arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.09558nucl-th
收藏

使用深度神经网络全局预测核电荷密度分布

Global prediction of nuclear charge density distributions using deep neural network

Tian Shuai Shang, Hui Hui Xie, Jian Li, Haozhao Liang

AI总结:

本文提出融合相对论密度泛函理论数据与1014个核实验电荷半径的DNN,以全局预测核电荷密度分布,并在验证集和最新实验数据上验证其精度。

摘要:

本研究开发了一种深度神经网络(DNN),用于生成核电荷密度分布;该网络利用来自相对论密度泛函理论的训练数据,并将1014个核体现有实验电荷半径纳入损失函数。该DNN在验证集上的电荷半径均方根(rms)偏差达到0.0193 fm。此外,DNN能够改进对电荷密度尾部和中心区域的描述,增强与实验结果的一致性。该模型的预测能力还通过其与近期电荷半径实验数据的吻合得到进一步验证。最后,这一经过改进的模型被用于预测更广泛核素图范围内的电荷密度分布,并给出参数化电荷密度、电荷半径以及电荷密度分布的高阶矩,为未来实验研究提供可靠参考。

英文摘要:

A deep neural network (DNN) has been developed to generate the distributions of nuclear charge density, utilizing the training data from the relativistic density functional theory and incorporating available experimental charge radii of 1014 nuclei into the loss function. The DNN achieved a root-mean-square (rms) deviation of 0.0193 fm for charge radii on its validation set. Furthermore, the DNN can improve the description in both the tail and central regions of the charge density, enhancing agreement with experimental findings. The model's predictive capability has been further validated by its agreement with recent experimental data on charge radii. Finally, this refined model is employed to predict the charge density distributions in a wider range of nuclide chart, and the parameterized charge densities, charge radii, and higher-order moments of charge density distributions are given, providing a robust reference for future experimental investigations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑