John G. Lewis, 1945-2019. Photo courtesy of Timothy O’Leary.
John G. Lewis, a SIAM Fellow and longtime member, died peacefully in his sleep on June 10 after a three-year battle with amyotrophic lateral sclerosis (ALS). He was 74 years old.
After his undergraduate studies and a short teaching stint at St. Olaf College in Minnesota, John enrolled in the Computer Science Department at Stanford University as a graduate student. In 1976, he completed his Ph.D. in numerical analysis with a thesis on the Lanczos algorithm under the direction of Gene Golub. Upon earning his degree, John joined the applied mathematics staff of The Boeing Company, where his influence shaped the development of mathematical algorithms and software. His ideas live on in the very fabric of the way Boeing solves its largest and most difficult analysis problems, and are especially evident in the company’s approach to large, sparse linear and nonlinear systems of equations.
I first met John when I joined the Boeing Math Group as a freshly-minted Ph.D. in the summer of 1986. At the time, he was a thriving, talented, and well-respected mid-career professional who immediately awed me. Having worked there for many years, John had managed to assemble a team of world-class researchers in computational linear algebra, including Phuong Vu, Horst Simon, Cleve Ashcraft, Roger Grimes, Barry Peyton, and Dan Pierce. Boeing was preparing to develop the 777, and the team—under John’s direction—was gearing up to implement the structural analysis and computational fluid dynamics necessary to do so. At the time, this work embodied some of the largest structural analysis problems ever solved. The group received the 1988 Gordon Bell Prize for achieving the then-unheard of performance of one gigaflop on a statics structures problem involving over a million variables. They managed to reduce the runtime from more than 15 minutes to less than 30 seconds.
In 1989, John was selected for the first cohort of Boeing’s Technical Fellowship program. He embodied the very ideals of the program: world-class technical knowledge and judgment, an uncanny ability to bring those talents to bear on Boeing’s toughest challenges, a willingness and capacity to mentor and teach others, and visionary technical leadership. Many scientists who successfully rose through the technical ranks at Boeing in subsequent years did so by following John’s example.
Although I was not a member of John’s amazing linear algebra team, we talked frequently and I learned a great deal from him. At one point early in my career, I was providing support to a team of airplane configurators (designers) attempting to develop an airplane wing planform design tool. The wing planform is essentially a two-dimensional top view of an airplane wing. Simple planforms are specified in terms of approximately 20 design variables, which unfortunately are not independent of one another and must satisfy 10 or so nonlinear relationships. The configurators wished to create an interactive tool enabling specification of any 10 design variables, with the remaining 10 to be interactively solved for in real time. I had no idea how to go about this, so I asked John for help. After a day or so, he came back with the Dulmage-Mendelsohn decomposition and a clear idea of its application to the problem. It worked perfectly and has been a cornerstone in Boeing airplane design ever since.
John always treated the problems and struggles of his colleagues and acquaintances like they were his own, and constantly made time for me and other junior staff members. Among the lessons I learned from John is the importance of continuing to think and act like a mathematician, even when surrounded by engineers with different backgrounds and approaches to problem solving. The point is not so much to maintain independence as it is to provide a perspective that could bring greater value to the company’s efforts. It was an early lesson for me in the significant benefit of diversity; careful, independent analysis; and the great technical progress that arises from the co-mingling and cultivation of other ideas with one’s own.
John was a member of SIAM throughout his entire professional life. He believed in the society’s merit, especially the advantages of membership to the careers of industrial mathematicians with the wherewithal to leverage applied mathematical research to solve the world’s toughest applied and industrial challenges.
John will be greatly missed by all who had the privilege of knowing him, learning from him, and experiencing his passion for mathematics, computing, and life. He will be missed even more by his wife Fran, his sons Steven and David, and his two grandchildren.