RHINE AI Model Speeds Simulations of Heavy-Element Formation in Neutron Star Mergers

Heavy elements such as gold and platinum are forged through the rapid neutron-capture process in extreme events including neutron star mergers. But simulating the complex nuclear reactions requires enormous computing resources and expense, forcing astrophysicists to simplify their models and creating discrepancies between theoretical predictions and observational data from space. Accurately reconstructing the moments when heavy elements form has long remained a major scientific challenge.

Germany’s GSI Helmholtz Centre for Heavy Ion Research unveiled a new AI model called RHINE on July 18, 2026. Developed with support from the FAIR international facility, which has received total investment of about €3.3 billion, the model uses neural networks to estimate nuclear-reaction heating rates in real time. It keeps prediction errors below 10% while sharply reducing computing costs, marking a key advance in linking ground-based experiments with observations from space.