Velvel Kahan - Cleve Moler on Mathematics and Computing

[poster]

Velvel Kahan's informal name in Yiddish, װעלװעל, means "little wolf." If he needs a more formal name, Velvel uses William.

Contents

• First Meeting
• Anonymous Referee
• Close enough
• Householder VII
• Householder XVII
• The Office
• Turing Award
• IEEE 754
• Bibliography
• Oral History

First Meeting

I first met Velvel in George Forsythe's office at Stanford in 1962. Velvel had come to California from Toronto for a meeting of the IBM computer users group, SHARE, and visited Stanford on his way.

When Forsythe introduced us, Velvel said, "Oh, Cleve Moler, I have wanted to see you. Do you have a minute?" We went to my office and Velvel asked "Do you have a copy of the report you wrote at JPL last summer?" I retrieved the report from my meager collection. Velvel opened it to page 12, pointed to an equation, and said

"That's wrong."He was right.

Anonymous Referee

I submitted the manuscript for my first paper, "Iterative Refinement in Floating Point", to the Journal of the ACM in 1966. Forsythe was the editor who handled the submission. He sent it to an anonymous referee

The manuscript was 10 pages long. The referee's report was 12. Guess who refereed the paper.

Close enough

In 1968, Velvel was thinking of leaving the University of Toronto. Forsythe told me he considered offering Kahan a position at Stanford, but he was afraid Velvel would be too disruptive. "I hope he comes to Berkeley," George said. "That's close enough."



Velvel Kahan, 1968. (Photo Credit: George M. Bergman, Archives of the Mathematisches Forschungsinstitut Oberwolfach.)

Householder VII

I always met Velvel at Householder seminars. In 1977, the Householder seminar was at Asilomar, a California state park south of Monterey.

Velvel would begin talks about the importance of reliable floating-point arithmetic with a story that went something like this.

A physician leaves Sacramento late one evening. He flies hisprivate plane home to Santa Barbara. After setting the autopilot,he falls asleep. When the plane is over the mountains north of theairport, the autopilot fails with a floating-point overflow andthe plane crashes. The doctor's estate sues you because your companywrote the autopilot software.



This photo shows Velvel telling that story with an overhead projector.



Householder XVII

In 1993, the Householder seminar was in California again, this time at Lake Tahoe. Here is a rare moment when Velvel is actually listening to me.



The Office

Both Velvel and Beresford Parlett have been on the faculty at U. C. Berkeley since 1968. Velvel's student, Jerome Coonen, took this photo when he volunteered to help clear out Velvel's office.



Turing Award

The citation for Velvel's Turing Award in 1989 reads:

For his fundamental contributions to numerical analysis.One of the foremost experts on floating-point computations.Kahan has dedicated himself to "making the world safe for numericalcomputations!"



https://amturing.acm.org/award_winne...an_1023746.cfm



IEEE 754

My friend Axel Ruhe was from Sweden. He once asked me where he should visit on his first trip to the western United States. I recommended

• The Grand Canyon
• Las Vegas
• The IEEE floating-point arithmetic committee.

The meetings of the IEEE floating-point committee were legendary. They would go on for hours at various spots in Silicon Valey. Suffering from jetlag, Axel went to sleep on the floor of the conference room.

Here is Velvel, remembering the committee. https://www.youtube.com/watch?v=ATCpecsyPE8.



Photo Credit: Alaina G. Levine, Heidelberg Laureate Forum.

Bibliography

Nelson Beebe has compiled this bibliography.



Oral History

The manuscript of Thomas Haigh's interview of Velvel in 2005 for SIAM's oral history project is 229 pages long. Take a look. https://history.siam.org/oralhistories/kahan.htm.

Get the MATLAB code (requires JavaScript)

Published with MATLAB® R2024b