In Memoriam: George Dantzig
George Bernard Dantzig1914-2005
Operations Research Loses a
Pillar of the Profession
Some thoughts of the late, great "Father of Linear Programming" on the past and future of operations research.
By Peter Horner
|
Life is full of uncertainty. How do you explain it, let alone plan for it?
|
Fortunately for the O.R. community, Dantzig's math skills improved.
Dantzig went on to earn an A.B. degree in mathematics and physics from the University of Maryland (where his father taught mathematics), an M.A. in mathematics from the University of Michigan and a Ph.D. in mathematics from the University of California-Berkeley in 1946.
It was while a grad student at Berkeley in the 1940s that Dantzig displayed the unique brand of genius that would eventually elevate him to almost mythical status in the O.R. community. Dantzig, believing he was working on a couple of "homework" assignments, instead solved two famous "unsolvable" problems that had stumped generations of statisticians. A legend was born.
During World War II, Dantzig served as chief of the Combat Analysis Branch for the Air Force statistical control unit, setting the stage for his greatest achievement. After earning his Ph.D., Dantzig returned to Washington as a mathematical adviser to the U.S. Air Force. In 1947 while working at the Pentagon, he developed linear programming in order to mechanize the planning process.
Dantzig left the Pentagon in 1952 for a job as a research mathematician for the RAND Corporation. He held that post until 1960 when he returned to U.C. Berkeley to chair the university's Operations Research Center. In 1966 Dantzig was lured away by the Operations Research Department at rival Stanford, his academic home for nearly 40 years, right up until the time of his death.
George Dantzig (far right) congratulates 1999 recipients of the Dantzig Dissertation Award.
Not surprisingly, Dantzig was showered with countless awards and honors throughout his storied career. He received the National Medal of Science — the nation's highest science award — in 1975 for his work on linear programming. That same year Dantzig was the first recipient of the INFORMS (then ORSA and TIMS) von Neumann Theory Prize for "fundamental theoretical contributions to operations research and management science." In 1994 INFORMS created the George B. Dantzig Dissertation Award in his honor.
On Aug. 27, 1999, OR/MS Today editor Peter Horner interviewed the legendary Dantzig as he neared his 85th birthday. The conversation ranged from Dantzig's famous "homework" problem in Jerzy Neyman's statistics class at Berkeley more than 50 years ago to his thoughts on today's O.R. grad students and the direction they may take the profession in the next 50 years.
Needless to say, there was plenty to talk about (and, in some instances, to laugh about), starting with the uncertainties of the real world, which Dantzig devoted most of his adult life attempting to plan.
Some people are born athletes. Some people are born politicians. Were you a natural-born mathematician?
My father named me after George Bernard Shaw. I guess you've heard of him. He was a great playwright, and my father wanted me to be a writer and a playwright. Instead, I ended up following in my father's footsteps. So maybe the answer is, yes.
Would you mind retelling the famous story of your "homework problem" in Jerzy Neyman's stat class at Berkeley?
I was a grad student at Berkeley working on my Ph.D. I wasn't very good at getting to his class on time. Neyman had a habit of putting homework assignments up on the blackboard at the start of class. When I came in late, I'd copy the problem, take it home and work on it.
On this particular day there were two problems. They seemed more difficult than usual. When I handed in the assignment, I apologized for taking so long. Neyman told me to throw the paper on his desk. If you knew Neyman, you knew his desk was always covered with a huge pile of papers. I threw the paper on the top of the pile and left, never expecting to hear about it again.
One Sunday morning a couple of weeks later he came running over to my house and banged on the door. We lived upstairs. I came down and opened the door. He rushed in and said he had written an introduction to the problems I had solved and was going to submit the paper for publication. It turns out that those two problems were two very well-known, unsolved statistical problems. I had solved them both.
That's quite a story.
It has since become sort of an urban legend. Many years later, this fellow (Don) Knuth — he's a very well-known computer scientist — was bicycling down the street one day at Stanford, and he comes up to me and says, "Do you know you are influencing religion in middle America?"
He said that he had heard a sermon while he was visiting in Illinois or some place like that in which the preacher told what was essentially my experience with the homework problems to make a point about positive thinking. The point was this: If I had known that those were famous unsolved statistical problems, I never would have tried to solve them.
How did the preacher end up with the story?
I was on a plane and happened to sit down next to this well-known minister from Los Angeles. He had some pieces of sandalwood paper, and on these pieces of paper he had written little statements that had to do with positive thinking. I told him the story about my homework problem, and he was quite struck with it. The next thing I knew I received his write-up of my story in the mail. He asked for my permission to use it. He had exaggerated the story all over the place. I wrote back and told him that it wasn't five famous unsolved problems, but only two. I wasn't a high school student, I was in college at the time. I tried to correct it, but he went ahead and published the false version.
Speaking of taking liberties, did you ever see the movie "Good Will Hunting"?
No, but I've heard about it.
The plot involves a school janitor who sees a complex math problem on the blackboard and proceeds to solve it, much to the amazement of the instructor. Sounds familiar. I was wondering if you received any royalties.
No, I didn't, but I have a fairly good reputation in other areas, and I made a living that way.
Tom Magnanti (right) presents Dantzig with one of the many awards that INFORMS bestowed upon the "Father of Linear Programming."
You've said in the past that your work on linear programming was a gradual process rather than a revelation, but wasn't there a defining moment somewhere along the line when you realized that you were on to something big?
Well, I guess you might say there was such a point. All during World War II, I was a branch chief at the Pentagon working under Tex Thornton. We worked on planning and scheduling problems for the Air Force. I was certainly aware that the basic types of things we were doing could also be applied to industry. In the immediate post-war period I got my Ph.D. and was looking for a civilian job, but the Pentagon people didn't want me to leave. A fellow by the name of Marshall Wood and another fellow by the name of Dal Hitchcock proposed that I stay and work on mechanizing the planning process. That was also just about the time we saw the first computers coming on the scene. So I guess that would be the defining moment in the sense that here we were planning in military terms essentially what would be the scheduling and planning of a whole nation.
At that time did you have any idea that linear programming would turn out to be such an important decision-making tool and play such a key role in the development of the field of operations research?
I think I realized that it had the potential to do all of that. Yes.
That must have been an exciting time for you.
Yes, it was very exciting.
You are often called the Father of Linear Programming, but you could also be considered the Uncle of Operations Research because LP has meant so much to the discipline.
I think a fair thing to say is that the idea of operations research came out of what was going on with military planning during the war, but it existed before linear programming.
Do you consider yourself first and foremost a mathematician or an operations researcher?
I consider myself a mathematician. I don't differentiate between pure mathematics and applied mathematics. I know other people will tell you there is a difference, but there isn't.
Pure mathematicians have been known to describe operations researchers as second-rate mathematicians playing with toy problems. Care to comment?
Mathematicians are snobs.
Present company excepted. Pure math involves clean problems, but real-world problems are messy.
Real-world problems are all dirty, that's correct. Any model of them is an approximation of reality; it's not reality.
What's your definition of operations research?
I call it the science of decision-making. That is, all the variations of abstracting a problem and putting it into mathematical form. This includes the people who are inclined to be mathematicians who are trying to solve these abstract problems all the way down to people who have the real problem and have to get an answer to the boss. Operations research is all those things, and, depending on whom you talk to, you'll get a different picture of what it is.
UMASS-Amherst Professor Anna Nagurney (right) is one of a countless number of members of the O.R. community influenced by George Dantzig.
In the real world, coming up with an optimal solution is often only half the problem. The hard part is convincing some decision-maker to implement the solution.
That's correct, because the decision-maker knows the model isn't exactly the problem he's concerned about. I think he's got a right to be skeptical.
Looking at the public policy arena, do you think politicians and operations researcher can ever see eye to eye, or is there an inherent conflict between the two?
That's a good question. I don't know exactly what a politician is other than a lawyer who is trying to make a living by pretending he's doing something. Getting back to your question, you asked if there is an inherent conflict between operations researchers and politicians, and I would say yes, definitely. One is trying to solve the problem honestly, and the other is trying to solve it ...
Dishonestly?
Not purposely.
Would a classically trained operations researcher make a good United States president?
The closest we ever had that I can remember was this (Jimmy) Carter fellow. He had some scientific training. On the other hand, the president who had the least education was the most successful in my way of thinking, and that would be Harry Truman. He barely made it out of high school.
How would the country be different with an operations researcher living in the White House?
I think you would see a more serious effort to solve problems. For example, I work on planning under uncertainty. However, most of the solutions to problems that I see out there don't take uncertainty into account properly, or they make short shrift of it. They're always locking the barn door after the horse is gone; they're always making some effort to make houses a little more earthquake-proof after the earthquake. They don't do a complete job, so you get hit again.
What's the best advice you ever received?
You mean, how to play the stock market?
No, I'm thinking in terms of your career.
Let me see. I guess it came from my wife. This was just before the war, World War II. I was working on my Ph.D. at Berkeley. We went home to the Washington, D.C., area for the summer. I knew that the war was coming so I looked around to see what the possibilities were of getting involved in the war effort. Someone told me there was a fella by the name of Tex Thornton at the Munitions Building who was interviewing people.
I was interviewed by Tex Thornton on the corner of Constitution Avenue and 20th Street. My wife was with me, and Thornton was looking across the street and telling me what he was planning to do in the old Munitions Building, before the Pentagon existed. He wanted to give me a job. After the interview, I was talking to my wife, and I asked her what she thought. She said that if I were you, I'd hitch my wagon to that star, meaning Tex Thornton. I would say that was the best advice I ever received.
Dantzig was a founding member of The Institute of Management Sciences in the early 1950s.
You were the president of The Institute for Management Sciences in 1966. What stands out about that experience?
Thank you for reminding me of the year, but nothing stands out about it.
That same year, 1966, you left Berkeley, where you headed the Operations Research Center, for arch-rival Stanford. What lured you away?
I had an offer from Yale as well. Berkeley was a strange place. Let me tell you a story. I once had a birthday party in my honor, and various people were trying to explain why I went from Berkeley to Stanford. Among them was Ken Arrow, the famous economist. He was telling the audience all kinds of reasons behind my decision to leave Berkeley for Stanford, none of them true.
This is what happened. Jerry Leiberman wanted me to come to Stanford, so he was showing me around the campus. I noticed that the building where I would work had a little sign in the parking lot that said, "Reserved for Staff Only." That was it. I did it for the parking place next to my office.
That's as good a reason as any, I suppose. Very practical.
That's not the end of the story. The day I got hired by Stanford and went down there, they took away the sign. And so it was back to the usual — driving around looking for a place to park.
Some things never change. Shifting gears, how do you measure the success of an individual operations researcher?
One way I pass judgment is by going to meetings and hearing what's going on. I don't mean at the formal presentations, but out in the halls where people are talking and blabbing with each other. Out there you can begin to size up who knows something and who doesn't.
Are you impressed with someone who's published a bunch of papers?
Probably not. Is anyone reading them? Some people are very prolific, turning out papers left and right, but they're probably writing the same paper over and over.
Who does impress you?
Well, I always pay attention to someone who solves a problem that I worked on and couldn't solve. That would be my criteria for a really great person. Ralph Gomory, for example, what he did in integer programming fits that category.
Operations research was born in the crucible of World War II and flourished in the post-war period. Fifty years later, is the glamour gone?
To a certain extent, yes. I think it's being eclipsed by all the glamour associated with the electronic age. People are drifting off into computer science and programming.
What advice would you give a young math whiz today?
You can't really advise young people where to go these days. They just seem to smell a good direction and follow the scent.
Would that be the scent of money?
A lot of them smell money. The best brains in America as far as I can see are going to Wall Street, so, yes, they're smelling money — in more ways than one. The interest in the pure mathematical side of the thing is not as intense as it once was.
What are you working on these days?
I told you before, I work on planning under uncertainty. That's the big field as far as I'm concerned; that's the future. Maybe I'm the only one who says that.
As you near your 85th birthday, does anything still amaze you?
The electronic revolution completely bowls you over. Computing power seems to increase tenfold every five years or so. That's quite amazing.
What about the incredible ascent of the Internet?
I don't think the Internet is anything more than an [electronic] Sears Roebuck Catalog.
What will be your message to the INFORMS crowd in Philadelphia that will gather to celebrate your 85th birthday?
Well, I know they want me to say as little as possible so I'll try to keep my end of the bargain. I'll probably say, "Go West, young man, get yourself involved in planning under uncertainty."
The sounds like good advice. There's never a shortage of uncertainty in this world.
That's correct, and all the problems that are solved under deterministic means have that fundamental weakness — they don't properly take uncertainty into account.
Happy 85th birthday.
If I make it...
More planning under uncertainty? I can say for certain that a lot of people are planning on you making it to 85 and well beyond.
|
OR/MS Today copyright © 2005 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Rd., Suite 220, Marietta, GA 30060 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 2005 by Lionheart Publishing, Inc. All rights reserved.