Remembering Joe Oliger

March 24, 2006

Joseph Oliger, 1941-2005

Joseph Oliger, a professor of computer science at Stanford University from 1974 to 2001, and director of the Research Institute for Advanced Computer Science (RIACS), NASA Ames Research Center, from 1991 to 1998, died on August 28 in Truckee, California, at the age of 64. He died of cancer, which had been diagnosed in 2004. He is survived by his sons, Nicholas and Jason, and a granddaughter, Zoe.

Joe was well known for his early research in numerical methods for partial differential equations and is remembered with special fondness by many former students in the computer science and mathematics departments and the Scientific Computing and Computational Mathematics Program at Stanford, as well as by many colleagues around the world. He was a member of the SIAM Council (1982�87) and Board of Trustees (1993�95) and, from 1976 to 1984, an editor of SIAM Journal on Numerical Analysis.

Joe was born in Indiana in 1941 and grew up on a farm. He received BS and MS degrees from the University of Colorado, Boulder. In the late sixties, he was working as a mathematician at the National Center for Atmospheric Research in Boulder, probably not aiming for an academic career. Then, in 1970, Heinz-Otto Kreiss met Joe at NCAR and quickly became aware of his talent, and the two began to work together.

A topic of quite intensive discussion at that time was whether higher-order methods should be used for large-scale problems like weather prediction, and Kreiss and Oliger set out to settle this question. They used Fourier analysis for finite-difference approximations of a simple hyperbolic model problem describing wave propagation, and estimated the number of points per wavelength that would be needed to achieve a certain accuracy. The result was quite clear: Except in the case of very low required accuracy and short integration times, use of fourth-order (and sometimes even higher) methods in space is certainly preferable to use of second-order methods. The important point is that the analysis provided simple tools for making the best choice of order given a desired accuracy. Because of Joe's affiliation with NCAR, the paper was published in Tellus (1972); it went on to become one of the most quoted papers in numerical PDEs.

At the same time, the development of Fourier methods was beginning. Here too, NCAR played a central role, as Steve Orszag, who had shown an interest in Fourier methods, was visiting for a year. Kreiss and Oliger included Fourier methods in their analysis, and found them to be very effective for periodic problems on regular grids. But they also showed that with variable coefficients, it is necessary either to write the equation in self-adjoint form or to include dissipative terms.

In 1973, Joe received his PhD from Uppsala University, Sweden, after spending a year with the department of meteorology in Stockholm. The same year saw the appearance of the famous GARP report, Methods for the Approximate Solution of Time Dependent Problems, by Kreiss and Oliger. Although a thin soft-cover report, it contained most of the latest theory for difference and Fourier methods, and was often quoted in the years that followed. Indeed, it sold out very quickly and is not easy to find today.

The success of the GARP report made clear the need for a new book on PDEs and their numerical solution, and Joe started working with Kreiss on this project. Later, I was invited to join as a co-author. We wrote several drafts over the years, but new material kept coming in. After many rewrites, the book, Time Dependent Problems and Difference Methods, more than 600 pages long, was finally published in 1995.
Joe's academic career at Stanford was particularly remarkable in that he advised so many students who went on to very successful careers of their own. Here is the full list of Joe's students (along with their years of graduation and current affiliations):

John Strikwerda, 1976 (Wisconsin); Ken Bube, 1978 (Washington); Tony Chan, 1978 (UCLA); Bill Coughran, 1980 (Google); Bob Higdon, 1981 (Oregon State); Bill Gropp, 1982 (Argonne); Marsha Berger, 1982 (NYU); Randy LeVeque, 1982 (Washington); Nick Trefethen, 1982 (Oxford); John Bolstad, 1982 (Livermore); Steven Caruso, 1986 (consultant); Chris Fraley, 1987 (Washington and Insightful Corporation); Wei-Pai Tang, 1987 (Boeing); Bill Skamarock, 1988 (NCAR); Pat Worley, 1988 (Oak Ridge); Amala Mahadevan, 1995 (Boston University); Patrick Wit-ting, 1995 (Dreamworks); Xiaolei Zhu, 1996 (Tykhe LLC); Margot Gerritsen, 1997 (Stanford); James Lambers, jointly advised by Gene Golub, 2003 (Irvine).

Many of these former students have gone on to have PhD students of their own, which puts the current number of Joe's academic descendants at well over a hundred. Even more impressive than the quantity and quality of Joe's students is the uniform way in which they remember him: as a caring, kind person who was never too busy to help them as much as he could, never mind how long the line of other students waiting outside his door.

In the 1990s, with a leave (half time) from Stanford, Joe served as director of RIACS. RIACS researchers remember him as someone who was very supportive and with whom they could work easily. Because of his stature in the computational PDE community, he had the confidence of NASA Ames management, which was very important to RIACS.

Joe retired from Stanford in 2001, when he moved to Truckee, in his beloved Sierra Nevada, where he could focus his energies on the outdoor life that he loved so much. He was incredibly strong, both physically and mentally. In his early years he was a sprinter at the top level nationally. He was also a rock climber, an activity for which his mental strength was extremely useful. He was well known in the climbing community for pioneering ascents, with his friend Steve Roper, in Yosemite in 1961. Later in life, he became a long-distance runner. He loved to cross country ski in the mountains; if darkness fell when he was far away from home, he would dig a hole in the snow and, sleeping soundly, spend the night.

Joe was very tolerant, a man of great integrity, always ready to help others and never known to describe another person in negative terms. He is warmly remembered by all who knew him. Margot Gerritsen has set up a memorial blog for Joe at Readers are invited to visit the site and read the many fond memories and stories posted by his friends and former students, and to contribute their own.

Joe was a remarkable man. We will miss him greatly.---Bertil Gustafsson, professor emeritus, Uppsala University, and visiting professor, Stanford University.

Donate · Contact Us · Site Map · Join SIAM · My Account
Facebook Twitter Youtube linkedin google+