METAL AI -- Machine learning for heavy element energy levels and emission intensities

The heaviest chemical elements, i.e. rare earths, lanthanides, and actinides, are made in the extreme physics of binary neutron star collisions. Astronomical observations of the aftermath of these collisions are a sensitive probe of high-energy particle interactions and fundamental forces. But we face a roadblock in trying to explore this extreme physics phenomenon; we need good atomic data for the heavy elements, but these do not exist in the quantity or quality we need. In this collaboration, we explore machine learning methods for the prediction of the atomic structures of these complex elements. With these data, we will be able to model the spectra of neutron star mergers from world-class observatories. We have access to the best observational data, calculation codes, and new experimental laser-plasma data. If the technique is successful, we should be able to develop methods to rapidly calculate accurate atomic data for all isotopes and explore a framework to calculate atomic data for elements that cannot be measured in the lab, such as rare and radioactive species, and ultra-heavy elements, that do not exist on earth, but could be created in these extreme environments.