Dubrulle, Augustin SIAM oral history

Augustin Dubrulle discusses his career as a producer of software libraries for scientific computing and numerical analysis. Dubrulle grew up in France, studying mathematics at the University of Lille and serving in Algeria before accepting a job in 1962 with the IBM Scientific Service Bureau in Paris. He describes the work of the service bureau and the state of scientific computing and numerical analysis in France during this period. In 1967 he moved to the IBM Scientific Center in Paris, but soon left for White Plains, NY, on secondment to work on a conversion of IBM’s Scientific Software Package (SSP) for PL/1. Dubrulle was part of a small group, which in 1968 moved to Houston. The group’s leader, Ed Battiste, left IBM to found leading scientific software firm IMSL. Dubrulle describes the origins of SSP in IBM Boblingen, its reception by users, and various updated and improved versions, including PL/Math and SL/Math. In 1970, Dubrulle moved to an IBM development office in Menlo Park, California, produced a mathematical software package called Math-Basic for the System 3 minicomputer, and gained permanent residency in the USA. He then worked for several years in the IBM Advanced Systems Development Division in Palo Alto on projects concerning scientific computations, including a stint in the Future Systems organization and a development for the measurement and modeling of computer performance. He then transferred to the Palo Alto Scientific Center to participate in a research project of computational geophysics for oil exploration. Upon completion of this assignment, he switched his interests to vector computations in support of IBM’s 3090VF vector supercomputer, for which he produced in 1989 a version of EISPACK. Dubrulle, who earned a Ph.D. in the mid 1980s under the direction of Cleve Moler, also discusses his relationship with prominent members of the mathematical software community, including Cleve Moler, Jack Dongarra, and Jim Wilkinson. After leaving IBM in 1992, he worked on a mathematical software library for Hewlett Packard’s Wide Word processor, which eventually became Intel’s IA64 architecture. He transferred to Convex, then in the process of being acquired by HP, where he continued to work on software libraries before retiring in 1998 when HP lost interest in this area.