一种基于边界识别的聚类算法

针对基于密度的聚类算法由高密度区到低密度区的处理顺序所带来的不能识别低密度对象类别的缺陷,通过对聚类过程中可能存在的边界识别进行讨论,提出了一种基于边界识别的聚类算法.该算法的思想是：同簇优先权高于密度优先权,即在选择下一个对象进行聚类时,在已聚类的对象中优先选择同一簇的对象,当对象沿某一方向扩展到达簇边界时停止扩展,转而向其他方向扩展,这种处理顺序能使得类别最大化.通过分析簇边界的密度变化特征,建立了边界识别准则,并根据该准则对数据进行聚类.通过在合成数据和美国加州大学提供的知识挖掘数据库数据集上的实验结果表明,所提算法能有效地处理低密度区域的数据,与识别聚类结构的对象排序算法相比,聚类效果可提高4%左右,而时间性能相当.     

农户小额信贷对农信社盈利能力的影响分析

随着小额信贷的发展,关于小额信贷的影响和绩效评价已成为当前国内外理论和实证研究关注的焦点.针对我国当前农信社为小额信贷主体的现状,从实证的角度分析了农户小额信贷业务的开展对农信社盈利能力的影响,并对更好地开展小额信贷业务提出了相关的政策建议.     

Application of backflush and micro-flow techniques to the analysis of C5—C13 hydrocarbons in crude oils and its geochemical significance

An Agilent 6890N GC equipped with both FID and Agilent 5975 MSD analyzer has been employed to analyze C₅—C₁₃ hydrocarbon fractions of crude oils. A technical combination of program temperature volatilizer injection, backflush and micro-flow controller afforded a fine separation of C₅—C₁₃ compounds on a PONA column with the heavy part of crude oils being cut off before entering the analytical column. Both GC-FID chromatogram and GC-MS mass chromatograms (MID and full scan) could be obtained at the same time. The retension time differences of nC₆—nC₁₃ alkanes between GC and TIC were in the range of 0.02—0.58 minutes. Totally 286 peaks have been assigned group compositions. Results on 8 typical oil samples from CNPC Key Laboratory Crude Oil Library showed that the characteristics of C₆—C₁₃ hydrocarbon group composition could be used in oil-oil correlation studies.     

Local orthogonal transform for a class of acoustic waveguides

There are some curved interfaces in acoustic waveguides. To compute wave propagation along the waveguides with some marching methods, flattening of the internal interfaces is needed. In this paper, a local orthogonal coordinate transform and an equation transform are constructed to solve the two-dimensional Helmholtz equation for the waveguides bounded by a flat top, a flat bottom and two curved internal interfaces with three layered media. The curved internal interfaces are flattened by the local orthogonal coordinate transform, and the corresponding transformed Helmholtz equation can be solved by some marching methods. This treatment can be extended in multilayered medium waveguides. The one-way reformulation based on the Dirichlet-to-Neumann (DtN) map is then used to reduce the boundary value problem to an initial value problem. Numerical implementation of the resulting operator Riccati equation uses a large range step method to discretize the range variable and a truncated local eigenfunction expansion to approximate the operators. This method is particularly useful for solving long range wave propagation problems in slowly varying waveguides with multilayered medium structure.     

中国产蕗蕨属(膜蕨科)的分类学修订

基于大量植物标本的研究,对中国产蕗蕨属植物进行了分类修订,认为中国产的蕗蕨属植物为17种并给出了各个种的形态描述,地理分布及分种检索表.     

一种低电压低功耗Rail-to-Rail CMOS运算放大器的设计

基于2 μm CMOS工艺!设计实现了一种2.4 V低功耗带有恒跨导输入级的RailtoRail CMOS运算放大 器。采用尾电流溢出控制的互补差分输入级和对称56类推挽结构的输出级，实现了满电源幅度的输入输出和恒 输入跨导；运用折叠共源共栅结构作为中间增益级，实现电流求和放大。整个电路在2.4 V的单电源供电下进行 仿真，直流开环增益为76.5 dB,相位裕度为67.6,单位增益带宽为1.85 MHz。     

数学概论讲演（1943～2004）

本书作者B．埃克曼是瑞士当代著名数学家，以研究拓扑、微分几何、同调代数和群论而著称。他毕业于瑞士苏黎世著名学府EHT（联邦技术学院），后来曾在瑞士洛桑大学及美国普林斯顿高等研究所任教和从事科研工作，1948年起任EHT的数学教授，     

Wavelet Numerical Solutions for Weakly Singular Fredholm Integral Equations of the Second Kind

Daubechies interval wavelet is used to solve numerically weakly singular Fredholm integral equations of the second kind. Utilizing the orthogonality of the wavelet basis,the integral equation is reduced into a linear system of equations. The vanishing moments of the wavelet make the wavelet coefficient matrices sparse,while the continuity of the derivative functions of basis overcomes naturally the singular problem of the integral solution. The uniform convergence of the approximate solution by the wavelet method is proved and the error bound is given. Finally,numerical example is presented to show the application of the wavelet method.     

Excitatory effect of histamine on neuronal activity of rat cerebellar fastigial nucleus Jn vitro

The cerebellar fastigial nucleus （FN） holds an important role in motor control and body balance. Previous studies have revealed that the nucleus is innervated by direct hypothalamocerebellar hletaminergic fibers. However, the functional role of histaminergic projection in cerebellar FN has never been established. In this study, we investigated the effect of histamine on neuronal firing of cerebellar FN by using slice preparations. Sixty-five FN cells were recorded from 47 cerebellar slices, and a vast majority of the cells responded to histamine stimulation with an excitatory response （58/65, 89.2%）. Perfusing slices with low-Ca^2＋/high-Mg^2＋ medium did not block the histamine-induced excitation （n=10）, supporting a direct postsynaptic action of histamine on the cells. Furthermore, the excitatory effect of histamine on FN neurons was not blocked by selective histamine H1 receptor antagonist triprolidine （n=15） or chlorpheniramine （n=10）, but was effectively suppressed by ranitidine （n=15）, a highly selective histamine H2 receptor antagonist. On the other hand, highly selective histamine H2 receptor agonist dimaprit （n=20） instead of histamine HI receptor agonist 2-pyridylethylamine （n=16） mimicked the excitatory effect of histamine on FN neurons. The dimaprit-induced FN neuronal excitation was effectively antagonized by selective histamine H2 receptor antagonist ranitidine （n=13） but not influenced by selective histamine H1 receptor antagonist triprolidine （n=15）. These results demonstrate that histamine excites cerebellar FN cells via the histamine H2 receptor mechanism and suggest that the hypothalamocerebellar histaminergic fibers may modulate cerebellar FN-mediated sensorimotor integration through their excitatory innervations on FN neurons.     

Mechanism of the interaction between Au nanoparticles and polymerase in nanoparticle PCR

Nanoparticle PCR is a novel method to optimize DNA amplification. It performs well in improving specificity, enhancing sensitivity and speed. Several mechanisms were proposed in previous studies: one was based on the interaction between gold nanoparticles (AuNPs) and DNA while the other was attributed to the heat transfer property of AuNPs. In this paper, we propose that the interaction between AuNPs and DNA polymerase can significantly influence PCR. First, the addition of DNA polymerase can eliminate the inhibitory effects of excess AuNPs. Second, the addition of AuNPs will increase yield of the desired PCR product and make the optimum concentration of DNA polymerase move to higher value. Third, while excess polymerase might inhibit amplification efficiency, AuNPs can reverse this process and the yield of PCR amplification. Based on these results we propose a possible mechanism that AuNPs might modulate the activity of polymerase and improve PCR amplification.