about 65 articles on computational mathematics in refereed publications.
Several of Paul's most important contributions began as solutions to projects he worked on with NCAR scientists. Early in his career, he determined an efficient way of computing a solution to a common partial differential equation that describes potential flow and a variety of other phenomena in science and engineering.
After publishing a number of papers on this solution, he created a software package known as FISHPACK. (The name is a subtle tribute to Siméon Poisson, the French scholar and originator of the classic equation whose many variants are solved by FISHPACK. Poisson means "fish" in French.) The package, widely distributed as an NCAR Technical Note, is still in use and is currently being updated to Fortran 90.
Another of Paul's signature contributions is the Fast Fourier Transform (FFT), which is perhaps the best-known algorithm in computational mathematics because of its wide-ranging applications in science, medicine, and engineering. After publishing a number of papers on the algorithm, Paul developed FFTPACK, which efficiently computes the fast complex, real, sine, cosine, and quarterwave transforms. He then turned to another software package, SPHEREPACK, which computes quantities required to solve partial differential equations that are posed on the sphere.
These packages, which can dramatically reduce the amount of computing time required to solve research problems, have been used widely both within and outside NCAR. FFTPACK, for example, has been accessed hundreds of thousands of times from the Netlib repository, an international collection of mathematical software, papers, and databases.
"It's very gratifying to see people who would have had to spend a week coding up and solving equations using these software packages to accomplish their work in an afternoon," Paul says. "That's the beauty of applied mathematics."
In the early 1990s, Paul teamed up with Bob Ward at the Oak Ridge National Laboratory to implement weather and climate modeling on massively parallel computers. The project was known as CHAMMP (for Computer Hardware, Applied Mathematics, and Mathematical Physics), which enabled the core group of researchers to label themselves CHAMMPIONS (CHAMMP Interagency Organization for Numerical Simulation). The project eventually drew Department of Energy funding for university researchers and NCAR divisions.
Paul traveled to the Siberian Academy of Sciences in Novosibirsk for various collaborations. On one of those visits in the early 1980s, Paul's hosts began speaking urgently to each other in Russian while they were strolling with Paul through the Siberian woods to the laboratory. They then told Paul there had been "an incident" between the United States and the Soviet Union. A Korean civilian airplane was missing off the Soviet Union's eastern coast, apparently shot down by the USSR.
The incident aggravated Cold War tensions. Airports began to close. Fortunately, Paul received a late-night phone call from Walter McIntyre, NCAR's head of computing services, who had made alternative travel arrangements to bring Paul home.
Paul recalls another interesting moment involving Walt Roberts, the founding director of NCAR. In the early 1970s, Paul served for a time as acting manager of SCD, and he attended meetings with Walt and other NCAR managers. At one of those meetings, several of the managers spoke in favor of a somewhat controversial proposal. But Walt, speaking last, quashed it because he worried that it might be viewed as a conflict of interest.
As Paul remembers, Walt said, "This is NCAR. We don't do anything that even hints of impropriety."
Walt's stand made an impression
on all the managers, including Paul. "That has stayed with me all these years," he says.
As Paul sees it, the greatest challenge to all NCAR employees is "simply to be deserving of employment at this prestigious and magnificent laboratory." He adds, "This is a national center. People expect the best."
Looking back, Paul says NCAR has been a great experience for him. And he is thankful for the decision he made back in 1962 to keep trying to find the location of his job interview.
"That's how random life is sometimes," he says. "I had a good job offer and I could very well have driven back to Denver without the interview, but a neuron fired and I decided to try once again."
· David Hosansky
Also in this issue...
Freezing drizzle: An aviation hazard that's no longer hard to see
Hurricanes and climate change: Is there a connection?
Farewell to a well-loved tree
Digital Image Library now easily accessible
Staff Notes home page | News Center