Ab initio calculations of light nuclei with quantified uncertainties
ORAL
Abstract
In ab initio nuclear structure theory, accurately calculating obervables with quantified uncertainties is essential for an understanding of nuclear properties. I discuss recent results for observables of p-shell nuclei obtained in the No-Core Shell Model (NCSM) using realistic chiral Effective Field Theory (EFT) interactions, with combined numerical and chiral EFT uncertainties.
For the uncertainty quantification of the NCSM calculations I present a comparison of model-space extrapolations for NCSM calculations of energies and radii of Li isotopes, and demonstrate that machine-learning (ML) approaches can provide reliable predictions with robust statistical uncertainties for both energies and radii. The interaction uncertainties can be estimated from order-by-order calculations in the chiral EFT. I present recent NCSM predictions of Boron radii with combined numerical and interaction uncertainties based on two different chiral EFT interactions up to N3LO. An extension to radius differences further allows us to investigate a potential proton halo in Boron-8, and provide predictions that relate directly to isotope shifts, which can be precisely measured in experiments. Finally, I discuss a novel ML method for quadrupole moments, that leverages the correlations between energies, radii, and E2 observables. By learning these correlations from NCSM calculations in accessible model spaces, this extension enables the prediction of converged E2 observables from converged energies and radii.
For the uncertainty quantification of the NCSM calculations I present a comparison of model-space extrapolations for NCSM calculations of energies and radii of Li isotopes, and demonstrate that machine-learning (ML) approaches can provide reliable predictions with robust statistical uncertainties for both energies and radii. The interaction uncertainties can be estimated from order-by-order calculations in the chiral EFT. I present recent NCSM predictions of Boron radii with combined numerical and interaction uncertainties based on two different chiral EFT interactions up to N3LO. An extension to radius differences further allows us to investigate a potential proton halo in Boron-8, and provide predictions that relate directly to isotope shifts, which can be precisely measured in experiments. Finally, I discuss a novel ML method for quadrupole moments, that leverages the correlations between energies, radii, and E2 observables. By learning these correlations from NCSM calculations in accessible model spaces, this extension enables the prediction of converged E2 observables from converged energies and radii.
–
Publication: M. Kn\"oll, M. Lockner, P. Maris, R.J. McCarty, R. Roth, J.P. Vary, T. Wolfgruber,<br>Benchmarking ANN extrapolations of the ground-state energies and radii of Li isotopes,<br>Phys. Rev. C 111, 064304 (2025)<br><br>T. Wolfgruber, T. Gesser, M. Kn\"oll, P. Maris, R. Roth,<br>High-Precision Ab Initio Radius Calculations of Boron Isotopes,<br>Phys. Rev. C 112, 014306 (2025)<br><br>M. Kn\"oll, M.L. Agel, T. Wolfgruber, P. Maris, R. Roth,<br>Machine Learning for Correlations of Electromagnetic Properties in Ab Initio Calculations,<br>arXiv: 2501.08959 [nucl-th], under review
Presenters
-
Pieter Maris
Iowa State University
Authors
-
Pieter Maris
Iowa State University