Systematic Study on the α \alpha -particle preformation factor in the theory of α \alpha -decay based on the Tabular Prior-data Fitted Network (TabPFN)

Panpan Qi, Xuanpeng Xiao, Gongming Yu, Haitao Yang, Qiang Hu · 2025

Read Systematic Study on the α \alpha -particle preformation factor in the theory of α \alpha -decay based on the Tabular Prior-data Fitted Network (TabPFN) by Panpan Qi, Xuanpeng Xiao, Gongming Yu, Haitao Yang, Qiang Hu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-14705.txt.

Introduction

Panpan Qi, Xuanpeng Xiao, and three co-authors combine a machine learning model with a classic nuclear model to predict how readily an alpha particle forms inside a heavy nucleus. That readiness, the preformation factor, strongly affects how fast the nucleus decays by alpha emission. The learning model was trained on nearly five hundred nuclei.

The predicted factors show clear odd-even patterns and dips at closed nuclear shells, as physics would suggest. They also line up with simple quantities tied to decay energy and fragmentation. Feeding the learned factors into the classic potential model sharpens half-life predictions a lot. The hybrid route offers a practical tool for estimating unknown decay rates.

Catalog record

CREATORPanpan Qi, Xuanpeng Xiao, Gongming Yu, Haitao Yang, Qiang Hu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-14705

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Systematic Study on the α \alpha -particle preformation factor in the theory of α \alpha -decay based on the Tabular Prior-data Fitted Network (TabPFN) Preprint: APS/123-QED Panpan Qi, Xuanpeng Xiao, Gongming Yu, Haitao Yang, Qiang Hu Abstract A hybrid approach combining the Tabular Prior-data Fitted Network (TabPFN) with the Coulomb and Proximity Potential Model (CPPM) is developed to investigate α \alpha -particle preformation factors P α P_{\alpha} and their impact on α \alpha -decay half-lives. The TabPFN model, trained on 498 nuclei, accurately learns the relationship between the properties of the nuclear structure and P α P_{\alpha} , achieving a root mean square deviation of σ rms = 0.211 \sigma_{\mathrm{rms}}=0.211 . The predicted factors reveal clear odd-even staggering and shell closure effects, and exhibit a linear correlation with Q α − 1 / 2 Q_{\alpha}^{-1/2} , extending the Geiger-Nuttall systematics. When incorporated into CPPM calculations, the machine learning-based P α P_{\alpha} values significantly improve half-life predictions. The capability of the model is demonstrated through predictions for superheavy nuclei ( Z = 117 Z=117 –120), suggesting N = 184 N=184 as a potential neutron magic number. Usage Secondary publications and information retrieval purposes. Structure You may use the description environment to structure your abstract; use the optional argument of the \item command to give the category of each item. I Introduction The phenomenon of \alpha -decay was first discovered by Rutherford in 1899. In 1928, Gamow and Condon theoretically established that \alpha -decay originates from the quantum tunneling effect 20 ; 24 . Since then, \alpha -decay has been recognized as a pivotal tool for investigating the structure of unstable atomic nuclei, offering critical insights into their internal dynamics. Consequently, the development of a robust theoretical framework for \alpha -decay is of paramount importance. Advances in \alpha -decay theory have primarily addressed aspects including potential barrier profiles, decay energies 31 ; 45 , nuclear medium effects 14 , nuclear deformation 23 , and the \alpha -particle preformation factor 67 ; 12 . Research on \alpha -decay theory has yielded numerous achievements, such as empirical formulas 57 ; 49 ; 51 ; 4 ; 47 , macro-microscopic models, and microscopic approaches 51 ; 50 ; 52 ; 32 ; 56 ; 16 ; 18 ; 62 . In recent years, machine learning (ML) algorithms have emerged as prominent alternative tools for investigating and predicting complex physical data, owing to their strong performance and increasingly vital role in this domain 37 ; 54 . In nuclear physics, ML techniques have been successfully applied to predict nuclear masses 7 ; 60 ; 61 , binding energies 44 , charge radii 59 , \beta -decay lifetimes 10 , and \alpha -decay rates 53 . In studies of heavy nuclei and density functional theory (DFT), ML facilitates the calibration of energy density functional (EDF) parameters 3 ; 21 and enables energy predictions for over a thousand nuclides with low computational cost 36 . In investigations of nuclear properties, ML is employed to assess model-related uncertainties in multi-model frameworks 11 ; in nuclear reaction and neutron star studies, research focuses on the equation of state (EoS) of nuclear matter 39 ; 40 . Deep learning techniques have been employed in studies of \alpha -decay half-lives 66 ; 55 ; 29 ; 30 . In Ref. 66 , an artificial neural network (ANN) was utilized to compute \alpha -decay half-lives; incorporating the angular momentum carried by \alpha particles and the quadrupole deformation parameters of parent nuclei reduced the root-mean-square deviation (RMSD) between predictions and experimental data from 0.581 to 0.334. In Ref. 55 , the XGBoost algorithm was adopted to predict \alpha -decay half-lives of superheavy nuclei, demonstrating the smallest mean square deviation compared to empirical formulas and experimental data. In Ref. 29 , the radial basis function (RBF) kernel support vector machine (SVM) was applied to predict nuclear \alpha -decay half-lives; analysis of 2232 data points showed excellent performance, with root-mean-square errors (RMSE) of 0.819 and 0.352 on two datasets, comparable to other ML methods. In Ref. 30 , a Bayesian neural network (BNN) combined with a complex \alpha -decay model was proposed, leading to improved half-life predictions through modifications to the preformation factor. Within the framework of \alpha -decay theory, the \alpha -particle preformation factor characterizes the probability of \alpha -cluster formation on the surface of the parent nucleus. Typically, P_{\alpha} is treated as an empirical constant, based on its smooth variation in open-shell regions 26 ; however, this assumption may introduce inaccuracies in reproducing half-lives for nuclides near closed-shell regions. In closed-shell nuclides (e.g., those with proton number Z=82 and neutron number N=126 ), P_{\alpha} exhibits significant variations 1 ; 13 , potentially leading to systematic deviations between theoretical and experimental \alpha -decay half-lives. Recent work by Zhang and Deng 15 demonstrated progress by extracting preformation factors via a semi-empirical formula, enhancing consistency between predictions and observations. The discrepancy between theoretical and experimental half-lives primarily stems from the treatment of P_{\alpha} , and given ML’s unique capability to address discrepancies between datasets and extract physical insights, its potential for investigating \alpha -particle preformation factors warrants further exploration. This work investigates the capability of the Tabular Prior-data Fitted Network (TabPFN) to extract \alpha -particle preformation factors, aiming to enhance predictions of \alpha -decay half-lives. Section II outlines the theoretical framework, describing the extraction of experimental preformation factors via the Coulomb and Proximity Potential Model (CPPM) and the TabPFN methodology. In Section III, we demonstrate that the TabPFN model effectively learns complex relationships between preformation factors and nuclear structure properties, capturing key features such as odd-even staggering and shell closure effects, and revealing a linear correlation with Q_{\alpha}^{-1/2} . The model’s extrapolation capability is validated against 41 newly evaluated nuclei, significantly improving the accuracy of predicted half-lives. Finally, the model is employed to predict preformation factors and half-lives for superheavy nuclei ( Z=117-120 ), suggesting N=184 as a potential neutron magic number. A summary is provided in Section IV. II THEORETICAL FRAMEWORK II.1 The extraction of experimental preformation factors P_{\alpha} exp under the framework of CPPM The \alpha -decay half-life of atomic nuclei is given by 48 : \displaystyle T_{1/2}=\frac{\ln 2}{\lambda}, (1) where \lambda denotes the decay constant, expressed as \displaystyle\lambda=P_{\alpha}\nu P. (2) Here, P_{\alpha} , \nu , and P represent the \alpha -particle preformation factor, assault frequency, and barrier penetration probability, respectively. This study begins by examining the relationship between the \alpha -decay constant \lambda and the half-life T_{1/2} . Combining Eqs. (1) and (2) with the experimentally measured half-life T_{1/2}^{\text{exp}} yields \displaystyle\lambda_{\text{exp}}=\frac{\ln 2}{T_{1/2}^{\text{exp}}}=P_{\alpha}^{\text{exp}}\nu P, (3) \displaystyle\lambda_{\text{cal}}=\frac{\ln 2}{T_{1/2}^{\text{cal}}}=P_{\alpha}^{\text{cal}}\nu P. (4) Within the unified decay theory framework, setting the theoretical preformation factor P_{\alpha}^{\text{cal}} to unity allows extraction of the experimental preformation factor as: \displaystyle P_{\alpha}^{\text{exp}}=\frac{T_{\alpha}^{\text{cal}}}{T_{\alpha}^{\text{exp}}}. (5) The Coulomb and Proximity Potential Model (CPPM) is employed as the unified decay theory framework, owing to its accurate description of nuclear fusion, fission, and particularly \alpha decay. The assault frequency \nu is set to 1.0\times 10^{22},\text{s}^{-1} 46 . The penetration probability P is computed using the semiclassical Wentzel-Kramers-Brillouin (WKB) approximation as \displaystyle P=-\frac{2}{\hbar}\int_{R_{\text{in}}}^{R_{\text{out}}}\sqrt{2\mu\left(V(r)-Q_{\alpha}\right)},dr. (6) Here, \hbar denotes the reduced Planck constant, \mu=\frac{m_{d}m_{\alpha}}{m_{d}+m_{\alpha}} is the reduced mass of the \alpha -daughter system, with m_{d} and m_{\alpha} representing the masses of the

Read the complete text