Using a novel computational modeling technique to test theoretical quantum physics, a team of researchers at the Department of Energy’s Oak Ridge National Laboratory have discovered properties of the atomic nucleus at new levels of detail — illuminating behavior of its protons and neutrons amid fundamental forces of nature.
The study has revealed how collective nuclear phenomena, such as the deformation of the nucleus, emerge from the most detailed knowledge available about the strong nuclear force — a fundamental force that holds atomic particles together within the nucleus.
This improved understanding of atomic nuclei emerged after modeling a variety of particle behaviors at differing energy levels, based on methods team members had been eager to test after years of research and development. The team’s findings improve scientists’ ability to predict nuclear properties with unprecedented accuracy. They also highlight the importance of including more details in atomic nuclei modeling efforts. These added details can contribute to better predictive capabilities in a variety of sectors, from energy production to national security.
“We’ve established a systematic approach to performing first-principles computations of the structure of nuclei,” said Zhonghao Sun of Louisiana State University, formerly of ORNL. “The new tools we introduced are truly game changers, allowing us to accurately compute the structure and decays of deformed nuclei from their building blocks, the structure of which lies on the frontier of nuclear science research.”
The team’s findings, published in the journal Physical Review X, advance knowledge of the detailed structure of atomic nuclei and the behaviors of their constituent protons and neutrons. This accomplishment required the stunning computational capacity of ORNL’s Frontier supercomputer, housed at the Oak Ridge Leadership Computing Facility, a DOE Office of Science user facility. Frontier performs at exascale, executing more than a quintillion calculations per second.
Atomic nuclei, the positively charged central portions of atoms, consist of clustered protons and neutrons that make up most of an atom’s mass. They exhibit various phenomena, behaving differently at the multiple energy scales at which they can be observed. These behaviors range from rotations at lower energies to short-range correlations between physical factors at higher energies. The short-range correlations help determine how protons and neutrons bind in the nucleus.
Gaute Hagen and Thomas Papenbrock of ORNL’s Theoretical and Computational Physics group collaborated on the team’s theoretical approach. “Our multiscale model revealed more than what was known of the deep structure of the nucleus,” Hagen said. Using probes at varying energy levels, the team resolved details of the nucleus in finer detail.
“At very low resolution, the nucleus might be viewed as a liquid drop that exhibits collective rotations,” Hagen said. “However, as resolution increases, more details about the nucleus’s internal structure are resolved, and more is learned about how protons and neutrons interact to build up the entire nucleus.”
It is particularly difficult to model the building blocks of atomic structure when scientists are starting from basic propositions, or first principles. Historically, building a unified model of the nucleus that captures a variety of different phenomena — such as small rotational energies and the large binding energy of the nucleus itself — has been a significant challenge.
“This has been a really longstanding challenge in our field,” said Hagen. “Typically, physicists have used existing frameworks to model different phenomena at differing energy scales. First principles theory has typically been very good at describing bulk properties, such as binding energies and charge radii — or measurement of proton distribution, which helps reveal the size of an atomic nucleus — but not so good at describing collective rotational states of the nucleus and the electromagnetic transitions between them.”
Papenbrock added, “Protons and neutrons are bound into atomic nuclei. That [requires] about 8 million electron volts per nucleon, while rotational energies are only a tiny fraction of that number. Getting both of these things right, in both the model and throughout the application itself, has been a real challenge for everyone who computes nuclei.”
With additional input from Andreas Ekström and Christian Forssén of Chalmers Technical University in Sweden, the team’s computational approach was to capture large energies by including short-range correlations of the nuclear state. They computed smaller details in a second step, which included long-range correlations. This step is computationally expensive, requiring leadership-class computing resources such as Frontier.
“Our study focused on relatively light nuclei, and many of these calculations were done on ORNL’s previous supercomputer, Summit,” said Hagen. “Frontier will now allow us to apply the same high-fidelity modeling to heavy nuclei because the computational cost increases with the mass of the atomic nucleus.”
Gustav Jansen of ORNL’s Advanced Computing for Nuclear, Particle, and Astrophysics group customized computations targeting Frontier’s graphics processing units, or GPUs, which accelerate application code at exascale and deliver most of Frontier’s computing power. His contribution enabled Frontier’s GPUs to engage with the Nuclear Tensor Contraction Library, which Jansen had developed for use with leadership-class computations. A tensor contraction is a single, fundamental operation, or a generalization of a multiplication matrix spanning multiple dimensions. The team’s use of this specialized library allowed for a huge variety of computations to occur simultaneously.
“Everything we do is written in terms of tensor contractions,” Jansen said. “It’s important that they’re done fast and efficiently on Frontier but in an architecturally independent way, so the scientific code doesn’t have to change when we go from one machine architecture to the next.” Running so many complex calculations at once demanded a major computing power commitment — approximately 20% of Frontier’s entire computation capacity.
The team also built high-fidelity models of nuclear properties. These computer-based designs are typically used in simulations or prototypes that closely mirror a final product. The models are run on laptops and are considerably less expensive to operate, though they still require the initial use of Frontier for their creation.
The computations revealed that a rare nucleus known as 30-neon has both spherical and deformed, football-like shapes that coexist. By performing millions of model evaluations, the team learned how the strong nuclear force also drives deformation of the nuclear shape. Applying this model to other nuclei will enable the discovery of additional knowledge, “wherever nuclei are used in curiosity-driven science,” said Hagen.
“We’re confident that our accurate and reliable predictions will bring new insights to the study of fundamental nuclear forces and nuclear structure,” said Sun.
The DOE Office of Science (Office of Nuclear Physics and Office of Advanced Scientific Computing Research) supported this research.
UT-Battelle manages ORNL for the DOE Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science. — Chris Driver
“This story is a deep dive of research described at https://www.ornl.gov/news/supercomputing-illuminates-detailed-nuclear-structure.”
This Oak Ridge National Laboratory news article "DEEP DIVE: Exascale computing illuminates detailed structure of atomic nuclei" was originally found on https://www.ornl.gov/news
