Machine Learning-Driven High-Precision Model for $α$-Decay Energy and Half-Life Prediction of superheavy nuclei

TL;DR

This work presents a physics-informed XGBoost framework to predict $Q_\alpha$ and $T_{1/2}$ for superheavy nuclei, integrating nuclear-structure features such as mass number, isospin asymmetry, shell proximity, angular-momentum hindrance, and deformation. It achieves higher predictive accuracy than traditional empirical formulas (Royer and UDL) and provides SHAP-based interpretability that aligns with the quantum tunneling and shell-structure physics behind $\alpha$-decay. The dual-model approach—separately predicting $Q_\alpha$ and $T_{1/2}$ with physically meaningful inputs—ensures robust extrapolation and physicochemical consistency across regions with sparse data. This methodology offers a principled, data-driven tool for exploring $\alpha$-decay systematics and extrapolating half-lives in exotic nuclei, with implications for nuclear structure and element discovery.

Abstract

Based on Extreme Gradient Boosting (XGBoost) framework optimized via Bayesian hyperparameter tuning, we investigated the α-decay energy and half-life of superheavy nuclei. By incorporating key nuclear structural features-including mass number, proton-to-neutron ratio, magic number proximity, and angular momentum transfer-the optimized model captures essential physical mechanisms governing $α$-decay. On the test set, the model achieves significantly lower mean absolute error (MAE) and root mean square error (RMSE) compared to empirical models such as Royer and Budaca, particularly in the low-energy region. SHapley Additive exPlanations (SHAP) analysis confirms these mechanisms are dominated by decay energy, angular momentum barriers, and shell effects. This work establishes a physically consistent, data-driven tool for nuclear property prediction and offers valuable insights into $α$-decay processes from a machine learning perspective.

Figures (3)

• Figure 1: SHAP summary plot showing the feature importance distribution in the decay energy prediction model. The horizontal axis represents SHAP values, indicating the impact of each feature on the model output per sample. Each dot corresponds to a single prediction instance, colored according to the feature value from low (blue) to high (red).
• Figure 2: Scatter plots comparing the predicted and experimental values of $\log_{10}(T_{1/2})$ using the XGBoost model, the Royer empirical formula, and the UDL formula. Each panel shows the correlation between model predictions and experimental data, with the dashed line representing the ideal prediction curve.
• Figure 3: SHAP-based feature importance for the $\alpha$-decay half-life prediction model.