高效选取线性独立的原子特征构建精确机器学习势函数

机器学习势由于具有与第一性原理计算相当的准确性,且低得多的计算成本,在原子模拟中极具前景. 然而原子机器学习势的可靠性、速度和可迁移性在很大程度上取决于原子构型的表示. 适当地选取用作机器学习程序输入的描述符是一个成功的机器学习表示的关键. 本文发展了一种简单有效的方法,可以基于训练数据固有的相关性,从大量待选的描述符中自动选取一组最佳的线性独立原子特征. 通过对几个具有较少冗余线性独立嵌入密度描述符的基准分子构建嵌入原子神经网络势的应用,证明了这种新方法的有效性和准确性. 该算法可以大大简化原子特征的初始选取,并极大地提高原子机器学习势的性能.     

有关大学物理微积分应用的若干问题

分析大学物理教学中的微积分思想方法及其一些重要概念,说明了用微积分解决物理问题的一般步骤,以及如何选取微元及在微元内如何建模,并举例说明了利用微积分的知识解决几类较为重要的物理问题.     

New ab initio Potential Energy Surfaces for Cl(2P3/2, 2P1/2)+H2 Reaction

采用多参考组态相互作用的计算方法和很大的基组构造了Cl+H₂反应体系的一组新的三维从头算势能面. 该势能面包含Cl+H₂体系的能量最低的三个绝热态，并在非绝热近似下转化为四个非绝热势能面. 另外，旋轨耦合矩阵元也基于Breit-Pauli Hamil-tonian计算得到. 对角化这四个非绝热势能和两个旋轨耦合矩阵元组成的全耦合Hamiltonian，得到了三个考虑旋轨耦合后的绝热势能面.基于这组新势能面的非绝热动力学计算结果与最新的实验符合得很好，很好地解释了     

人体微量元素硒营养

硒是人体必需的微量元素。环境系统中的硒含量差异较大，缺硒现象普遍。利用天然食品摄取硒营养是安全，有效，可行的途径。     

HOBr/DOBr态到态光解动力学的量子波包研究

本文基于高水平的多参考组态相互作用计算，报导了HOBr的最低三个单重态的三维势能面. 并且对该分子的光解离过程用实波包传播的方法进行了量子动力学研究，不仅计算结果得到了吸收光谱，而且还产生了光解产物的内态和角度分布. 结果与在紫外区域测得的HOBr的总吸收截面定量吻合，并很好的重现了实验上在266 nm处观察到的振动冷和转动热的光解OH/OD产物. 此外，预测了OH/OD的反冲各向异性参数接近平行跃迁的极限值，表明在266 nm处从基态(1¹A'')到2¹A''态的面内跃迁是一个快速的解离过程. 这与实验上从测量转动取向得出的结论一致. 然而，想要实现与实验上的定量一致，还需要考虑自旋和电子角动量的影响.     

小分子光解动力学的理论研究

光解过程是化学中的核心问题之一。量子态分辨的光解动力学可以使人们在原子与分子的层次上深刻理解光解反应的机制。态-态水平的光解动力学在过去四十年中取得了长足的进步,实验和理论的相互结合极大地促进了我们对光解反应本质的认识。本文综述了小分子态-态光解动力学的理论研究进展,总结了H2O和CH3I这两个最具代表性体系的态-态光解动力学研究成果,并提出了该领域未来面临的挑战。     

A Multiple Scattering Theory for Proton Penetration

We extend the electron small-angle multiple scattering theory to proton penetration. After introducing the concept of narrow energy spectra,the proton energy loss process is included in the proton deep penetration theory. It precisely describes the whole process of proton penetration.Compared to the Monte Carlo method,this method maintains the comparable precision and possesses much higher computational efficiency. Thus, it shows the real feasibility of applying this algorithm to proton clinical radiation therapy.     

Cu(100)表面上H2/D2散射的六维态-态量子动力学研究：位点平均模型的有效性

基于由optPBE-vdW密度泛函计算的数千个数据点拟合得到的精确的神经网络势能面,本文采用含时波包方法对H₂/D₂在刚性Cu(100)表面上的态-态散射进行了六维量子动力学计算. 与以往的理论和实验比较了H₂和D₂在Cu(100)中的振转(非)弹性散射的结果. 特别是通过将六维的(非)弹性散射几率与从十五个位点加权平均四维几率的结果比较,测试了在该体系中位点平均近似模型的有效性. 具体来说,位点平均模型很好地重现了振动弹性散射几率,尽管对于高能下的振动非弹性散射结果没那么好. 结果说明在未来研究重双原子或多原子分子从金属表面的态-态散射动力学时,可以使用位点平均模型来降低全维计算过高的成本.     

加速神经网络势能面的构建:一种杂化的训练算法

Machine learning approaches have been promising in constructing high-dimensional potential energy surfaces (PESs) for molecules and materials. Neural networks (NNs) are one of the most popular such tools because of its simplicity and efficiency. The training algorithm for NNs becomes essential to achieve a fast and accurate fit with numerous data. The Levenberg-Marquardt (LM) algorithm has been recognized as one of the fastest and robust algorithms to train medium sized NNs and widely applied in recent NN based high quality PESs. However, when the number of ab initio data becomes large, the efficiency of LM is limited, making the training time consuming. Extreme learning machine (ELM) is a recently proposed algorithm which determines the weights and biases of a single hidden layer NN by a linear solution and is thus extremely fast. It, however, does not produce sufficiently small fitting error because of its random nature. Taking advantages of both algorithms, we report a generalized hybrid algorithm in training multilayer NNs. Tests on H+H₂ and CH₄+Ni(111) systems demonstrate the much higher efficiency of this hybrid algorithm (ELM-LM) over the original LM. We expect that ELM-LM will find its widespread applications in building up high-dimensional NN based PESs.     

Whistler-Mode Waves Growth by a Generalized Relativistic Kappa-Type Distribution

The instability of field-aligned Whistler-mode waves in space plasmas is studied by using a recently developed generalized relativistic kappa-type （KT） distribution. Numerical calculations are performed for a direct comparison between the new KT distribution and the current kappa distribution. We show that the wave growth for the KT distribution tends to occur in the lower wave frequency （e.g.,ω 〈~0.1Ω） due to a larger fractional number of the resonant electrons ηrel （which controls the wave growth）, while primarily locating in the higher wave frequency for the kappa distribution. Moreover, the relativistic anisotropy Are1 by the KT distribution is found to be smaller than that by the kappa distribution, leading to a smaller peak of wave growth. The results present a further understanding of plasma wave instability particularly in those plasmas where relativistic electrons are present.