基于不动点转移的SA动态演化模型  

SA Dynamic Evolution Model Based on Static-Point Transition

作　　者：王映辉[1] 刘瑜[2] 王立福[3] 

机构地区：[1]陕西师范大学计算机学院,西安710062 [2]北京大学信息遥感与地理信息系统研究所,北京100871 [3]北京大学信息科学技术学院软件研究所,北京100871

出　　处：《计算机学报》2004年第11期1451-1456,共6页Chinese Journal of Computers

基　　金：国家"八六三"高技术研究发展计划项目基金 (2 0 0 1AA113 171);国家"九七三"重点基础研究发展规划项目基金 (2 0 0 2CB3 12 0 0 6);国家博士后基金 (2 0 0 40 3 5 0 2 5 1);陕西省自然基金 (2 0 0 3F3 5 )资助 .

摘　　要：构造性和演化性是软件的两个基本特性 .而软件演化包括静态演化和动态演化两个方面 .动态演化更为复杂 ,这种复杂性决定了 ,对动态演化的研究首先应从宏观层面入手 .软件体系结构SA作为软件的蓝图和支撑骨架 ,为人们宏观把握软件的动态演化提供了一条有效的途径 .该文描述了构件———连接件组成的SA动态语义网络模型 ,分析了SA动态语义网络模型中的浸润过程 ,给出基于不动点的浸润过程收敛的判定 ,提出了邻接矩阵过滤和原子过滤的概念 ,阐明了基于邻接矩阵原子过滤的SA动态语义网络浸润步的原子性 .指出SA动态演化过程可用一系列邻接矩阵原子过滤在时刻上相继的逻辑衔接来描述 .最后给出了两个层面上对SA动态演化波及效应的分析方法 .为基于矩阵变换的SA动态演化的进一步研究和计算机自动量化描述奠定了基础 .Construction and evolution are two basic properties of software. Software evolution includes static and dynamic evolution. Software dynamic evolution is more complex, and this complexity decides that research of software dynamic evolution should be begun with macroscopical level. Software architecture SA acts as blueprint and skeleton of software, which is an availability approach to grasp macroscopical software architecture and evolution based on SA. SA dynamic semantic network model based on components and connectors is described. Soak process in SA dynamic semantic network is analyzed, and decision condition about convergence of soak process is addressed. A concept of adjacency matrix atomic filtration and atomic property of soak step based on adjacency matrix atomic filtration is put forward, so SA dynamic evolution is described by a series of continuous adjacency matrix atomic filtrations. At last, an analysis approach of ripple-effect of SA dynamic evolution is described based on two of levels. All the above are credible foundation of SA dynamic evolution research in the future, and are automatic transition foundation based on matrix in computer.

关 键 词：软件体系结构 动态演化 SA动态语义网 邻接矩阵 浸润 不动点 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]