DPG Phi
Verhandlungen
Verhandlungen
DPG

Köln 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

HK: Fachverband Physik der Hadronen und Kerne

HK 25: Focus Session II: Accurate Nuclear Charge Radii of Light Elements

HK 25.3: Hauptvortrag

Mittwoch, 12. März 2025, 14:50–15:15, HS 1 Physik

Precision Radii from the No-Core Shell Model via Neural Networks — •Robert Roth — Institut für Kernphysik, TU Darmstadt, Germany

The no-core shell model (NCSM) is one of the most rigorous and universal ab initio methods for light nuclei. It is based on the solution of the many-body problem in a finite model space, characterized by a single truncation parameter. For sufficiently large truncation parameters, observables are guaranteed to converge to the exact solution. The convergence pattern varies for different observables, and radii are particularly challenging due to their sensitivity to the long-range behavior of the wave functions. Therefore, fully converged calculations are only possible for few-nucleon systems. For p-shell nuclei, we obtain NCSM sequences that reveal a convergence pattern, but not a fully converged observable. To overcome this limitation, we have developed artificial neural networks (ANNs) that predict converged energies and radii based on NCSM convergence patterns using large sets of NCSM calculations for few-nucleon systems as training data. We demonstrate the application of the ANN-enhanced NCSM for radii of boron and carbon isotopes with different families of chiral two- plus three-nucleon interactions. In addition to the model-space uncertainties that are extracted from the ANNs in a statistical manner, we use Bayesian methods to asses chiral truncation uncertainties. This approach delivers precise NCSM predictions for radii with fully quantified theory uncertainties. Finally, we explore the generalization of ANN-enhanced NCSM calculations to other electromagnetic observables.

Keywords: ab initio nuclear structure theory; no-core shell model; artificial neural networks; charge radii; chiral EFT interactions

100% | Mobil-Ansicht | English Version | Kontakt/Impressum/Datenschutz
DPG-Physik > DPG-Verhandlungen > 2025 > Köln