Trained as a mathematician, Mr. Golub was an innovator in the field of numerical analysis. He created software formulas known as algorithms that made it possible to apply the first widely available generation of commercial computers to scientific and engineering problems as varied as predicting the weather and prospecting for oil. Mr. Golub was one of a small group of Stanford scientists who founded the university's computer science department in 1965.
His contributions were significant in an area known as matrix computation, said Cleve Moler, the chief scientist of Mathworks, a software firm that makes tools for mathematical calculation.
"Matrix computation is the foundation of modern scientific computing," Mr. Moler said. In particular Mr. Golub created a widely used linear algebra algorithm known as singular value decomposition, which has applications in signal processing and in statistics.
Singular value decomposition, or S.V.D., which he designed in 1964, has a variety of applications including search engines and data analysis. It has been called the Swiss Army knife of numerical analysis because of its versatility.
Mr. Golub was a revered figure in the tightly knit community of mathematical computer scientists. He played a central role in connecting scientists who were working on problems in the field, said Michael L. Overton, a professor of computer science and mathematics at the Courant Institute of Mathematical Sciences in New York.
Mr. Golub was born in Chicago in 1932 to parents who had immigrated from Latvia and Ukraine. He studied at the University of Illinois, where he received a doctorate in 1959. He went to Stanford as a visiting assistant professor in 1962 and taught there for 45 years.
He is survived by his brother, Alvin Golub of Northbrook, Ill.
By JOHN MARKOFF
Published: December 10, 2007 © New York Times