AI Improves Simulations of Heavy Element Formation

Many of the chemical elements found throughout the universe are forged during extreme cosmic events, including supernova explosions and neutron star mergers. These enormous explosions generate the energy needed to produce heavy atomic nuclei through a process known as rapid neutron capture, or the r-process.

During the r-process, atomic nuclei rapidly absorb free neutrons. Some of those neutrons then transform into protons, allowing the nuclei to grow larger and eventually form many of the heavy elements found in nature.

Simulating these reactions is one of the biggest challenges in nuclear astrophysics because the calculations require tremendous computing power.

"Researchers around the world strive to make these complex reactions understandable through theoretical simulations. However, modeling all parameters requires incredible computing power, which is why the models often have to be simplified," said Dr. Oliver Just, first author of the study and a researcher in the "Nuclear Astrophysics & Structure" department at GSI/FAIR. "Our new model RHINE, which uses artificial intelligence, offers an efficient alternative."

Deep Learning Speeds Up Complex Nuclear Calculations

The new system, called RHINE (r-process heating implementation in hydrodynamic simulations with neural networks), relies on machine learning (ML), specifically a deep learning neural network, to estimate how much energy is released during nuclear reactions in the r-process while hydrodynamic simulations are running.

This energy release, often called heating, plays an important role in determining how matter is expelled during stellar explosions. It can influence both the speed of the ejected material and the light produced afterward. In neutron star mergers, that brilliant glow is observed as a kilonova.

Instead of performing every nuclear calculation during each simulation, the AI is first trained using an extensive library of reference calculations that include complete nuclear reaction networks. Once trained, it can accurately estimate the heating rates with only a fraction of the computational effort.

"First the ML models are trained using a large number of reference calculations produced with a full set of nuclear reactions. Subsequently, the models are adopted in running hydrodynamical simulations to approximate the heating rates during the r-process with minimal effort," explained Dr. Zewei Xiong, also a scientist in GSI/FAIR's "Nuclear Astrophysics & Structure" department and a key developer of the machine learning models.

"With detailed comparisons, we validated our ML scheme against reference data. The high degree of agreement suggests that the use of ML models can save a tremendous amount of computing time. We also deduced from the results that r-process heating is an important effect that should be better accounted for in future modeling."

Connecting Future Experiments With Cosmic Observations

The researchers say RHINE could enable much more detailed simulations in the future while dramatically reducing the computing resources required. Those improved models may eventually help connect experiments at the upcoming FAIR research facility with observations of stellar explosions and neutron star mergers made by astronomers.

The RHINE source code has been made publicly available so other researchers can build on the work. The project was co-funded, among other organizations, by the European Research Council (ERC).