80-Year-Old Seeds Gave Rise to Today's Supercomputers

Earlier this spring, Los Alamos National Laboratory brought online its newest supercomputer, Venado. While not the world's fastest supercomputer, Venado holds the distinction of being one of just two worldwide to integrate powerful superchips that use AI technology to accelerate computing speeds, usually at lower cost and power consumption, than earlier chip technology. Some complex computing problems that previously took months will now take only minutes and help speed research on everything from climate science to DNA.

It might surprise people to know that the seeds of this warp-speed computing power were planted back in the Manhattan Project by a Polish-American mathematician named Stanislaw Ulam.

Ulam was born in 1909 to Jewish parents in what is today Ukraine but was then part of Poland. In 1939, he and his younger brother were put on a ship bound for the United States just days before Hitler invaded Poland.

He joined the Manhattan Project in Los Alamos in 1944 and set about pioneering the Lab's earliest computing efforts.

From his calculations that informed the first atomic weapons designs - and later, the design of the H-bomb - to his adaptation of statistical sampling methods for programming the world's first supercomputers, Ulam's work continues to impact Los Alamos's national security mission today.

"As astounded as I'm sure Ulam would have been at the power and speed of Venado, with his remarkable talent for embracing the new and the previously unimaginable, I can imagine him quickly thinking up clever and unexpected uses of such an amazing tool," said National Security Research Center historian Nic Lewis, an expert on the history of Cold War computing and Department of Energy laboratories' role in the evolution of electronic computing technologies.

Ulam worked on hydrodynamic calculations, or calculations that describe the movement of materials, for implosion-type atomic weapons such as Fat Man. He served the Lab from those early days until 1965, when he transitioned to university teaching. Throughout his career, Ulam, who died on May 13, 1984, defied professional categories and made significant contributions to science and mathematics.

An unconventional intelligence

Gifted with an unconventional brilliance, one of his most significant talents, Lewis said, was his ability to find "logical shortcuts to seemingly intractable problems." He was also unique in his ability to inspire others through seemingly casual conversations, as his thought process required interaction with other people. "He was notorious for just walking into someone's office and costing them hours of productive time, but his ideas were always inventive, even if most weren't practical. When his ideas were good, they were very good," said Lewis.

He added that Ulam's collaborative nature imbued his professional and personal life: "He was a social butterfly, had a mischievous sense of humor and was extremely charming. He fit in effortlessly wherever he went."

A problem-solver under pressure: Fat Man and the Monte Carlo method

As a mathematician, Ulam did not enjoy the tedious, number-crunching aspects of his field - a predilection that honed his capacity for developing effective shortcuts to solve complex problems. This capacity turned out to be critical for the Manhattan Project's success.

"His most important Manhattan Project contributions," Lewis said, "were around calculations he performed with (fellow mathematician) John von Neumann on the implosion concept, figuring out the complexities of implosion before the Lab had mechanical computing tools."

This work proved invaluable in 1944 when attempts to develop a gun-type plutonium weapon - on which the Thin Man design was based - proved unsuccessful. Ulam's calculations demonstrated there was a viable alternative to Thin Man. Ulam also advocated for the Lab's early leasing of the essential IBM computing machines, ensuring they were in place when first Lab director J. Robert Oppenheimer reorganized his teams around the production of the implosion-type weapon, which became Fat Man.

The mathematical calculations that made Fat Man possible led to Ulam's later adaptation of statistical sampling for electronic computers. Taking his inspiration from the random characteristics of card games and roulette, Ulam realized it wasn't necessary to calculate every possible outcome of a particular game when you could feed a representative sample into a computer and have it calculate a range of likely outcomes in a fraction of the time.

"The answer might not be exact," Lewis said, "but it would usually be good enough for making decisions." Ulam's longtime Lab collaborator Nicholas Metropolis named this method Monte Carlo, after the famous Monaco gambling location. The method became a foundation for modern computational physics, risk analysis and problem-solving in many other fields.

2024-06-20

Stanislaw Ulam in 1966 with the "Fermiac," a hand-operated computer for studying the motion of neutrons, which was designed by Ulam's colleague Enrico Fermi and partially based on Ulam's Monte Carlo method.

Unlike several of his Los Alamos contemporaries, Ulam's contributions to science and mathematics did not lead to a Nobel Prize or even a nomination for those prestigious awards.

"He had 150 publications to his name," Lewis said, "but he tended to flitter around topics that interested him at a very high, theoretical level, rather than delving deeply into an idea for any length of time. Nobels tend to go toward groundbreaking, extended studies where the impact of the discovery is readily apparent. Mathematics in general isn't easy to fit into that mold, and Ulam's style of work was even less likely to be considered for a nomination."

However, in addition to his work on the Manhattan Project, Monte Carlo and H-bomb, Ulam made dozens of other contributions to mathematics and science throughout his career. He conducted virtual experiments that started the field of non-linear studies - methods of understanding complex and seemingly random systems and phenomena - which contributed to the founding of the Lab's Center for Non-Linear Studies in the early 1980s, shortly before his death.

"In a very real sense," Lewis said, "advancements like the Venado supercomputer are a continuation of Ulam's legacy of finding innovative applications of computing technologies to answer the Lab's most important scientific questions."

/Public Release. This material from the originating organization/author(s) might be of the point-in-time nature, and edited for clarity, style and length. Mirage.News does not take institutional positions or sides, and all views, positions, and conclusions expressed herein are solely those of the author(s).View in full here.