算法

We will show several previous results of the span of simple graphs and multigraphs, and we also will present a new algorithm to compute spT of multigraphs, furthermore we will discuss the bound of the span of a special multigraph and the algorithm of spT.

本文将给出一些关于简单图和重图的已知结论，同时还将给出一种计算重图的sp_T的新算法，并将讨论取特殊T集时重图的T-匝算法以及完全图K_n的T-匝算法。

On the design of algorithms, a novel exact hierarchical delay analysis method for general circuits is proposed; based on the sensitization theorem for sequential circuits, an exact minimizing clocking method is proposed; based on Boolean process, a waveform simulation method considering interconnecting delay for logic circuit and a parallel waveform simulation method are proposed; a new method that transforms bit-level waveform polynomial to word-level polynomial model is proposed; a multiple valued synthesis algorithm based on multiple valued Boolean process and a wire-centered delay synthesis policy are proposed, in which timing planning, floorplanning, wire planning and optimal clock skew in early design are considered; a two-layers channel routing method for minimizing crosstalk under grid mode is proposed; based on the transition numbers theorems for waveform polynomial, a new method for generation of test with noise effects is proposed.

算法设计方面，提出了一种精确的通用电路层次化延时分析方法；基于时序电路的敏化定理提出时序电路最小时钟周期精确确定方法；提出基于Boolean过程论的考虑互连延迟的逻辑电路波形模拟方法，在分析了波形模拟适合并行化基础上，进一步提出一种并行波形模拟算法；提出一种将位级电路波形多项式描述转化成字级多项式描述的新方法；提出一种基于多值Boolean过程的多值电路综合算法以及一种将前期设计定时规划、前期设计的布局规划和线网结构化方法及低偏移的时钟分配等技术相结合的面向互连延时的综合策略；提出一种串绕最小化的网格模式下的双层通道布线方法；从波形多项式描述跳变数的定理出发提出了一种考虑噪声效应的测试生成新方法。

Multistation passive fusion tracking algorithms is given, based on EKF algorithms, and central fusion structure is selected.

将推广卡尔曼滤波算法与集中式融合跟踪算法相结合，用于被动式多站跟踪，给出了基于EKF的被动式多站集中式融合跟踪算法。

In this paper, myopic task schedule algorithm has been developed based on original task schedule algorithm, and its behavior has been simulated and its efficiency has been analyzed.

本文在普通任务调度算法的基础上提出了近视眼任务调度算法，并对近视眼任务调度算法进行仿真实现和性能分析。

In light of the convergence theorem obtained in Chapter 2, a result on topology preservation of 2-dimensional SOM algorithm with continuous inputs is given. In Chapter 4, an equivalence relation on n dimensional Euclidean space, called equi-action relation, is introduced and a number of properties on this relation are studied.

指出了已有的有关SOM算法拓扑保序定义的缺陷，并在此基础上，给出了一个具有连续输入的二维SOM算法的拓扑保序新定义；利用第二章中的收敛性定理，我们进而获得了一个具有连续输入的二维SOM算法的拓扑保序性定理。

Then proposes a video image pre-processing method based on adaptive neighborhood statistics. The proposed pre-processing conquers the low efficiency of classical adaptive neighborhood filter and the sensitivity to threshold in other methods.

提出了一种基于自适应邻域统计的视频图像预处理算法，该算法解决了自适应邻域滤波中邻域构建的效率问题和传统预处理算法的门限敏感问题。

Considering the actual need of hydro-structure design, one algorithm for drawing isoline map or nephogram is proposed in this paper. Compared with other algorithms, this algorithm is easy to implemente, and the increase of series of isoline will not result in the observable increase of the time for computing the isoline.

此算法依靠单元共边共点的特性保证等值线或云图的连续性，相对等值线追踪算法而言，此算法简单易行，而且实际测试结果表明等值线级数的增加不会导致其计算时间的显著增长。

In conjunction with MICQ, a nephogram clustering algorithm based on PSO is proposed, of which the coding and fitness function are projected. In addition, nowadays, many researches of nephogram classification usually focus on some single feature, and the classifications are achieved based on these unique features, correspondingly. This strategy would probably ignore the potential which is provided by the combined classification based on multiple features.

结合该参数，本文将一种基于群体智能的随机优化技术：微粒群优化算法，引入到云图的分类领域，从算法编码和目标函数的构造方面进行了设计，提出了一种基于微粒群优化的云图聚类算法，为后一阶段的云图分类打下基础。

The optimal models and algorithms for making up pairs and non-pairs locomotive turn-around diagram are given out. Specially, a more efficient way to consider both the optimum and the equilibrium of this problem is also presented as follows: at first formulate the network flow model of this problem and solve it by the method of the minimal cost flow and then use the labeling method to readjust the solution.

对成对和不成对机车周转图编制问题进行了探讨，给出了其最优配置的数学模型和算法；在此基础上，提出了把均衡性与最优化综合考虑的不成对机车周转图的改进算法——首先建立网络流模型用最小费用流算法求得问题的初始最优解，然后用标记法对该解进行调整，使其在最优的基础上达到均衡。

The convergence speed of forward algorithm is as fast as ten times of backward algorithm.Additionally,the decrypted image quality and noiseproof ability for the forward algorithm are also higher than the backward algorithms obviously.

不仅如此，整个算法的收敛速度比后向算法快一个数量级，而且无论迭代结果的解密图象质量还是抑制噪声能力也都明显优于后向算法。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。