Using a novel simulation model based on machine learning, an international research team at GSI/FAIR has succeeded in gaining a deeper understanding of element formation in stellar events such as neutron star mergers. For the first time, the scientists used deep learning with a neural network to model the energy release during r -process nucleosynthesis in hydrodynamic simulations. The results are published in the journal Physical Review D.
Many of the chemical elements we know are created in massive stellar events such as the explosions of stars or neutron star mergers. These events release incredible amounts of energy which allow for the production of heavy nuclides. One key nuclear production process is the so-called rapid neutron-capture ( r -process), in which free neutrons are captured by existing nuclei and convert into protons — thus creating larger, heavier atomic nuclei.
“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,” says Dr. Oliver Just, first author of the publication and researcher in the department “Nuclear Astrophysics & Structure” at GSI/FAIR. “Our new model RHINE, which uses artificial intelligence, offers an efficient alternative.”
RHINE ( r -process h eating i mplementation in hydrodynamic simulations with ne ural networks) uses machine learning (ML) — specifically, a neural network based on deep learning — to describe the energy release from nuclear reactions in the r -process in hydrodynamic simulations of the events. This “heating” could have a significant impact on the dynamics and velocity distribution of the material ejected by the explosion and thus also on the electromagnetic radiation which, in the case of neutron star mergers, can be observed as a so-called kilonova.
“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,” Dr. Zewei Xiong explains the method. He is also a scientist in GSI/FAIR’s department “Nuclear Astrophysics & Structure” and played a key role in the design of the ML 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.”
In the future, thanks to the use of the new RHINE model, more detailed simulations could be conducted which could directly link the results of experiments at the future FAIR facility with observations of stellar explosions and neutron star mergers.
The RHINE source code is publicly available for use. Among others, the project was co-funded by the European Research Council (ERC).
Physical Review D
Computational simulation/modeling
𝑟 -process heating implementation in hydrodynamic simulations with neural networks
16-Apr-2026