Paul Alper writes:
I found this at the 15:57 mark of one of Bill Press’s videos but do not know if this bit of history is well known in the MCMC universe. This is Marshall Rosenbluth criticizing Metropolis (and others). The text is taken from an interview of Rosenbluth in 2003 by Kai-Henrik Barth:
Barth:
So basically the Monte Carlo we use today in high energy physics and so on is based on your early work in the ‘50s?
Rosenbluth:
Well, it’s related to it, certainly. I mean, even before I got into it people were using Monte Carlo to trace neutron tracks in complicated reactors and so on. And photon transport using these statistical randomized methods, which was obviously the way to do complicated multi-dimensional calculations. And then Teller had the idea of maybe one should apply these techniques to statistical ensembles, so then I worked out this appropriate algorithm for doing that. We did the first papers on that. That was almost totally classical physics. It’s been a very widely growing field ever since. This meeting at Los Alamos in June had, maybe 200 or 300 people from all over the world.
Barth:
Your collaborators for these papers were Edward Teller and Nick Metropolis and your former wife?
Rosenbluth:
Yes. She actually did all the coding, which at that time was a new art for these new machines. You know, no compilers or anything like that.
Barth:
And it’s also listing A.H. Teller.
Rosenbluth:
That was Teller’s wife, who during the war had been one of these computer [women]— he wanted her to get back into the work, but she never showed up. So she was basically—
Barth:
Put on the paper for it?
Rosenbluth:
Yes. As was Metropolis, I should say. Metropolis was boss of the computer laboratory. We never had a single scientific discussion with him.
That’s funny, as I’d thought that a lot of it was Stanislaw Ulam’s idea. Ulam is not on the 1953 paper of Metropolis at al., but he published a paper in 1949 with Teller Metropolis on the Monte Carlo method. That earlier paper did not have all the details but I’ve always assumed that the method already existed at the time of the 1949 paper and that it just took them a few years to write it up, perhaps because of security issues (no joke, given that these methods were developed for the H-bomb project and of course the Russians had their own, competing program).
In his book, Ulam gives the impression that Teller had a habit of grabbing credit for others’ successful ideas. Anyway, I searched for *Ulam* in the Rosenbluth interview, and this is what I found:
Barth:
Could you fill us in about your participation, your role, in the development of the Ulam-Teller invention [which made modern hydrogen weapons possible]?
Rosenbluth:
Yes. I was a junior member and I had been doing all the detailed physics calculations on the booster and the GEORGE shot, which were similar, and I wasn’t even aware at the time until months later of this joint Ulam-Teller paper. So I can only give my own impression, which is that Teller really understood all the physics of what was going on—he was really a physicist—and that Ulam didn’t. And what I had seen what Ulam suggested was very vague and it seemed to me didn’t constitute any kind of a real invention, whereas what Teller came up with, with the help of Freddie de Hoffmann, me, and Dick Garwin, and others, was a pretty complete design before even the very detailed calculations had been done. So from my point of view, it really should be called “the Teller invention.” Ulam was a mathematician, not a physicist. A very smart and charming guy, but he was one of these people when he wandered into your office you had a feeling you were going to have a pleasant conversation for the next half hour, but you wouldn’t learn anything and nothing new would get done. On the other hand, when Teller came by you knew you were going to have sort of an intensive grilling and some new and relevant ideas thrown at you. So from this perspective, I couldn’t understand really why Ulam was receiving equal billing with Teller. But I don’t know what went on in the higher level discussions.
Barth:
As far as I understand, from Rhodes’ book, Ulam had the idea of radiation pressure first. But I’m not certain about that.
Rosenbluth:
No, I don’t think he did. . . .
Hmm . . . I really wasn’t there, of course, but based on my sympathy with Ulam I’m inclined to think that Rosenbluth is pretty much just reporting what Teller was telling him.
Otherwise, I dunno, maybe we need to rename Stan and call it Ed. I’d hate to do that, though, as people might then think it’s named after Ed Wegman . . .
For those of your readers who have not seen it, I’d highly recommend Gian-Carlo Rota’s portrait of Ulam, “The Lost Cafe”. Rota used to often say, with respect to Ulam, Teller, and the hydrogen bomb, that he knew personal secrets that he would take to his grave. Which, to the best of my knowledge, he did. I have no idea what that quip implies.
http://fas.org/sgp/othergov/doe/lanl/pubs/00285736.pdf
Mark, thanks for sharing that link. I really enjoyed reading GCR’s memoir. Extremely interesting.
This is hilarious and really well written.
“I remember once watching him at the blackboard trying to solve a quadratic equation. He furrowed his brow in rapt absorption, while scribbling formulas in his tiny handwriting. When he finally got the answer, he turned around and said with relief, “I feel I have done my work for the day.””
Mark:
Thanks for the link. Rota’s article was interesting but I was a bit put off by his mathematician-like way of ranking everyone in terms of ability. It wasn’t enough for him to say what each person did and could do; he had to compare and rank everyone.
https://statmodeling.stat.columbia.edu/2015/03/17/1980-math-olympiad-program-now/
“””
#1 math team kid in our team ….
Noam was obviously the best of all of us ….
I was probably about the 20th best out of 24. Had I practiced, I think I could’ve been 10th or 15th (then again, if everyone had practiced, maybe I would’ve been 23rd or 24th) ….
he was one of the top kids in the program ….
All 3 were, like me, near the bottom of the pack ….
In the middle of the pack ….
another guy who was trying his best but was no star ….
also in that middle-of-pack range ….
one of the top kids but not the very top ….
Obviously the most talented math kid in the group ….
“””
Anon:
Yes, exactly. That’s how we all thought back then. It was the mathematician way, and being on the math team didn’t help.
I reread the Rota article. It’s wonderful. I don’t know why I didn’t appreciate it enough the first time. It’s refreshing to see something written on a great scientist that’s not pure hero worship. I recently read a book about Neumann and it just made me want to barf. Not Neumann’s fault; it’s the fault of the biographers who want to just present an overwhelmingly positive take.
I did notice this funny in retrospect bit in Rota’s article:
The bomb, sure, that will never be forgotten, unfortunately. But the other part: Ulam’s problem books may already be forgotten. But the Monte Carlo method is now huge. It’s funny to think that in 1987, when Rota wrote that article, the Monte Carlo method was seen as a bit of an afterthought (albeit one with “substantial practical applications”), whereas now it’s a big deal in mathematics as well.
You have to take a lot of what is written by Ulam, Teller, and Rosenbluth about themselves and one another with a grain of salt. They were strong personalities that, on the whole, did not suffer from the stress of self doubt. A favourite quote concerns an interviewer who asked Ulam if he had worked with Professor Teller. Ulam replied “No, Professor Teller worked with me!”.
Andrew, are you aware of this restriction on the page that you quote from:
“This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.”
I hope you got permission from the AIP before posting this!
I was pretty impressed with the contemptuous comments of Dr. R; I’m glad this level of pettiness is not limited to low-stakes disciplines, but shows up wherever human beings show up.
Also impressed with the presenter of the slide, who says he knew Dr. R and (so) is inclined to believe him. If he had known Teller and Metropolis as well, he’d really have gotten his knickers into a bunch.
The phrase of the week for me is Stephen McNeill’s “They were strong personalities that, on the whole, did not suffer from the stress of self doubt.” I especially like the “on the whole”.
Plenty of people put things on the internet that they would later rather not have quoted.
I know some of physicists who worked with Teller, and one in particular was intimately involved in the early development of fusion weapons, and worked closely with Teller. Yes he’s very old. We have discussed the matter of who gets most of the credit. It was Teller, and not Ulam. Rosenbluth’s story is essentially correct.
Andrew, Your last sentence is not right, either historically or scientifically.
Mca:
I’m not at all saying you’re wrong–as noted above, I didn’t know any of these people—but, given that all these guys argued so publicly about who deserved the credit, it doesn’t count a lot for me that a guy who worked with Teller thought that Teller deserves most of the credit. From the other direction, Ulam’s friend Rota, in the article linked to in comments above, says Ulam came up with the “seed idea that finally worked,” after “demolishing” various misguided ideas from Teller. So I think it’s hard to say. I agree that the recollections of a friend of Teller are relevant to the discussion, I don’t see them as definitive.
Andrew,
The people I know who worked with Teller are actually quite critical of him on a personal level, so I’m not talking about the “recollections of a friend.” I’m afraid that Rota is simply wrong. It’s hard to clear the record because the technical details are still SRD (secret restricted data)– classified. Who did what and when is known, and I’m afraid that Ulam did not come up with the “seed idea that finally worked.”
My impression is that in a lot of these cases, the idea was one whose time had come (perhaps due in part to technology, or current interest, or advances in related fields), several people were on the point of inventing it and while it is fun to pick a winner, assigning priority to one person or group is pretty much a matter of luck. I don’t know the details of this particular case though.
http://en.wikipedia.org/wiki/Stigler's_law_of_eponymy
That’s been my experience, both firsthand and secondhand. But then computational linguistics and logic programming aren’t exactly nuclear physics.
I also think it’s very hard to judge when an idea crosses the finish line. I’ve heard objections to arXiv along the lines of “it just lets people stake a claim without working through the details.” I think those conjectures are important, but it muddies the waters of precedence, by which scientists like to judge themselves for hiring, promotion, tenure, and grants.
It’s also difficult to judge who did what with respect to originality vs. crank turning when you have a working group (or even two people) discussing issues over a long period of time.
The 1949 article that I’m familiar with was written by Nicholas Metropolis and S. Ulam. It appeared in the Journal of The American Statistical Association in September of that year (Vol. 44, No. 247, pp. 335-341).
Here’s the title and abstract:
The Monte Carlo Method
Nicholas Metropolis and S. Ulam
Los Alamos National Laboratory
We shall present here the motivation and a general description of a method dealing with a class of problems in mathematical physics. The method is, essentially, a statistical approach to the study of differential equations, or more generally, of integro-differential equations that occur in various branches of the natural sciences.
There are three references in the paper: one by von Neumann and Ulam, one by Goldberger, and two by C. J. Everett and Ulam.
Edward:
Yes, I’d misremembered Ulam’s coauthor in that reference, thanks for pointing it out.
Pingback: “Maybe we should’ve called it Arianna” « Statistical Modeling, Causal Inference, and Social Science