Berni Julian Alder, theoretical physicist and inventor of molecular dynamics, 1925–2020

Berni Julian Alder, one of the leading figures in the invention of molecular dynamics simulations used for a wide array of problems in physics and chemistry, died on September 7, 2020. His career, spanning more than 65 years, transformed statistical mechanics, many body physics, the study of chemistry, and the microscopic dynamics of fluids, by making atomistic computational simulation (in parallel with traditional theory and experiment) a new pathway to unexpected discoveries. Among his many honors, the CECAM prize, recognizing exceptional contributions to the simulation of the microscopic properties of matter, is named for him. He was awarded the National Medal of Science by President Obama in 2008.

Berni Julian Alder in 2015. Image credit: Lawrence Livermore National Laboratory.

Alder was born to Ludwig Adler and Otillie née Gottschalk in Duisburg, Germany on September 9, 1925. Alder’s father Ludwig was a chemist who worked in the German aluminum industry. When the Nazis came to power in 1933, Alder, his parents, elder brother Henry, and twin brother Charles fled to Zurich, Switzerland. In 1941, they further emigrated to the United States (becoming “Alders” in the process), where they settled in Berkeley, California. From then on, he and his family lived and worked in the Bay Area, which they considered their “slice of heaven.” Alder completed his senior year of high school there and then did his undergraduate studies at the University of California, Berkeley. Alder’s education was interrupted by his service in the US Navy as a radar technician in the Pacific Theater. Later, he and his wife Esther raised their two sons and a daughter in the Bay Area community of El Cerrito.

As a Berkeley undergraduate, Alder’s mentor was the great chemist Joel Hildebrand, who influenced his early thinking about chemical systems. Later, circa 1951, as a student of J. G. Kirkwood at the California Institute of Technology, Alder began to explore the idea of Monte Carlo sampling applied to atomistic systems. This early work brought him to the attention of Edward Teller who, with Nicolas Metropolis, Marshall Rosenbluth, Arianna Rosenbluth, and Augusta Teller, had invented the famous “Metropolis” Monte Carlo sampling algorithm that was to revolutionize statistical physics (1). Teller recruited Alder to the just-founded Lawrence Radiation Laboratory (later, the Lawrence Livermore National Laboratory), where he, along with other young talented scientists, built a unique “can do” scientific culture that thrives to this day.

Beginning in the mid-1950s, Alder, who possessed a striking, intuitive understanding of many body statistical systems, began—with his collaborators—a series of remarkable numerical simulations of a simple model system, a collection of hard spheres. These carefully chosen simulations, though at first seemingly oversimplified, repeatedly got to the heart of key physical questions. This novel reliance on numerical simulation in research was natural at Livermore, which had embraced the use of advanced computing from its founding days. Because his calculations involved simple classic dynamics that could be paused and restarted at any stage with modest memory overhead, Alder was able to develop a very effective system of running jobs in the background (termed “free standby” decades ago) that accumulated results at statistical precisions that were famously decades ahead of their time.

For nearly 70 years, Alder applied an inquisitive, open minded, and dauntless research method, based on a kind of Socratic dialogue with his research group, that led them to several important discoveries in statistical mechanics. It is now a commonplace truism that computational simulations have become a third pillar of the scientific method (along with theory and experiment). However, it can also be said that true discoveries that changed our fundamental understanding of nature, originating in such simulations, have remained quite infrequent. As the theorist Leo Kadanoff once remarked, there have been only a few big discoveries in physics driven by simulation since the late 1940s. Alder and his colleague Tom Wainwright were arguably responsible for several of these: Discovering in 1957 that a liquid–solid (freezing) phase transition could occur in system with only repulsive interactions (2), the even more remarkable result in 1962 that that a two-dimensional (2D) system with short-range forces could have a subtle ordering phase transition (3–5), and the 1967 to 1970 discovery (6) that nonequilibrium fluids relax to equilibrium far more slowly than previously thought.

Alder and Wainwright’s extraordinary 1962 result that a 2D system of hard disks with short-range interactions could indeed have a freezing/melting phase transition, was remarkable in two ways.

Berni Alder (standing) with his collaborators, Mary Ann Mansigh and Tom Wainwright, in 1962 with their computational results showing the 2D melting phase transition. Image credit: Lawrence Livermore National Laboratory.

First, the result flatly contradicted powerful arguments going back to Landau and Peierls that there couldn’t be a true, long-range ordered state in a 2D system with continuous symmetry. This contradiction was an essential motivator of Kosterlitz and Thouless’s 1973 theory (K-T) of phase transitions in 2D systems mediated by topological, vortex-like defects (4, 5). Second, the basic vortex-mediated K-T phase transition mechanism and its elaboration to include further intermediate “hexatic phases” (7, 8) are all continuous phase transitions. However, in their 1962 simulations, Alder and Wainwright found clear numerical evidence for a first-order transition to a fluid. Their simulation data, obtained from following 870 disks for ∼10⁷ collisions on the Livermore LARC computer, still stands up well after 50 years when compared with the current state-of-the-art simulations featuring ∼10⁶ disks, definitively confirming the first-order transition (9).

Having successfully resolved important questions with hard spheres, in the 1970s Alder began investigating the possibilities for simulating many-body systems at the more fundamental quantum level. The work came to fruition in 1980 with the definitive calculation of the simplest model of interacting electrons, namely the electron gas (10), which crucially enabled the success of the density functional method in modern computational physics, chemistry, and materials science. Alder’s collaborators went on to develop methods to treat quantum systems at nonzero temperature, including the phenomena of Bose condensation and superfluidity (11).

At Livermore, from the 1950s onward, Alder played a major role in building up the laboratory’s program in materials equations of state, a major need of the laboratory’s defense mission. In 1963, with Teller and others, Alder cofounded the University of California, Davis Department of Applied Physics, which focused on training graduate students in areas relevant to the laboratory’s programs, such as plasma and high-pressure physics, radiation transfer, and laser fusion. Alder was also one of the founders and the editor of the Journal of Computational Physics.

Alder mentored many successful students and post doctorates over the decades, and it was always a special treat to join with him and his group at their lunchtime meeting in the Livermore cafeteria. He always conducted these discussions in a friendly manner, while still thoughtfully questioning assumptions and conclusions. Whether one was a beginning graduate student, a senior scientist, or anything in between, it didn’t matter. One always came away from these dialogs with a deepened understanding of physics.