Newsletters - MATLAB News & Notes
Colleagues Celebrate MATLAB Founder, Cleve Moler, at 60
by Mary Beth McDonald
 Cleve Moler is chairman and chief scientist at The MathWorks. Before leaving academic life for the computer industry, Moler was a professor of math and computer science for almost 20 years at Michigan, Stanford and the University of New Mexico. He then spent five years with two computer hardware manufacturers, the Intel Hypercube organization and Ardent Computer, joining The MathWorks full-time in 1989. Cleve's professional interests center on numerical analysis and mathematical software. In addition to being the author of the first version of MATLAB, he is one of the authors of the LINPACK and EISPACK scientific subroutine libraries. He is also co-author of three textbooks on numerical methods. |
He entered Caltech at a time when the very term numerical analyst was just coming into use. He then went to Stanford to further his studies under George Forsythe. In 1967, Cleve Moler co-authored his first book-a book that everyone agrees made a profound impression. It was also one of the first textbooks to combine theoretical analysis with production-grade software. MATLAB® founder and MathWorks chief scientist, Cleve Moler, recently turned 60. To celebrate this occasion, a special event was held October 8-9, 1999. The festivities were kicked off on Friday evening when friends, university colleagues, students, and co-workers assembled on the top floor of The MathWorks new building at 3 Apple Hill in Natick, MA for a reception. The reception brought together people associated with Cleve's undergraduate and graduate school days and his academic and professional careers. Posters adorned the walls chronicling Cleve's contributions to computational science. Most spectacular was the luminous ice sculpture of the L-shaped membrane, the focus of Cleve's doctoral work. Early on Saturday morning, in the tradition of academia, leaders in mathematics and computer science assembled to pay tribute to Cleve by talking about the discoveries and self-discoveries they made while working with him as a teacher, author, mentor, student, or friend. Remembering when Cleve was a studentThe master of ceremonies was Gene Golub, director of the Scientific Computing/Computational Mathematics Program in the School of Engineering at Stanford University. Professor Golub opened the proceedings with his recollections of Cleve as a graduate student. He described Cleve as being one of those students who you know instinctively will have a successful career and said, "At seminars, he would ask the deepest and most penetrating questions-he was clearly a leader during his graduate years." Golub also recalled Cleve visiting Stanford on several occasions when he was in academic life and remembered: "One year Cleve and Pete Stewart completed their work on the QZ algorithm, which is certainly one of the most elegant algorithms of modern numerical linear algebra. Another time, he taught a course based on one of the earliest versions of MATLAB. This course may have inspired the beginnings of The MathWorks. Cleve always made himself available to students and showed great patience even with rather low-level questions. It's wonderful to review his many contributions to our community. There is no better!" |
 |
| Gene Golub (standing) chatting with Michael Overton, Sven Hammarling, and Michael Saunders. |
Longtime colleagues
Margaret Wright, former president of the Society for Industrial Applied Mathematics (SIAM) and a distinguished member of the Bell Labs Computing Sciences Research Center, first met Cleve in 1965, when she was an undergraduate at Stanford. Wright remembered being in Cleve's course on elementary numerical analysis. She recalled, "The course was very lively and I immediately found a new interest." She met Cleve again in 1972 during a Ph.D. summer program run by Cleve and Jim Wilkinson at the University of Michigan.
Whenever Wright is talking about solving linear systems, she always remembers Cleve saying, "'It depends on whether you want a small residual or a small error in the solution.' He said this over and over. And today, I find myself repeating this very same statement as well, and each time I do, I distinctly hear Cleve's voice repeating it in his own inimitable way."
Wright continued by saying:
"Cleve has always been very active professionally and is a longtime devoted member of SIAM. During 1995 and 1996, when I was serving as the president of SIAM, Cleve was the vice president, and today he sits on the board. Whenever Cleve runs for office, he always gets more votes than anyone else.
MATLAB is an essential ingredient in my professional life. It is inconceivable for me to imagine not using MATLAB. It is the natural language of working scientists and engineers. Everything I do, in terms of computations, is infinitely easier because of MATLAB. It lets me do things that I never would have dreamed possible."
Another long-time colleague, Gilbert Strang, professor of mathematics at MIT and president of SIAM, said: "My working contact with Cleve came from his generous offer to prepare Linear Algebra Teaching Codes to go with my textbook. This was a Sunday morning job. Not one morning! We had to do things that MATLAB would never countenance-avoid pivoting no matter how small the pivot, ignore instability of all kinds, even find determinants for Cramer's Rule (Cleve had to close his eyes for a crime like that)."
"But the result was fun, and now (updated by Steve Lee with counts of repeated eigenvalues and independent eigenvectors and more) the codes are on the linear algebra page (http://web.mit.edu/ |
 |
Cleve's Ph.D. students
Another group of people to pay homage share a special bond-each was a former Ph.D. student under Cleve's direction. The three speakers were Stanley Eisenstat, professor of computer science, Yale University; Jack Dongarra, distinguished professor, Department of Computer Science, University of Tennessee and distinguished scientist, Oak Ridge National Laboratory; and Charles Van Loan, professor of engineering and chair of the Department of Computer Science, Cornell University. Charles Van Loan recalling his days as a student.
 |
|
 |
|
 Jack Dongarra giving his presentation. |
Eisenstat gave a brief history of methods for solving the L-shaped membrane eigenvalue problem, beginning with Cleve's thesis from 1965, "Finite Difference Methods for the Eigenvalues of Laplace's Operator." Dongarra named his talk "26 Years, 3 Months, and 30 Days or How I Learned to Quit Worrying and Love Numerical Software" (that's how long he has known Cleve).
Charles Van Loan recalled, "In my senior year at the University of Michigan, I took a two-semester sequence in numerical analysis from Cleve Moler. We used Cleve's book with Forsythe on linear equation solving and I bought a copy of Wilkinson's Algebraic Eigenvalue Problem. That was my start in matrix computations, where I have worked ever since. Cleve was a great teacher and I inherited from him all my perspectives on computing and mathematics."
More words from Charles Van Loan:
"Cleve Moler has played a very critical role in my academic career ever since I first walked into his office while an undergraduate at the University of Michigan. As teacher, Ph.D. advisor, and force behind MATLAB, Cleve has defined the way I look at mathematics and computing.MATLAB affects the way we do research in scientific computing because it encourages experimentation with interesting mathematical ideas. Visualization and vector-level thinking are supported in a way that permits focus on high-level issues. It is by clearing such a wide path from research to applications that MATLAB has been such an uplifting force in computational science."
MATLAB Toolpack colleagues
The next to speak was Alan Edelman, associate professor of applied mathematics and member of the Laboratory for Computer Science at MIT. Edelman described MATLAB*P (formerly called MITMatlab) as a "small software research project aimed at bringing the benefits of interactivity to parallel supercomputing, particularly for large problems." Edelman also noted that "for every step of the way, for every detailed choice of syntax and functionality," members of his team ask themselves, "What would Cleve Moler do if he had to make this decision?"
Like several other speakers, Nick Higham, professor of applied mathematics at the University of Manchester and Larry Shampine, professor of mathematics at Southern Methodist University acknowledged two important and influential books: Computer Solution of Linear Algebraic Systems by G. E. Forsythe and C. Moler (1967), and Computer Methods for Mathematical Computations by Forsythe, Malcolm, and Moler (1977), which was also among the first textbooks to blend theory and practice.
Larry Shampine, who is well known for his research in ordinary differential equations, gave his own tribute to the MATLAB community. "Nowadays everyone recognizes that MATLAB is an enormously useful tool for both teaching and research. Cleve got it going, but many of us here have joined him in making MATLAB an important contribution to the scientific community of the world. I'm sure you're all just as proud as I am to have been associated with him in his effort."
Cleve at The MathWorks
 |
| Enjoying dinner together are (from left to right) Al and Susan Dollberg, Cleve and Patsy Moler, Nick Trefethen, John Gilbert, and Jack Little.
|
Approaching the end of the day, but not the end of the tributes, Jack Little, president and CEO of The MathWorks, took the podium to tell of the countless projects Cleve has initiated at The MathWorks-beginning with the invention of a language that uses a single data type, to having modern data types work in a matrix world. Among the projects are the MATLAB Compiler, the Symbolic Toolbox, and the MATLAB Student Edition-all of which today form the basis of whole development groups.
Jack Little chose the following passage from George Sand to best describe Cleve's contributions, "Simplicity is the most difficult thing to secure in this world; it is the last limit of experience, and the last effort of genius." He then went on to say:
"What Cleve has done with MATLAB has truly changed the world, how the world does its computations, and learns its computations. Here at The MathWorks, and for many people around the world, MATLAB has enabled us to have an impact in many specialized areas by building off Cleve's fundamental contributions. We thank Cleve for giving us the opportunity to affect the world just as he has."
After the talks, a banquet, and some fun with a game of "MATLAB Jeopardy," Jack Little announced the final tribute: a graduate student fellowship in engineering with a focus on control systems at Stanford University in the name of Cleve B. Moler. The fellowship was instituted by The MathWorks and its distributors in Japan, Sweden, Italy, Germany, and the Czech Republic. This is a fitting tribute to a man who has inspired so many careers in the field.
