论文部分内容阅读
对于大系统,特别是刚性系统的仿真算法,隐式法与显式法相比,前者在精度、稳定域、解题的阶次、对刚性系统适应性等方面更为优越。文中介绍了三种所述系统的隐式并行算法:分级分散法、分级分解法和分解的牛顿法。其中,分解的牛顿法及其软件LEIDEC 的应用对于所述系统结构无依赖性,通过实例计算,表明它的运算速度比分级分解法高数倍,可得到广泛的应用。
For the simulation of large systems, especially rigid systems, the implicit method is superior to the explicit method in terms of accuracy, stability domain, order of solving problems and adaptability to rigid systems. In this paper, three kinds of implicit parallel algorithms of the system are introduced: Newton method of grading scatter method, fractionation method and decomposition method. Among them, the decomposed Newton method and the application of its software LEIDEC have no dependence on the structure of the system, and the example computation shows that the computing speed of the LEIDEC is several times higher than that of the hierarchical decomposition method and can be widely applied.