Learning Nuclear Structure with AI: Radii and Collectivity

Abstract

Low-energy nuclear structure is encoded in a broad body of experimental information across the chart of nuclides. Learning how this information is organized across observables and nuclei can provide a data-driven empirical baseline for theoretical extrapolations and experimental design. Here, we develop held-out ensembles based on NuCLR (Nuclear Co-Learned Representations), a multi-task model of nuclear data, to study charge radii and electric-quadrupole transition strengths. Out-of-fold (OOF) validation shows that shared representation improves performance over single-task learning, yielding a charge-radius RMS deviation of 0.0147 fm and a B(E2) RMS deviation of 0.192 e²b² across hundreds of nuclides, competitive with state-of-the-art nuclear models. Our error bars estimate the expected prediction accuracy across the nuclear chart, highlighting regions where new data would encode information beyond the learned patterns. NuCLR thus serves as a data-driven surveyor of nuclear structure and a step toward a shared, multi-observable foundation model of the nuclear chart.

Nuclear moments, charge radii, and magnetization distribution parameters of Ag isotopes from laser spectroscopy and \textit{ab initio} electronic-structure calculations
Leonid V. Skripnikov, Bram van den Borne, Michail Athanasakis-Kaklamanakis +2
arXiv30 Sep 2026PreprintarXiv licenceFull textPhysical chemistry Atomic, molecular & optical physicsarXiv
Probing collectivity in nuclear binding energies from the viewpoint of the F-spin formalism
H. Emami, H. Sabri
Scientific Reports15 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Full textNuclear physicsCrossref
Ab Initio Emergence and Collapse of Nuclear Collectivity near N = Z = 40
X. C. Cao, C. F. Jiao, R. Z. Hu
arXiv15 Sep 2026PreprintCC BY 4.0Full textNuclear physicsarXiv
Collapse of Nuclear Collectivity along the N = Z Line
M. A. Bentley, R. Taniuchi, R. Wadsworth +29
Physical Review Letters17 Oct 2026Peer-reviewedCC BY 4.0Full text2 citedNuclear physicsHAL
Nuclear Charge Radii from Precision Bound-Electron g-Factor Measurements
Bastian Sikora
arXiv1 Nov 2026PreprintCC BY 4.0Full textAtomic, molecular & optical physicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{giacalone2026learning,
  title = {{Learning Nuclear Structure with AI: Radii and Collectivity}},
  author = {Giuliano Giacalone and Sokratis Trifinopoulos and Mike Williams},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.17838},
  eprint = {2609.17838},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.17838},
}

RIS

TY  - UNPB
TI  - Learning Nuclear Structure with AI: Radii and Collectivity
AU  - Giuliano Giacalone
AU  - Sokratis Trifinopoulos
AU  - Mike Williams
PY  - 2026
DA  - 2026/09/15
PB  - arXiv
DO  - 10.48550/arxiv.2609.17838
UR  - https://arxiv.org/abs/2609.17838
AB  - Low-energy nuclear structure is encoded in a broad body of experimental information across the chart of nuclides. Learning how this information is organized across observables and nuclei can provide a data-driven empirical baseline for theoretical extrapolations and experimental design. Here, we develop held-out ensembles based on NuCLR (Nuclear Co-Learned Representations), a multi-task model of nuclear data, to study charge radii and electric-quadrupole transition strengths. Out-of-fold (OOF) validation shows that shared representation improves performance over single-task learning, yielding a charge-radius RMS deviation of 0.0147 fm and a B(E2) RMS deviation of 0.192 e^2b^2 across hundreds of nuclides, competitive with state-of-the-art nuclear models. Our error bars estimate the expected prediction accuracy across the nuclear chart, highlighting regions where new data would encode information beyond the learned patterns. NuCLR thus serves as a data-driven surveyor of nuclear structure and a step toward a shared, multi-observable foundation model of the nuclear chart.
ER  -

Abstract

Related

Cite