Forsythe, George Elmer
Forsythe, George Elmer
FORSYTHE, GEORGE ELMER
(b. State College, Pennsylvania, 8 January 1917, d. Stanford, California, 9 April 1972),
mathematics, computer science, education.
Forsythe is considered one of founders of computer science as an academic discipline. Trained as a numerical analyst and meteorologist, he pioneered the use of digital electronic computers to solve complex mathematical equations. The Stanford University Computer Science Department, which he formed in 1965, was exemplary, both in intellectual scope and institutional structure.
Early Life and Training . George Elmer Forsythe was born 8 January 1917 in State College, Pennsylvania, to Dr. Warren Ellsworth Forsythe, an academic physician and cofounder of the American Student Health Association (ASHA), and DeEtta Brodie Forsythe. George was raised in Ann Arbor, Michigan, where his father worked as director of the University Health Service of the University of Michigan. Evidence of his interest in computing first appeared in the seventh grade, when he attempted to use a hand-cranked desk calculator to see how digits repeated in large decimal expansions (e.g., 10000/7699). At Swarthmore College, he received a liberal arts education, earning a BA in mathematics in 1937. There he also met his future wife and lifelong intellectual collaborator, Alexandra (Sandra) Winifred Illmer. Following graduation, he and she both pursued doctorates in mathematics at Brown University in Providence, Rhode Island.
At Brown, Forsythe wrote a dissertation in numerical analysis entitled Riesz Summabilitly Methods of Order r, for R (r)< 0, Cesaro Summability of Independent Random Variables, under the direction of Jacob David Tamarkin and the probability theorist William Feller. While Forsythe received the support of his advisors and peers, and finished his PhD in four years, Illmer struggled in an environment that was inhospitable to female mathematicians. Ultimately, she left Brown and completed a master’s degree at Smith College. Forsythe and Illmer were married in a Quaker ceremony on 14 June 1941, the same day that Forsythe received his PhD. In the fall of 1941, the Forsythes moved to California so that George could start work as an instructor for the Stanford University Department of Mathematics.
Weather, Numerical Analysis, and Computers . Forsythe’s first career at Stanford was terminated by the United States’ entry into World War II. During the war, he served in the U.S. Army Air Forces (USAAF) as an instructor in theoretical meteorology for a training unit based at the University of California at Los Angeles (UCLA). With the unit’s head, Jörgen Holmboe, and his colleague William Gustin, he adapted the training lectures into Dynamic Meteorology(1945), a mathematics-oriented introductory textbook to theoretical meteorology. While working with the meteorologists, Forsythe began to develop a serious interest in employing electronic computers to solve complex applied mathematical problems. He was particularly inspired by the work of Lewis Fry Richardson, a British mathematician and meteorologist who had proposed in his 1922 Weather Prediction by Numerical Process that the weather might be forecast by means of a vast computation of differential equations—then far beyond human capability—provided that enough upper-air weather observations were available as input data. Pairing Richardson’s proposal with the mechanical computing technology that he saw in Wallace John Eckert’s 1940 Punched Card Methods in Scientific Computation, Forsythe came to believe that such machines could dramatically advance science and industry.
Immediately following the war, Forsythe pursued his interest in computational machines at Boeing, where he supervised the implementation of a tabulating machine set up for data processing. Sandra Forsythe also worked in private industry but left to raise the couple’s two young children, future botanist Warren L. “Tuck” Forsythe (b. 1944) and cultural anthropologist Diana E. Forsythe (b. 1947). In 1948, George Forsythe returned to the UCLA campus, this time to join the short-lived but highly influential Institute for Numerical Analysis of the National Bureau of Standards. While at the institute, he published several papers on numerical analysis, some in collaboration with Feller, and organized a major effort to translate contemporary and historical Russian mathematical texts. He was also a major contributor to the institute’s 1951 symposium on Simultaneous Linear Equations and the Determination of Eigenvalues, to which he presented a bibliographic paper, “Tentative Classification of Methods and Bibliography on Solving Systems of Linear Equations” that long served as an important reference for numerical analysts.
Most of Forsythe’s energy in the early 1950s, however, was devoted to helping develop and write programs for the institute’s Standards Western Automatic Computer (SWAC), an early digital electronic computer with 256 37-bit words of Williams-Kilburn tube memory, which became operational in 1950. As he became more familiar with SWAC’s operation, Forsythe focused on the problem of expressing algebraic equations on computers, often collaborating in this area with his wife and the mathematician Marcia Ascher. In the mid-1950s, Forsythe began to publish the algorithms he employed to communicate algebraic equations on SWAC; these established him as one of the foremost experts on implementing higher mathematics on digital electronic computers. After the institute was incorporated by UCLA in 1954, Forsythe joined the university’s faculty.
Establishing Computer Science at Stanford and the ACM . In 1957, John G. Herriot, a fellow Tamarkin student and computer enthusiast, recruited Forsythe to Stanford’s Mathematics Department. At Stanford, Forsythe’s initial task was to help Herriot meet the rapidly growing demand for computer courses and train graduate students in numerical analysis. There Forsythe also became vocal on the issue of how students, both inside and outside the Mathematics Department, should be educated to use computers. His widely read 1959 American Mathematical Monthly paper, “The Role of Numerical Analysis in an Undergraduate Program,” suggested dozens of ways to incorporate computers into university courses and asserted their pedagogical value: “The automatic computer really forces that precision of thinking which is alleged to be a product of any study of mathematics” (p. 655). Meanwhile, with Sandra, he also wrote extensively on teaching computers to secondary school students; by the 1960s, Sandra had become a major authority on teaching computers to high schoolers. Her most widely read books were her Computer Science series (1969–1973), which included a general primer and introductions to the FORTRAN and Basic programming languages.
Despite his increased involvement with computing, Forsythe still regarded himself as a mediator between theoretical mathematics and application-oriented engineering, rather than a computer scientist per se. His 1960 textbook Finite-Difference Methods for Partial Differential Equations (written with Wolfgang Wasow), emphasized the role that the numerical analyst would play in bridging the gap between mathematical theory and practical application, the latter category including the use of digital electronic computers. By 1961, however, he had concluded that training in numerical analysis alone would not suffice to generate mathematicians or engineers capable of conducting meaningful work on computers within a reasonable amount of time.” To provide students with “the general-purpose mental tools” he believed they needed, he founded that year the Division of Computer Science within the Mathematics Department. Though the division’s offerings were at the outset hardly distinct from those of the broader department, Forsythe envisioned a curriculum that included not only more traditional mathematical training in numerical analysis and automata theory, but also offerings from experts in programming, data processing, business games, adaptive systems, information theory, information retrieval, recursive function theory, and computer linguistics.
With strong institutional support from Stanford provost Frederick Terman and dean Albert H. Bowker, who saw the establishment of computer science as resonant to their own aims of reorienting the university toward federally funded, commercially stimulating, interdisciplinary research, Forsythe imported about a dozen young computer experts of complementary strengths from diverse areas of academe and industry to fulfill his vision of a broad but nevertheless cohesive discipline of computer science. Between 1961 and 1965, he recruited, among others: John McCarthy, from the Massachusetts Institute of Technology’s Mathematics Department; Niklaus Wirth, from the University of California, Berkeley’s Electrical Engineering School; Edward A. Feigenbaum, from Berkeley’s Business School; Gio Wiederhold, from the aeronautics industry; Bill Miller, a high-energy physicist at Argonne National Laboratories; Eugene Golub, from Lawrence Livermore Laboratories; and Donald Knuth, a mathematics graduate student at the California Institute of Technology. Under Forsythe’s leadership, members of the division produced some of the most well-known computer science concepts, programming languages, and projects. These included McCarthy’s and Feigenbaum’s pioneering work in Artificial Intelligence (AI), Wirth’s development of the programming language Pascal, and Knuth’s authorship of the reference The Art of Programming. Collectively, they also established an exemplary culture that remained firmly grounded in mathematics while eschewing mathematicians’ aversion to experimentation.
In 1961, Forsythe was also appointed director of Stanford’s Computation Center, a facility that housed most of the machines used by the university’s researchers. Forsythe’s responsibilities included acquiring equipment, determining how various scientific projects could incorporate computers, and aggressively encouraging scientists to computerize their research agendas. The targets of his most vigorous proselytizing were life scientists, the group he believed had the farthest to travel in terms of computerizing their research but also the most to gain from computers. Beyond offering biologists the resources of the Computation Center, he led an effort in 1962 and 1963 to convince the National Institutes of Health (NIH) to grant Stanford about one million dollars toward supporting an IBM 7090 that would be used primarily by pharmacologist Keith F. Killam, Jr., neurologists Kao Ling Chow and Frank Morrell, psychologist Karl Pribram, and medical statistician Lincoln E. Moses. Ultimately the NIH turned down Forsythe’s proposal on the grounds that the project was too exclusive and that biomedical researchers did not have sufficient knowledge of computers to use them nonsuperficially. Although Forsythe did not again become involved in biomedical computing endeavors, his proposal served as a model for Joshua Lederberg’s successful 1965 attempt to establish a large, federally funded computing center in Stanford’s Medical School.
While Forsythe was forming the Division of Computer Science and running the Computation Center, he remained a prolific contributor of both algorithms and prose to the Communications of the Association for Computing Machinery (ACM). In January 1964, he was appointed editor of the algorithms section of Communications and just six months later was elected president of the ACM, then the largest, most influential professional society within the theoretical computing and software development communities. During his two-year term as ACM president, Forsythe used his position not only to establish computer science as a fully recognized academic discipline but also to define the discipline’s priorities. Repeatedly he encouraged universities (and their sponsors) to form computer science departments along the lines of his Stanford Division: closely associated with mathematics, unhesitant to blend theory and experimentation, and attuned to the vast importance of research to the development of computer technology. From his ACM pulpit, Forsythe also called for the academic community to recognize algorithms as pieces of scientific scholarship to be refereed and edited. Although Forsythe’s “scientific” approach to computing was popular within the ACM, his motion to rename the organization the Association for Computing and Information Sciences did not muster the required two-thirds of the membership’s votes. Following his presidency, Forsythe initiated the Education Section of Communications and served as its first editor.
The Computer Science Department . By January 1965, Forsythe’s Computer Science Division had garnered enough productive faculty and had drifted far enough from the Mathematics Department’s immediate interests that the Stanford administration chose to accept his petition to recognize it as the Computer Science Department. Though Stanford’s was not the first computer science department, it was widely recognized as one of the most important intellectual centers of the field. Under Forsythe’s direction, the Computer Science Department continued the division’s exemplary role in defining the priorities of computer science, but it also established a widely emulated template for how the field’s departments would approach issues of research funding and collaboration with other disciplines.
Broadly, Forsythe ran Computer Science as a research institute that happened to train graduate students rather than as a traditional department. In line with Provost Terman’s decision to reduce Stanford researchers’ dependency on university resources, he provided department members only half-salaries, and insisted they look to federal funding agencies and private enterprise to raise the other half, as well as their research funds. Consequently, Stanford computer sciences formed longstanding and deep ties to federal agencies such as the Advanced Research Projects Agency (ARPA), the National Science Foundation (NSF), and the Office of Naval Research (ONR). Reliance on such sources provoked complaints that Stanford computer scientists would be constrained by and perhaps beholden to nonacademic forces, but the steady flow of funds into the department allowed it to expand and conduct expensive research even during times of relative poverty. Moreover, external support allowed department members to form “dukedoms,” united only by teaching concerns, in which they pursued their interests largely insulated from departmental politics as well as each other’s sometimes contradictory agendas. Forsythe further leveraged the department by setting up joint appointments, especially with the Engineering School, with the aim of providing computer scientists with access to pertinent research as well as more funding resources.
When he was not preoccupied with forging ties to his department’s sponsors, an effort requiring hundreds of letters and memoranda each month, Forsythe continued to pursue his interest in formalizing the pedagogy of computer science. In the late 1960s, he served as editor of Prentice Hall’s Series in Automatic Computation, a comprehensive 75-book set of computer science textbooks, monographs, and edited volumes. Among them were Computer Solution of Linear Algebraic Systems, which Forsythe coauthored with his student Cleve B. Moler in 1967, and Computer Methods for Mathematical Computation, coauthored with Moler and another student, Michael Malcolm (1977). Forsythe also remained a prolific author of algorithms and essays that attempted to stimulate computer use among a wider academic audience. Within mathematics he explored the challenges of solving quadratic and partial differential equations on computers and became active in the Society for Industrial and Applied Mathematics, to which he was elected trustee in 1971.
Around his fiftieth birthday, in 1967, Forsythe was diagnosed with terminal skin cancer. Even ill, he maintained his exceptional productivity until the last two weeks of his life in the early spring of 1972. Consequently, his apparently sudden death came as a profound shock to his colleagues. In his eulogy for Forsythe, Donald Knuth declared, “It is generally agreed that he, more than any other man, is responsible for the rapid development of computer science in the world’s colleges and universities.”
The Archival Collection, George and Alexandra Forsythe Papers, 1936–1979 (Stanford University Archives, SC 098), includes forty linear feet of professional and personal correspondence, notebooks, and publications. This forty linear foot collection is split into three series. Series 1, George Forsythe Papers, 1938-1972, includes professional and personal correspondence, publications, notebooks, and notes Forsythe took on various subjects, starting in college and continuing to just before his death. Series 2, Alexandra Forsythe Papers, 1970-1979, includes Alexandra Forsythe’s professional and personal correspondence, lectures, and teaching materials.Series 3, Addenda, 1936-1972, includes the Forsythes’ publications as well as annotated books, foreign language editions of their published work, and additional notes and correspondence.
WORKS BY FORSYTHE
Riesz Summability Methods of Order r, for R (r)< 0, Cesaro Summability of Independent Random Variables. PhD diss., Brown University, 1941.
“Riesz Summability Methods of Order r, for R (r)< 0.” Duke Mathematical Journal 8 (1941): 346–349.
With A. C. Schaeffer. “Remarks on Regularity of Methods of Summation.” Bulletin of the American Mathematical Society 48 (1942): 863–865.
“Cesàro Summability of Random Variables.” Duke Mathematical Journal 10 (1943): 397–428.
With Jörgen Holmboe and William Gustin. Dynamic Meteorology. New York: Wiley, 1945.
“Bibliographical Survey of Russian Mathematical Monographs, 1930–1951.” National Bureau of Standards Report 1628, 25 March 1952 (see also: Supplement, Rep. 1628A, 25 March 1952).
“A Numerical Analyst’s Fifteen-foot Shelf.” Mathematical Tables and Other Aids to Computation 7 (1953): 221–228.
With A. I. Forsythe. “Punched-Card Experiments with Accelerated Gradient Methods for Linear Equations.” National Bureau of Standards Applied Mathematics Series 39 (1954): 55–69.
With Marcia Ascher. “SWAC Experiments on the Use of Orthogonal Polynomials for Data Fitting.” Journal of the Association for Computing Machinery 5 (1958): 9–21.
“Bibliography on High School Mathematics Education.” Computers and Automation 8 (May 1959): 17–19.
“The Role of Numerical Analysis in an Undergraduate Program.” American Mathematical Monthly 66 (1959): 651–662.
With Wolfgang Wasow. Finite-Difference Methods for Partial Differential Equations. New York: Wiley, 1960.
“Engineering Students Must Learn Both Computing and Mathematics.” Journal of Engineering Education 52 (1961): 177–188.
“An Undergraduate Curriculum in Numerical Analysis.” Communications of the Association for Computing Machinery 7 (April 1964): 214–215.
“President’s Letters to the ACM Membership.” Communications of the Association for Computing Machinery 7 (1964): 448, 507, 558, 633–634, 698; 8 (1965): 3, 143–144, 422–423, 541, 591, 727; 9 (1966): 1, 244, 325.
“Algorithms for Scientific Computation.” Communications of the Association for Computing Machinery 9 (April 1966): 255–256.
With Cleve B. Moler. Computer Solution of Linear Algebraic Systems. Englewood Cliffs, NJ: Prentice-Hall, 1967.
“What to Do Till the Computer Scientist Comes.” American Mathematical Monthly 75 (1968): 454–462. Winner of the Lester R. Ford Award, 1969.
“Solving a Quadratic Equation on a Computer.” In The Mathematical Sciences: A Collection of Essays, edited by the National Research Council’s Committee on Support of Research in the Mathematical Sciences (COSRIMS) with the collaboration of George A. W. Boehm. Cambridge, MA: MIT Press, 1969.
With Michael A. Malcolm and Cleve B. Moler. Computer Methods for Mathematical Computation. Englewood Cliffs, NJ: Prentice-Hall, 1977.
Forsythe, Diana E. Studying Those Who Study Us: An Anthropologist in the World of Artificial Intelligence. Stanford, CA: Stanford University Press, 2001. Includes personal accounts of George and Alexandra Forsythe’s careers from the perspective of their daughter; emphasizes role of gender in shaping those careers.
Harriot, John G. “In Memory of George E. Forsythe.” Communications of the Association for Computing Machinery 15 (1972): 719–720.
Householder, A. S. “Obituary: George E. Forsythe.” SIAM Journal on Numerical Analysis 10–2 (April 1973).
Knuth, Donald E. “George Forsythe and the Development of Computer Science.” Communications of the Association for Computing Machinery 15 (1972): 721–727.
McCorduck, Pamela. “An Interview with Edward Feigenbaum.” The Charles Babbage Institute, OH 14, 12 June 1979.
Moler, Cleve B. “A Memory of George Forsythe.” SIGNUM Newsletter 7 (October 1972): 8–9.
Salisbury, David, and Gio Wiederhold, “George Forsythe, His Vision and Its Effects.” Stanford News Service, 26 November 1997. Available from http://www-db.stanford.edu/pub/voy/museum/ForsytheNews.html.
Varah, James. “The Work of George Forsythe and His Students.” In Proceedings of the ACM Conference on History of Scientific and Numeric Computation, Princeton, NJ, 13–15 May 1987, pp. 139–150. New York: ACM Press, 1987.
———. “The Influence of George Forsythe and His Students.” In A History of Scientific Computing. ACM Press History Series, pp. 31–40. Reading, MA: Addison-Wesley, 1990.