一维单原子链空位的一种新统计方法

视一维单原子链中原子为谐振子，运用玻尔兹曼分布给出了空位的一种新统计方法．     

低磁场下激光泵浦超极化惰性气体研究

为了获得高极化度的惰性气体,采用激光泵浦的方法可以使没有放射性的惰性气体同位素(如129Xe、3He)的自旋极化度得到极大增强.圆柱形的玻璃泵浦室内充满天然氙气、氮气、铷蒸气和氦气的混合气体,压力维持在1 MPa左右.泵浦室放置在30Gs的均匀磁场内,并被加热到318～352K.中心波长为794.7 nm的圆偏振激光均匀照射泵浦室对碱金属及惰性气体进行泵浦极化.其获得的非平衡极化度远远高于在相同磁场里玻尔兹曼平衡值.增强的核自旋极化度,可以使惰性气体同位素核磁共振的灵敏度提高104～105倍.     

经典统计力学中的正则分布及系综理论

本文梳理了经典统计力学理论中正则分布的推导问题。从孤立体系的哈密顿方程和相空间出发，强调了相空间坐标可分性是推导正则分布的前提。讨论了温度因子的引入、热力学极限和系综概念的物理本质。进一步阐述了正则分布与玻尔兹曼分布、微正则系综和正则系综的关系。     

基于AT89S52玻尔兹曼常数测量仪的设计

应用PN结电流电压特性关系,采用硅三板管接成共基极电路,由集成运放实现电流电压转换以实现弱电流测量．利用AT89S52单片机所具有的智能测控特点,由AD转换对数据进行采样,采用最小二乘法对数据进行处理,从而较准确测定玻尔兹曼常[k=0．41-1．44）×10^-23J／K],误差小于5％．测量仪有手动、半自动及自动测量三种模式．     

融合词频-逆向文件频率的受限玻尔兹曼机推荐算法

针对数据稀疏性导致推荐算法准确度不高的难题,提出一种融合词频-逆向文件频率(Term frequency-inverse document frequency, TF-IDF)的受限玻尔兹曼机(Restricted Boltzmann machine, RBM)推荐算法,利用受限玻尔兹曼机构建用户项目二维评分矩阵,利用余弦相似度计算方法得出初始推荐评分,最后融合词频-逆向文件频率算法生成最终推荐结果集。对MovieLens1M的电影评分数据进行实验,结果显示,该文提出的混合推荐算法的平均绝对误差(Mean absolute error, MAE)和均方根误差(Root mean square error, RMSE)分别为0.602 8和0.622 5,比传统受限玻尔兹曼机分别提高3.22%与6.06%,也优于对照混合推荐模型的准确率。该算法能提高用户评分预测精度,进一步提升推荐质量。     

用格子Boltzmann方法研究交通流BML模型

用格子玻尔兹曼研究二维决定论文通流BML模型，得出模型满足的玻尔兹曼方程，并对其进行了计算机模拟，获得了与用CA方法研究该模型相类似的结论，提供了格子玻尔兹曼方法用于研究交通流的可行性。     

基于高斯受限玻尔兹曼机的非线性过程故障检测

针对非线性工业过程,提出了一种基于高斯受限玻尔兹曼机(GRBM)模型的故障检测方法.该方法从海量过程数据中提取出GRBM隐层特征信息,通过隐层特征再构建出重构数据,并依据重构误差在残差空间中构建检测统计量,形成了非线性过程故障检测算法.仿真结果表明,基于GRBM的故障检测方法不仅比传统的核主元分析(KPCA)方法具有更好的故障检出率,并且针对大数据量问题具有更强的处理能力.     

New Boundary Treatment Methods for Lattice Boltamann Method

In practical fluid dynamic simulations, the boundary condition should be treated carefully because it always has crucial influence on the numerical accuracy, stability and efficiency. Two types of boundary treatment methods for lattice Boltzmann method (LBM) are proposed. One is for the treatment of boundaries situated at lattice nodes, and the other is for the approximation of boundaries that are not located at the regular lattice nodes. The first type of boundary treatment method can deal with various dynamic boundaries on complex geometries by using a general set of formulas, which can maintain secon-order accuracy. Based on the fact that the fluid flows simulated by LBM are not far from equilibrium, the unknown distributions at a boundary node are expressed as the analogous froms of their corresponding equilibrium distributions. analogous forms of their corresponding equilibrium distributions. Therefore, the number of unknowns can be reduced and an always-closed set of equations can be obtained for the solutions to pressure, velocity and special boundary conditions on various geometries. The second type of boundary treatment is a complete interpolation scheme to treat curved boundaries. It comes from careful analysis of the relations between distribution functions at boundary nodes and their neighboring lattice nodes. It is stable for all situations and of second-order accuracy. Basic ideas, implementation procedures and verifications with typical examples for the both treatments are presented. Numerical simulations and analyses show that they are accurate, stable,general and efficient for pracitical simulations.     

等温气压公式的力学推导方法

用力学的方法推导温气压公式，并解决玻尔兹曼分布不能处理的几种情况。