Viscosity Characteristic in Metallic Melts with Medium／Short-Range Order Structures

Viscosities and liquid structure of alloys Cu75Al25, Cu87Sn13 and Al-12.5%Si and pure metals Cu and Sn are investigated by using torsional oscillation viscometry and high temperature x-ray diffractometry. The viscosities of pure metals and eutectic alloy melts along with the short-range order structure are found to follow the Arrhenius law in a wide range of temperature above the liquidus. The breakpoints in Arrhenius plots emerge when the structures of alloy melts are transformed from the medium-range order structure to the short-range order structure. It has been found that the change of the viscosity of the metallic alloys is a characteristic of microstructure transformation in the related melts.     

Al_(70)Cu_(30)合金液态RDF高斯分解与局域结构计算

运用密度泛函B3LYP方法对AlmCun(m,n≤4)团簇结构进行量子化学计算,得到团簇的稳定构型。对不同温度下X-射线衍射实验所得的径向分布函数(RDF)在0.2～0.4nm范围内做高斯分解,结合计算结果对高斯峰的意义做了解释,认为计算所得部分团簇构型可以代表Al70Cu30合金熔体的内部局域结构。     

二元合金Ag_(50)Rh_(50)从液态到非晶的分子动力学计算

本文对贵重金属银铑合金Ag50Rh50的液态结构和激冷过程进行了分子动力学模拟研究原子间作用势采用紧束缚势模拟在施加了周期性边界条件的常压状态下进行。采用了偶关联函数、键对分析技术和键取向序参数以分子动力学模拟计算方法揭示了Ag50Rh50的液态结构存在原子偏聚特征以及在快速凝固过程形成原子偏聚的不均匀非晶并与同族过渡金属进行了非晶形成能力的比较。     

温度对液态重金属Pb结构变化的影响

利用分子动力学模拟技术 ,详细考察了在不同温度下液态金属Pb的结构特点。结果发现液态金属Pb的有序度随温度降低不断增强 ;液态金属Pb中存在着不同类型的原子集团 ,原子集团是以不同类型的键对存在的。利用键对分析技术 ,计算出不同温度下的键对类型数和二十面体的两类键趋向序参数 ,从微观上分析了液体与固体的结构相关性。说明了液态金属在不同温度下有不同的结构形式 ,而不象人们想象得那样杂乱无章。     

具有协同约束的共生迁移学习算法研究

传统迁移学习方法通常直接利用相关领域中的数据来辅助完成当前领域的学习任务,而忽略了领域间互相学习的能力.针对此类问题,提出了一种具有协同约束的共生迁移学习方法(Collaborative Constraints based Symbiosis Transfer Learning,CCSTL).在协同约束的基础上,引入共生迁移机制实现领域间的交替互动学习,进而实现源领域和目标领域间的知识迁移,从而提高受训分类器的分类性能.在模拟数据和真实数据集上的实验结果表明了新算法的有效性.     

面向非静态数据分类的演进支持向量机

时间自适应支持向量机(TA-SVM)方法在处理非静态数据集时表现出良好的性能,但仅根据邻接子分类器相似而获得的相关信息并不充分,由此可能会导致训练所得模型不可靠,限制其应用能力.该文通过定义子分类器序列的相关性衰减函数,提出新的面向非静态数据分类问题的演进支持向量机(Evolving Support Vector Machines, ESVM).ESVM 使用衰变函数以体现子分类器之间的相关程度,通过约束所有子分类器之间的带权差异以求得变化更光滑的子分类器序列,契合了数据中隐藏的渐变概念.在各种数据缓慢变化场景的对比实验中,该文的 ESVM 方法优于 TA-SVM 方法.     

基于共同决策方向矢量的多源迁移及其快速学习方法

多源迁移学习提取了多个相似领域之间有用信息,提高了学习效率,但存在计算核矩阵的空间和时间复杂度较高的问题.提出了一种多源迁移学习方法,该方法基于结构风险最小框架理论,以共同决策方向矢量为基准,将多个相似领域的决策方向矢量嵌入到支持向量机的训练过程中,提高了目标领域分类器的分类性能.并结合核心向量机理论提出了共同决策方向矢量核心向量机,实现对大样本数据集的快速分类学习.模拟和真实数据集实验表明了所提算法的有效性.     

Al70Cu30合金液态RDF高斯分解与局域结构计算

运用密度泛函B3LYP方法对Al_mn₃₀(m，n≤4)团簇结构进行量子化学计算，得到团簇的稳定构型。对不同温度下X-射线衍射实验所得的径向分布函数(RDF)在0.2～0.4nm范围内做高斯分解，结合计算结果对高斯峰的意义做了解释，认为计算所得部分团簇构型可以代表Al₇₀Cu₃₀合金熔体的内部局域结构。     

Reverse Monte Carlo study on structural order in liquid and glassy AlFe alloys

This paper reports that anomalous local order in liquid and glassy AlFeCe alloy has been detected by x-ray diffraction measurements. The addition of the element Ce has a great effect on this local structural order. The element Ce favours interpenetration of the icosahedra by sharing a common face and edges. It argues that frustration between this short-range order and the long-range crystalline order controls the glass-forming ability of these liquids. The obtained results suggest that a system having a stronger tendency to show local icosahedral order should be a better glass-former. This scenario also naturally explains the close relationship between the local icosahedral order in a liquid, glass-forming ability, and the nucleation barrier. Such topological local order has also been analysed directly using the reverse Monte Carlo method. It also estimated the fraction of local ordered and disordered structural units in a glassy AlFeCe alloy.     

Electric Conductivity of Phosphorus Nanowires

We present the structures and electrical transport properties of nanowires made from different strands of phosphorus chains encapsulated in carbon nanotubes. Optimized by density function theory, our results indicate that the conductance spectra reveal an oscillation dependence on the size of wires. It can be seen from the density of states and current-voltage curves that the structure of nanowires affects their properties greatly. Among them, the DNA-like double-helical phosphorus nanowire exhibits the distinct characteristic of an approximately linear I - V relationship and has a higher conductance than others. The transport properties of phosphorus nanowires are highly correlated with their microstructures.