arXiv subjects

Panpan Qi

Publications and source records attributed to Panpan Qi.

7 recordsLinked to original sources

arXiv2026

Physics-guided residual correction of $\alpha$-decay half-lives based on the effective liquid drop model

To improve the prediction accuracy of $\alpha$-decay half-lives in heavy and superheavy nuclei, a physics-guided residual-correction framework combining the effective liquid drop model (ELDM) with machine-learning methods is proposed. The ELDM is first used as the macroscopic baseline for describing the barrier-penetration process, and XGBoost and TabPFN models are then employed to learn the residual deviations between ELDM predictions and experimental data. To incorporate microscopic nuclear-structure information, several physically motivated descriptors are constructed, including deformation-related quantities, Geiger--Nuttall-related features, and minimum orbital angular momentum. The results show that machine-learning residual correction significantly improves the predictive performance of the ELDM baseline. Among all models, TabPFN-term3 achieves the best accuracy, reducing the RMSE and MAE to 0.348 and 0.248, corresponding to improvements of 38.60\% and 40.46\%, respectively. Residual-distribution and feature-ablation analyses further indicate that the corrected predictions are closer to experimental values and that physically motivated descriptors play an important role in learning nonlinear residual structures. Overall, the proposed ELDM-based residual-correction framework can effectively compensate for missing microscopic nuclear-structure effects while preserving physical interpretability, providing a feasible strategy for high-precision $\alpha$-decay half-life prediction.

nucl-th↗

arXiv2025

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

A hybrid approach combining the Tabular Prior-data Fitted Network (TabPFN) with the Coulomb and Proximity Potential Model (CPPM) is developed to investigate $\alpha$-particle preformation factors $P_{\alpha}$ and their impact on $\alpha$-decay half-lives. The TabPFN model, trained on 498 nuclei, accurately learns the relationship between nuclear structure properties and $P_{\alpha}$, achieving a root mean square deviation of $\sigma_{\mathrm{rms}} = 0.211$. The predicted factors reveal clear odd-even staggering and shell closure effects, and exhibit linear correlations with both $Q_{\alpha}^{-1/2}$ and the fragmentation potential $V_{\mathrm{frag}}$. When incorporated into CPPM calculations, the machine-learning-based $P_{\alpha}$ values significantly improve half-life predictions. Similar improvements are also obtained when deformation effects are included in the potential barrier description. The capability of the model is further demonstrated through predictions for superheavy nuclei ($Z = 117$--120), suggesting $N = 184$ as a potential neutron magic number.

nucl-th↗

arXiv2025

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

Based on Extreme Gradient Boosting (XGBoost) framework optimized via Bayesian hyperparameter tuning, we investigated the {\alpha}-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 $\alpha$-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 $\alpha$-decay processes from a machine learning perspective.

nucl-th↗

arXiv2025

Bayesian optimization and nonlocal effects method for $\alpha$ decay of superheavy nuclei based on CPPM

We combine nonlocal effects with Bayesian Neural Network (BNN) methods to enhance the prediction accuracy of $\alpha$ decay half-lives. The results indicate that accounting for nonlocal effects significantly impacts the half-life calculations, while the BNN method markedly improves prediction accuracy and demonstrates strong extrapolation capabilities. Furthermore, we discuss the impact of nuclear deformation (the quadrupole deformation factor $\beta_2$) on machine learning predictions. Through Shapley Additive Explanations (SHAP), we conducted a quantitative comparison of six input features within the BNN, revealing that the $\alpha$ decay energy $Q_\alpha$ is the primary driving factor affecting the half-life $T_{1/2}$. Leveraging the remarkable extrapolation ability of the BNN, we successfully predicted the $\alpha$ decay half-lives of the isotope chain ($Z=118, 120$), uncovering a significant shell effect at neutron number $N=184$. For the isotopic chains ($Z=118, 120$), the predicted $\alpha$ decay half-lives and $Q_{\alpha}$ values satisfy the Geiger-Nuttall (G-N) linear relationship. This result further confirms the predictive reliability of the proposed model. Keywords: $\alpha$ decay, half-lives, nonlocal effects, Bayesian Neural Network, Coulomb and proximity potential model

nucl-th↗

arXiv2025

Systematic study of {\alpha}-decay half-lives of superheavy nuclei based on Coulomb and proximity potential models with temperature effects

By employing the Coulomb proximity potential model (CPPM) in conjunction with 22 distinct proximity potential models, we investigated the temperature dependence and the effects of proton number and neutron number on the diffusion parameters that determine the {\alpha}-decay half-lives of superheavy nuclei. The results indicate that the Prox.77-3 T-DEP proximity potential model yields the best performance, with the lowest root mean square deviation ({\sigma}=0.515), reflecting a high consistency with experimental data. In contrast, Bass77, AW95, Ngo80, and Guo2013 display larger deviations. The inclusion of temperature dependence significantly improves the accuracy of models such as Prox.77-3, Prox.77-6, and Prox.77-7. The -decay half-lives of 36 potential superheavy nuclei were further predicted using the five most accurate proximity potential models and Ni's empirical formula, with the results aligning well with experimental data. These predictions underscore the high reliability of the CPPM combined with proximity potential models in the theoretical calculation of {\alpha}-decay half-lives of superheavy nuclei, offering valuable theoretical insights for future experimental investigations of superheavy nuclei.

nucl-th↗

arXiv2025

Systematic calculation on alpha decay and cluster radioactivity of superheavy nuclei

In the Coulomb and Proximity Potential Model (CPPM) framework, we have investigated the cluster radioactivity and alpha decay half-lives of superheavy nuclei. We study 22 different versions of proximity potential forms that have been proposed to describe proton radioactivity, two-proton radioactivity, heavy-ion radioactivity, quasi-elastic scattering, fusion reactions, and other applications. The half-lives of cluster radioactivity and alpha decay of 41 atomic nuclei ranging from 221Fr to 244Cm were calculated, and the results indicate that the refined nuclear potential named BW91 is the most suitable proximity potential form for the cluster radioactivity and alpha decay of superheavy nuclei since the root-mean-square (RMS) deviation between the experimental data and the relevant theoretical calculation results is the smallest ({\sigma}= 0.841). By using CPPM, we predicted the half-lives of 20 potential cluster radioactivity and alpha decay candidates. These cluster radioactivities and alpha decays are energetically allowed or observable but not yet quantified in NUBASE2020.

nucl-th↗

arXiv2020

Dynamic Malware Analysis with Feature Engineering and Feature Learning

Dynamic malware analysis executes the program in an isolated environment and monitors its run-time behaviour (e.g. system API calls) for malware detection. This technique has been proven to be effective against various code obfuscation techniques and newly released ("zero-day") malware. However, existing works typically only consider the API name while ignoring the arguments, or require complex feature engineering operations and expert knowledge to process the arguments. In this paper, we propose a novel and low-cost feature extraction approach, and an effective deep neural network architecture for accurate and fast malware detection. Specifically, the feature representation approach utilizes a feature hashing trick to encode the API call arguments associated with the API name. The deep neural network architecture applies multiple Gated-CNNs (convolutional neural networks) to transform the extracted features of each API call. The outputs are further processed through bidirectional LSTM (long-short term memory networks) to learn the sequential correlation among API calls. Experiments show that our solution outperforms baselines significantly on a large real dataset. Valuable insights about feature engineering and architecture design are derived from the ablation study.

cs.CR↗