讨论了一类求解非线性方程组算法的并行性能。与传统的算法不同之处是用一个块对角矩阵作为迭代矩阵，且该矩阵可由一个仅包含向量内积和矩阵与向量乘积的递推关系简便计算得到。 在对算法进行描述之后，分析了算法的并行执行过程，给出了算法的并行加速比和对存储的需求分析，数值计算表明理论分析与数值结果相符合，算法具有较好的并行度和较低的存储要求。可适用于一般和大规模科学与工程的高性能计算。

该文讨论了一类求解大规模非线性方程组算法的并行性能及其在曙光并行机上的实现过程。与传统的算法不同之处是用一个块对角矩阵作为迭代矩阵，且该矩阵可由一个仅包含向量内积和矩阵与向量乘积的递推关系简便计算得到。在对算法进行描述之后，分析了算法的并行加速比和存储需求，讨论了算法在基于消息传递的MPI并行环境下的实现流程，数值计算表明理论分析与数值结果相符合，算法在分布式并行环境下具有较好的并行度和较低的存储要求，可适用于大规模科学与工程的高性能计算。

本文提出了一种基于TMN 的集中式路由管理新方法C2MRVT，并运用排队论理论对C-MRVT和传统MRVT方法进行了性能分析比较，最后给出了该方法的实现方案和模拟结果。理论分析和模拟结果表明，C2MRVT是一种有前景的、技术可行的信令网路由管理方法。

主要讨论了基于OSA/Parlay标准的业务管理系统,特别是计费和账务系统。分析了OSA的体系结构、VHE 模型和网络互连技术，讨论了组成OSA 的三大模块和业务管理流程。在此基础上，给出个人多媒体移动业务管理模型原型系统，分析了系统中每个模块的功能以及系统中数据库的构造和关系。讨论了采用XML Web Service 技术进行系统的实现。

Grid is a distributed computing and resource environment. Security is an important issue in the grid environment. In this paper, we present a prototype of the grid security architecture. It shows that the architecture is scalable, and meets the security requirements of the grid.

Security is an important issue in research and appliance of grid computing.Grid security model is composed of a series of mechanism and strategy to solve various practical security problems. This paper presents a grid security model based on virtual organization referring to GSI(a security component in Globus). The model consists of a logical model and a physical model.

Solving nonlinear equations is a problem often needed to be dealt with in the practical engineering application. This paper discusses preconditioning methods based on Block Broyden method. It first introduces the Block Broyden method and preconditioning technique. Then it presents four different preconditioners for the Block Broyden method and discusses the implementation process. It also analyzes the time complexity of the four algorithms. Finally. it solves the nonlinear systems arising from the Bratu problem by the four preconditioning algorithms. Experimental results show that the preconditioning technique is effective for the Block Broyden method and that the preconditioner SSOR has the fastest solving speed and the best performance among the four preconditioners. Therefore. it can be used in the large-scale problems arising from scientific and engineering computing.