Laboratory Science Entwined with Rise in Computing
LIVERMORE scientists have been using computer simulations to attain breakthroughs in science and technology since the Laboratory’s founding. High-performance computing remains one of the Laboratory’s great strengths and will continue to be an important part of future research efforts.
To meet our programmatic goals, we demand ever more powerful computers from industry and work to make them practical production machines. We develop system software, data management and visualization tools, and applications such as physics simulations to get the most out of these machines. High-performance computing, theoretical studies, and experiments have always been partners in Livermore’s remarkable accomplishments.
The Laboratory’s cofounders, Ernest O. Lawrence and Edward Teller, along with Herbert York, the first director, recognized the essential role of high-performance computing to meet the national security challenge of nuclear weapons design and development. Electronic computing topped their list of basic requirements in planning for the new Laboratory in the summer of 1952. The most modern machine of the day, the Univac, was ordered at Teller’s request before the Laboratory opened its doors in September. The first major construction project at the site was a new building with air conditioning to house Univac serial number 5, which arrived in January 1953.
Edward Teller, whose centennial we are celebrating this year, greatly appreciated the importance of electronic computing. His thinking was guided by his interactions with John von Neumann, an important pioneer of computer science, and his prior experiences using “human computers” for arduous calculations. Teller was attracted to and solved problems that posed computational challenges—the most famous being his collaborative work on the Metropolis algorithm, a technique that is essential for making statistical mechanics calculations computationally feasible. His work demonstrated his deeply held belief that the best science develops in concert with applications.
This heritage of mission-directed high-performance computing is as strong as ever at Livermore. Through the National Nuclear Security Administration’s (NNSA’s) Advanced Simulation and Computing Program, two of the world’s four fastest supercomputers are located at Livermore, and they are being used by scientists and engineers at all three NNSA laboratories. The prestigious Gordon Bell Prize for Peak Performance was won in 2005 and 2006 by simulations run on BlueGene/L, a machine that has 131,072 processors and clocks an astonishing 280 trillion floating-point operations per second. Both prize-winning simulations modeled physics at the nanoscale to gain fundamental insights about material behavior that are important to stockpile stewardship and many other programs at the Laboratory.
The article entitled A Quantum Contribution to Technolgoy features Livermore-designed computer simulations that focus on the nanoscale beginning with first principles: the laws of quantum mechanics. The use of large-scale simulations to solve quantum mechanics problems was pioneered in 1980 by Livermore scientist Bernie Alder in collaboration with David Ceperley from Lawrence Berkeley National Laboratory.
To predict how materials will respond under different conditions, scientists need accurate descriptions of the interactions between individual atoms and electrons: how they move, how they form bonds, and how those bonds break. These quantum molecular dynamics calculations are extremely demanding. Even with the Laboratory’s largest machines, computational scientists, such as those in Livermore’s Quantum Simulations Group, must design clever modeling techniques to make the run times feasible (hours to days) for simulating perhaps only 1 picosecond of time (a trillionth of a second).
Outstanding science and technological applications go hand-in-hand in this work. As described in the article, our scientists are using quantum simulations to evaluate nanomaterials to reduce the size of gamma-ray detectors for homeland security, provide improved cooling systems for military applications, and help design even smaller computer chips. Yet another quantum simulation project is examining materials to improve hydrogen storage for future transportation.
These examples merely scratch the surface of the novel uses for nanotechnologies that scientists can explore through simulations. One can only imagine what possibilities might be uncovered in the future as computational power continues to increase and researchers become ever more proficient in nanoscale simulations and engineering. True to its heritage, Livermore will be at the forefront of this nascent revolution.