The death of Mike Kascic is much more than the loss of a subject area editor for this journal. It is a real loss to all of us involved in the field of supercomputing.
Mike's early education was classical. His talent in mathematics was recognized at an early age, but his interests never focused exclusively on mathematics. He had a serious interest in linguistics as well as continuing interests in physics, astronomy, music, and theology. These interests had a substantial effect on his later work.
After getting his B.S. in mathematics and physics from St. Joseph's University, he went on to get his M.S. in mathematics from the Courant Institute. He then took a position with North American Aviation and, while on a fellowship from them, earned his Ph.D. in mathematics (functional analysis) from U.C.L.A. He then returned to the east coast and to an academic environment as a John Wesley Young Research Instructor at Dartmouth. He continued his research into operators on duals of LF spaces and taught at Stevens Institute of Technology where he became an assistant professor of mathematics.
The industry-academia pattern of his career continued as he went to Boole and Babbage as a software analyst and then to Control Data Corporation in 1976. In that position he was able to use all of his talents and draw upon his previous experience: he taught, pursued his own research in algorithms, and did joint work with researchers in other fields. His best known teaching was a series of courses on vector numerical algorithms. He reached a wide audience, and few can forget his flamboyant style. More important than specific techniques was the approach Mike always emphasized. His focus was always on doing what had not yet been done. Although he was not one to turn up his nose at a ten percent speedup, he was always looking for the sort of speedup that would allow somebody to do science that had not been done before. He constantly insisted that the only way to get such speedups was to rethink the entire solution, from the underlying science and mathematical abstraction through the computer architecture. Syntactic transformation of codes would not suffice: There was a need for what he termed 'semantic vectorization."
Mike's collaborative research was very influential in several areas. Perhaps his greatest influence was in the area of fluid dynamics. Not the least of his contributions was getting scientists to think of supercomputers as objects worthy of study. He always maintained that the existence of vector processors should affect the way we approach problems, and he was able to win many people over to that position. In doing so he never seemed to lose sight of the forest for the trees. Perhaps his favorite quote was Richard Hamming's statement that "the object of computing is insight and not numbers."
As a teacher he was generous with his time, and he helped many of us to broaden our vision. His death is a loss to the entire community, and we miss him very much.