摘要
利用神经网络模型学习、模拟随机两体系综(TBRE)下的原子核基态自旋分布,并对学习后的模型输入特征进行了分析。这是核物理中利用神经网络模型进行分类的典型应用。研究表明,采用本工作的单隐藏层神经网络模型,精确地描述每个随机相互作用系综内的样本仍比较困难。然而,神经网络模型却能够相对较好地描述基态自旋的统计性质,这可能是因为神经网络模型学习到了TBRE中基态自旋分布的经验规律。
The neural network model is used to learn and simulate the ground state spin distribution of the nucleus under stochastic two-system ensemble(TBRE),and the input characteristics of the learned model are analyzed.This is a typical application of classification using neural network models in nuclear physics.We show that it is still difficult to accurately each the sample within random interaction ensemble using the single hidden layer neural network model in this paper.However,the NN model describes the statistical properties of the ground state spins reasonably well,probably because the NN model learned the empirical law of the ground state spin distribution in TBRE.
作者
刘登
ALAM Noor A
肖越
雷杨
覃珍珍
LIU Deng;ALAM Noor A;XIAO Yue;LEI Yang;QIN Zhenzhen(School of Mathematics and Physics,Southwest University of Science and Technology,Mianyang,621010,Sichuan,China;School of Defense and Technology,Southwest University of Science and Technology,Mianyang,621010,Sichuan,China)
出处
《原子核物理评论》
CAS
CSCD
北大核心
2024年第1期385-395,共11页
Nuclear Physics Review
基金
国家自然科学基金资助项目(12105234)。
关键词
神经网络
随机两体系综
原子核基态自旋
neural network
two-body random ensemble
angular-momentum distribution of nuclear ground state
