Stanislaw Ulam: Monte Carlo, the H-Bomb, and a Mind That Survived Everything

Photo: Pranjall Kumar / Pexels
Stanislaw Ulam was a Polish-American mathematician who left his mark on two very different parts of the twentieth century. In 1946, while recovering from emergency brain surgery, he conceived the Monte Carlo method, a way of answering hard questions by running many random trials. Five years later he supplied the compression idea that made the two-stage hydrogen bomb workable. Born in 1909 in Lwów, then part of Austria-Hungary, he lost most of his family to the Holocaust and spent his later career at Los Alamos. This profile separates what is documented from what is legend.
From Lwów to the Scottish Café
Ulam was born on April 13, 1909, in Lemberg (now Lviv, Ukraine), into a wealthy Polish Jewish family. According to the MacTutor biography at the University of St Andrews, he began teaching himself mathematics from books at 14, starting with Euler's Algebra, and published his first papers in 1929 while still an undergraduate at the Lwów Polytechnic Institute (MacTutor). He took a doctorate there in 1933 under Kazimierz Kuratowski, a thesis he later said he wrote in six hours.
The city mattered as much as the schooling. Lwów hosted one of the most productive mathematical communities in Europe, and its members worked in cafés. The best known is the Scottish Café, where Stefan Banach and his colleagues wrote problems into a notebook that became the Scottish Book. It recorded 193 problems between 1935 and 1941, and Ulam contributed 40 on his own, 11 with Banach and Stanisław Mazur, and 15 with others. Hugo Steinhaus kept a copy through the war and sent it to Ulam in 1957, and Ulam translated it into English (Wikipedia).
Think of the café as an open-source repository before the term existed: anyone could add a problem, anyone could try one, and the record was kept in plain sight. Ulam's early specialty was measure theory, the rigorous mathematics of assigning sizes to sets, and his work on measurable cardinals sits at the border between set theory and analysis. That foundation matters later, because Monte Carlo is at heart an application of probability measures.
| Date | Event |
|---|---|
| April 13, 1909 | Born in Lemberg (Lwów) |
| 1929 | First papers published in Fundamenta Mathematicae |
| 1933 | Doctorate at the Lwów Polytechnic under Kuratowski |
| January 1936 | Arrives in New York at von Neumann's invitation |
| August 1939 | Sails to the US with his brother Adam, shortly before the invasion of Poland |
| 1944 | Arrives at Los Alamos |
| January 1946 | Encephalitis and emergency surgery in Los Angeles |
| 1949 | Metropolis and Ulam publish "The Monte Carlo Method" |
| March 9, 1951 | Teller-Ulam report on staged implosion |
| May 13, 1984 | Dies in Santa Fe |
Leaving Europe just in time
In 1935, John von Neumann invited Ulam to visit the Institute for Advanced Study in Princeton. Ulam landed in New York on January 7, 1936, and then spent academic years at Harvard as a member of its Society of Fellows, returning to Poland for summers. The arrangement ended in the summer of 1939. Ulam and his brother Adam, then 17, sailed for the United States in August, arriving less than two weeks before Germany invaded Poland (MacTutor).
The family left behind did not survive. His mother Anna had died in Vienna in 1938, and MacTutor reports that his father Józef was killed in 1941 when the Germans occupied Lwów, along with his sister Stefania. Ulam's memoir, Adventures of a Mathematician (1976), is famously light in tone, and readers have long noted how little of the loss it states directly.
In America he worked at Harvard and then at the University of Wisconsin, where he became an assistant professor in 1940. He married Françoise Aron, a French-born woman he had met in Cambridge, Massachusetts, on August 19, 1941. Their daughter Claire was born in Santa Fe in 1944.
How did a Lwów topologist end up at Los Alamos?
Von Neumann was the connection. Accounts differ on the details: MacTutor says von Neumann asked Ulam to join war work after the two met in a Chicago railroad station in 1943, while another biography records an invitation from Hans Bethe in October 1943 and an arrival at Los Alamos in February 1944. Both agree that the move came from the mathematicians' network that had formed before the war. At Los Alamos, Ulam worked on hydrodynamic calculations for the explosive lenses of an implosion-type weapon, and in Edward Teller's group on the "Super," the early idea for a hydrogen bomb.
The illness that produced Monte Carlo
In January 1946, Ulam was newly appointed as a professor at the University of Southern California. One morning in Los Angeles he woke unable to speak. Doctors diagnosed encephalitis, an inflammation of the brain, and within hours he underwent dangerous surgery. His friend Paul Erdős is said to have visited during the recovery.
While convalescing, Ulam played solitaire and asked a question that sounds trivial: what are the chances that a game of Canfield solitaire comes out? The combinatorics were too tangled to work out by pencil. A simpler route occurred to him. Lay out the game a hundred times and count the wins. Then he made the leap that mattered. The same logic, he realized, applied to the physics problems at Los Alamos, such as how neutrons diffuse through a bomb core (Wikipedia, Monte Carlo method).
An analogy helps. Political pollsters do not ask every voter; they ask a random sample and trust the arithmetic of sampling to estimate the whole. Monte Carlo does the same with a simulated world. Instead of solving every branch of a problem, you sample many random paths through it and let the average stand in for the answer.
How does a random experiment produce a precise answer?
The standard classroom example estimates pi. Draw a square, inscribe a quarter circle, and scatter random points across the square. The fraction landing inside the quarter circle approximates pi divided by four, so multiplying by four estimates pi. More points give a better estimate, but slowly: the error shrinks in proportion to one over the square root of the number of samples, so quadrupling the samples only halves the error (Wikipedia, Monte Carlo method).
That slow convergence sounds like a weakness, and for simple problems it is. The strength is that the rate does not depend on the number of dimensions. For problems with dozens or thousands of variables, where grid-based methods become hopeless, the same one-over-root-N behavior holds. That property is why the method survived and spread.
Who named it, and how did it become a real tool?
Later in 1946, Ulam described the idea to von Neumann, and the two began planning calculations. Because the work was classified, it needed a code name. Nicholas Metropolis suggested "Monte Carlo," after the casino in Monaco where Ulam's uncle Michał borrowed money from relatives to gamble. In March 1947, von Neumann proposed a statistical approach to neutron diffusion, and in spring 1948 he, Metropolis, and others programmed the ENIAC computer to run the first fully automated Monte Carlo calculations, on a fission weapon core. Metropolis and Ulam published the first open paper on the method, "The Monte Carlo Method," in the Journal of the American Statistical Association in 1949.
The sequence shows how the method really emerged. Ulam supplied the idea, von Neumann the mathematical formalism and the push toward computers, and Metropolis the programming and the name. As with many scientific advances, credit reflects a team more than a lone inventor, though Ulam's own account places the solitaire insight with him.
Where is Monte Carlo used now?
The history section of the Monte Carlo article notes that the method spread from nuclear weapons work into physics, physical chemistry, and operations research, aided by funding from the RAND Corporation and the US Air Force. Today, simulation by random sampling is a routine tool in fields that deal with uncertainty or with systems too complex for closed-form solutions. The examples in the table are general categories, not a ranking.
| Field | How random sampling is used |
|---|---|
| Neutron and radiation transport | Follow simulated particles as they scatter and are absorbed |
| Operations research | Test how schedules and supply chains behave under random delays |
| Particle physics | Simulate detector responses to compare against real data |
| Finance | Price instruments and test portfolios across many simulated futures |
| Statistics | Sample from complicated probability distributions |
| Game-playing AI | Estimate which moves lead to wins by simulating play-outs |
The last row connects to a famous result: systems such as AlphaGo combined neural networks with Monte Carlo tree search, a method that decides on moves by sampling many simulated continuations of a game. The statistical programming language Stan, widely used for Bayesian modeling, is named after Ulam.
The Teller-Ulam design and the hydrogen bomb
The first American idea for a thermonuclear weapon, a bomb that gets much of its energy from fusing light nuclei, was the "Classical Super." Fusion fuel was placed around or inside a fission trigger in the hope that the fission explosion would ignite it. The approach proved unworkable: the heat of a fission explosion alone would not ignite the fuel.
According to the Wikipedia account of the design, Teller and Ulam devised the concept in January 1951, with Ulam's breakthrough idea being that the mechanical shock of a fission explosion could be used to compress a separate package of fusion fuel. Teller added the crucial refinement that radiation from the fission bomb, not just hydrodynamic shock, could do the compressing, and later suggested a fissile "spark plug" inside the fusion stage. Together they submitted a classified report on March 9, 1951, titled "On Heterocatalytic Detonations I. Hydrodynamic Lenses and Radiation Mirrors" (Wikipedia, Teller-Ulam design).
Think of the arrangement as a two-stage rocket for energy. The first stage, the fission primary, does not itself do the main job. It exists to squeeze the second stage, the fusion secondary, to the temperatures and densities where fusion can proceed.
What happened at Ivy Mike?
The first full test, code-named Ivy Mike, took place on November 1, 1952, at Enewetak atoll in the Pacific, with a yield of 10.4 megatons of TNT equivalent. The device used liquid deuterium, kept cold by about 20 short tons of cryogenic equipment, and weighed roughly 80 short tons in all, so it was a laboratory installation rather than a deliverable weapon. Its yield was more than 450 times that of the bomb dropped on Nagasaki (Wikipedia, Teller-Ulam design).
Who deserves the credit?
The question has been disputed for decades. Hans Bethe, who knew both men, offered a widely quoted compromise: "Ulam is the father, because he provided the seed, and Teller is the mother." Teller took a harder line and is quoted as saying, "I contributed; Ulam did not." Other scientists disputed how much of the credit belonged to Teller, and Teller himself later called the collaborative emphasis in his 1955 article a "white lie."
The design remains official secret territory. The US government has long declined to confirm details, and its attempt in 1979 to block a magazine article by Howard Morland was dropped after similar information appeared elsewhere. Anything beyond the broad outline above, including the exact materials and geometry, is not publicly confirmed. For the physics of the first stage, see how nuclear fission works, and for the fuel of the second, the explainer on nuclear fusion. The wartime scientific community that produced both is also visible in profiles of Hans Bethe and J. Robert Oppenheimer.
The Fermi-Pasta-Ulam-Tsingou problem
One of Ulam's most influential contributions was not a weapon at all. In the summer of 1953, Enrico Fermi wanted a physics problem to test the Los Alamos MANIAC, one of the first electronic digital computers. With John Pasta, Ulam, and the programmer Mary Tsingou, he set up a simulated chain of masses connected by springs with a slightly nonlinear force.
The expectation was thermalization. Classical statistical mechanics says that if you pluck one vibrational mode, the energy should spread out and eventually be shared evenly among all modes, just as cream spreads in coffee. The computer showed something else: a complicated, quasi-periodic behavior in which the system repeatedly returned close to its initial state. The Los Alamos report appeared in 1955, after Fermi's death in 1954, and Ulam is a named author (Wikipedia).
Why did this puzzle matter?
The result was surprising enough that it took a decade to explain. In 1965, Martin Kruskal and Norman Zabusky linked the system's continuum limit to the Korteweg-de Vries equation and its soliton solutions, solitary waves that pass through one another without changing shape. The solitons, they argued, prevent the energy from spreading. The episode is often described as launching the study of nonlinear systems and as one of the earliest uses of digital computers in mathematical research. Over much longer times the chain does eventually thermalize.
Tsingou's work was largely overlooked until 2008, when Thierry Dauxois published new details and called for the problem to carry her name. Modern references use the full "Fermi-Pasta-Ulam-Tsingou" label. The phrase is a small reminder of how often the person writing the code dropped out of the credit line. For another example of how deterministic rules can generate unexpected behavior, see the article on the butterfly effect and chaos.
The Ulam spiral
In 1963, according to the standard account, Ulam was sitting through a dull scientific talk and began doodling. He wrote the integers in a square spiral on graph paper and circled the primes. They did not scatter evenly. They bunched along diagonal lines. With Myron Stein and Mark Wells, he then used the MANIAC II computer at Los Alamos to extend the picture to about 100,000 points, published in 1964. Martin Gardner's March 1964 "Mathematical Games" column in Scientific American popularized the image and put it on the magazine's cover (Wikipedia, Ulam spiral).
The pattern is partly explained. Diagonal, horizontal, and vertical lines in the spiral correspond to quadratic polynomials, and some quadratics are unusually rich in primes. Euler's polynomial n squared plus n plus 41 is the famous example: starting the spiral at 41 produces a diagonal of 40 consecutive primes. Divisibility by small primes such as 3 and 5 explains why some diagonals are richer than others. What remains open is a precise, proven account of how often such polynomials yield primes. The relevant 1923 Hardy-Littlewood Conjecture F, which predicts the density of primes along such lines, is still unproven.
One historical footnote: a similar arrangement was published in 1932 by Laurence Klauber, a triangular array that showed comparable lines. Ulam's version became famous because it was visual, computer-assisted, and easy to reproduce. For a wider look at the mathematics of this era, see the biographies of Kurt Gödel and Alan Turing.
Project Orion and nuclear propulsion
Ulam's imagination reached into space travel. According to the Wikipedia history of Project Orion, he proposed the general idea of nuclear pulse propulsion in 1946, and in 1947 Frederick Reines and Ulam wrote a Los Alamos memorandum with preliminary calculations. In August 1955, Ulam and C. J. Everett co-authored a classified paper proposing that fission bombs detonated at a distance could propel a spacecraft (Wikipedia, Project Orion).
The design was a ship that rode a series of explosions. Each blast struck a heavy pusher plate on the underside, with large shock absorbers smoothing the jolt for the rest of the vehicle. Ted Taylor led the work at General Atomics, and Freeman Dyson took a year away from the Institute for Advanced Study to join it. ARPA agreed in July 1958 to sponsor the project at about one million dollars a year, then decided in late 1959 to withdraw support. The 1963 Partial Test Ban Treaty, which prohibited nuclear explosions in the atmosphere and in space, greatly reduced political support, and the project was canceled in 1964.
Pure mathematics under the applied fame
Ulam's applied work often overshadows a body of pure mathematics that would distinguish a career by itself. The Wikipedia biography lists contributions in set theory, topology, functional analysis, ergodic theory, group theory, number theory, combinatorics, and graph theory. Named results include the Borsuk-Ulam theorem, the Mazur-Ulam theorem, the Kuratowski-Ulam theorem, Hyers-Ulam-Rassias stability, Ulam numbers, lucky numbers, and the Ulam-Warburton cellular automaton (Wikipedia).
Between 1936 and 1939, working with John Oxtoby at Harvard, he did the ergodic theory work that appeared in the Annals of Mathematics in 1941. The cellular automaton that carries his name is a grid of cells that update by local rules, an idea related to the self-replicating machine theory that von Neumann and Ulam both worked on.
A cellular automaton is a good illustration of Ulam's style. Simple local rules, repeated, produce complex global patterns. That is also the spirit of the Ulam spiral, the FPUT chain, and Monte Carlo: take something simple, repeat it many times, and watch what emerges.
After the bomb: Colorado, memoir, and death
Ulam remained at Los Alamos for two decades. He then moved to the University of Colorado, first as chair of mathematics and later as professor of biomathematics at its medical school, and spent winter semesters at the University of Florida after retiring from Colorado. He gave the SIAM John von Neumann Lecture in 1963, and in 1976 the Polish government-in-exile in London awarded him the Commander's Cross with Star of the Order of Polonia Restituta. He was elected to the American Academy of Arts and Sciences, the US National Academy of Sciences, and the American Philosophical Society.
His autobiography, Adventures of a Mathematician, appeared the same year. He died in Santa Fe, New Mexico, on May 13, 1984, aged 75, of an apparent heart attack.
Did the encephalitis change his mind?
The question has a long history. In a 1987 article, the mathematician Gian-Carlo Rota, who knew Ulam, argued that the 1946 attack altered his personality and working style, shifting him toward speculative conjectures and away from sustained calculation. Rota cited the recollections of Ulam's friends Paul Stein and John Oxtoby. Françoise Ulam disputed the claim. A fair summary is that Ulam's career after 1946 is clearly different in emphasis, but whether the illness caused the change is contested, and the evidence consists of recollections rather than data.
What people get wrong about Ulam
He did not invent the hydrogen bomb alone, and neither did Teller. The clearest statement of the documented record is that the staging and compression idea came from Ulam, the radiation mechanism from Teller, and the two-author March 1951 report records the combination. Anyone who assigns the entire invention to one man is choosing a side in a dispute rather than reporting a fact.
Monte Carlo is not just "random guessing." The method works because it uses randomness systematically, with error that shrinks predictably as the number of samples grows. The same property that makes it slow on simple problems makes it robust on complicated ones.
The Ulam spiral is not a mystery of nature. Parts of it are well understood through quadratic polynomials, while the deeper density questions remain unproven. It is an unsolved problem in the ordinary mathematical sense, not an unexplained anomaly.
FPUT was not a failed experiment. The team expected thermalization and got recurrence. That surprise is the discovery, and it helped open the field of nonlinear dynamics.
The takeaway
Ulam's habit of mind was to turn hard questions into experiments that a computer could run. A card game became a method for neutron physics. A doodle on graph paper became a research problem about primes. A chain of springs became the starting point of nonlinear science. The same habit sat inside the most consequential and most contested episode of his life, the staging idea behind modern thermonuclear weapons, where credit remains disputed and the details remain classified. The usual image of a mathematician is someone who proves theorems alone. Ulam's legacy is closer to a method of working: sample, simulate, notice what surprises you, and write it down.
FAQ
Q: Did Ulam or Teller deserve more credit for the H-bomb design? A: Both made essential contributions, and the sources conflict on emphasis. In January 1951 the two devised the concept, with Ulam's idea that a fission explosion's shock could compress a separate fusion package, and Teller identified radiation as the better compressing mechanism. Bethe called Ulam the father and Teller the mother, while Teller said he contributed and Ulam did not. The design details remain classified.
Q: What exactly is the Monte Carlo method? A: It is a way to estimate answers by repeated random sampling. Instead of solving a problem analytically, you simulate it many times with random inputs and average the results. Error shrinks in proportion to one over the square root of the number of samples, and that rate does not depend on how many dimensions the problem has.
Q: How did Ulam's encephalitis lead to Monte Carlo? A: In January 1946 he woke in Los Angeles unable to speak, was diagnosed with encephalitis, and had emergency surgery. While recovering, he played solitaire and wondered how to estimate the odds of winning by playing many games. He then realized the same approach could handle neutron diffusion, and he described it to von Neumann later that year.
Q: What is the Ulam spiral? A: It is a square spiral of the integers with the primes marked. Ulam noticed in 1963 that primes line up along diagonals. The lines correspond to quadratic polynomials, some of which are rich in primes, such as Euler's n squared plus n plus 41. The full density questions behind it are tied to the unproven Hardy-Littlewood Conjecture F.
Q: What was the Fermi-Pasta-Ulam-Tsingou problem? A: It was a 1953 computer experiment on a chain of masses connected by slightly nonlinear springs. Researchers expected the energy to spread evenly among vibrational modes. Instead the system kept returning near its starting state. Kruskal and Zabusky later tied this to solitons in 1965, and Mary Tsingou, who programmed the simulation, was named in the title after 2008.
Q: What was Project Orion, and what was Ulam's role? A: Orion was a design for a spacecraft pushed by a series of nuclear explosions behind a pusher plate. Ulam proposed the general idea of nuclear pulse propulsion in 1946, and with C. J. Everett wrote a classified paper on it in August 1955. The project ran into funding problems in 1959, lost political support after the 1963 Partial Test Ban Treaty, and was canceled in 1964.
Sources
- MacTutor History of Mathematics, University of St Andrews: Stanislaw Ulam
- Wikipedia, Stanislaw Ulam (biography, with citations to Ulam's memoir and Rota's 1987 article)
- Wikipedia, Monte Carlo method (history, pi example, convergence)
- Wikipedia, Teller-Ulam design (March 1951 report, Ivy Mike, credit dispute)
- Wikipedia, Fermi-Pasta-Ulam-Tsingou problem
- Wikipedia, Ulam spiral and Project Orion
- Ulam, S.M., Adventures of a Mathematician (Scribner's, 1976)
- Metropolis, N. and Ulam, S., "The Monte Carlo Method," Journal of the American Statistical Association 44(247), 335-341 (1949)
Read next:
- John von Neumann, Ulam's closest long-term collaborator
- Enrico Fermi, his partner in the FPUT experiment
- Niels Bohr, the Danish counterpart to the Los Alamos generation
- Richard Feynman, who worked alongside Ulam at Los Alamos
Follow new articles
Add SciHub101 to your feed reader to see new science and technology stories.
About the Author
SciHub101 is an independent science and technology publication founded and edited by Buğra Sözeri, a software engineer and app developer based in Türkiye. AI tools assist with drafting and research. Source documentation in the existing archive is uneven, and articles are being reviewed and corrected. The Editorial Team byline identifies editorial responsibility; it does not certify specialist or independent human review. Where we compare products, anything built by our own team is labelled as such — see the editorial guidelines and About page for how corrections and disclosures work.
Related Articles

Tim Berners-Lee: How One Proposal at CERN Built the World Wide Web
Tim Berners-Lee invented the World Wide Web at CERN in 1989, released it free to the world in 1993, and has spent 30 years fighting to keep it open and equitable.

Satyendra Nath Bose: The Physicist Behind Every Boson
Satyendra Nath Bose invented a new statistics for quantum particles in 1924, corresponded directly with Einstein, and never won a Nobel Prize. Here is his full story.

Hans Bethe: The Physicist Who Explained Why Stars Shine
Hans Bethe discovered the CNO cycle that powers stars, led Manhattan Project theory, won the 1967 Nobel Prize, and spent 70 years at Cornell advocating against nuclear weapons. He lived to 98.

Enrico Fermi: Father of the Nuclear Age
Enrico Fermi built the first nuclear reactor, won the Nobel Prize, and pioneered estimation science. Discover the life of Italy's greatest modern physicist.

John von Neumann: The Man Who Built the Modern World
John von Neumann invented game theory, defined computer architecture, and shaped the atomic bomb. Meet the mind that colleagues called beyond human.

The Manhattan Project Science Explained
The Manhattan Project combined nuclear physics, engineering, secrecy, and wartime urgency. This article explains the science without glorifying the weapon.