Swarztrauber, Paul N. SIAM oral history

Paul N. Swarztrauber grew up in Illinois and gained a BS in Engineering Physics from the University of Illinois in 1959. During his studies he used the ILLIAC computer and studied mathematical computing under Lloyd Fosdick. After graduation Swarztrauber spent three years in the US Air Force, and he discusses his work at Kirtland Air Force Base in New Mexico. In 1963, he went to work as a scientific programmer at the embryonic National Center for Atmospheric Research in Boulder, Colorado. Swarztrauber discusses the initial computing environment at NCAR and its development over the years. While at NCAR, he enrolled in the PhD program of the University of Colorado, Boulder, earning his PhD in 1970 under the direction of Robert Richtmyer. This led to a gradual shift of responsibilities at NCAR from application programming to mathematical research. He has published over sixty refereed papers in areas including partial differential equations, computational fluid dynamics, parallel computation and communication algorithms, computational aspects of weather and climate modeling, the fast Fourier transform, and, multi-computer design. He is also the author or coauthor of several large scientific software packages. Swarztrauber discusses his role in three significant packages, including for each their relationship to NCAR’s needs, related original mathematical research, their origins, collaborators, source of algorithms, documentation, distribution methods, relations with users, testing procedures and development through multiple revisions. These were FISHPACK (with Roland Sweet and John C. Adams) for the approximate solution of separable elliptic partial differential equations, FFTPACK for the fast Fourier transform of periodic and other symmetric sequences, and SPHEREPACK (with John C. Adams) for the modeling of geophysical processes using spherical coordinates. Swarztrauber also discusses his work on the CHAMMP project to evaluate scientific applications for the massively parallel machines during the late 1980s, and his resulting attempts to conceptualize and design a new architecture for multiprocessor units he calls the “communication machine.”