摘要
模拟电路的仿真问题最终归结为对线性代数方程组的求解。利用分块化方法可以降低求解过程中Jacobi矩阵的维数,从而有效降低求解时间。如何降低求解线性方程组的迭代次数,是有效降低求解时间的另一重要问题。首先详细分析了用于求解模拟电路代数方程中Jacobi矩阵的划分问题,然后提出一种改进的隐式迭代方法。最后,通过实验分析了算法中内迭代次数Iin对总迭代次数的影响,该结论对提高整体加速比具有指导意义。
Analog circuit simulation comes down to the solution of linear algebraic equations. PartitioningJacobi matrix and reducing the number of iterations are major issues and effective ways to make the simulation cycle shorten. This paper analyzed the partition of Jacobi matrix in detail, then proposeda modified implicit iteration algorithm. Finally, an example was presented to show the effect of the number of inner iteration Iin on the whole iteration number, from which a conclusion was reached to be guidance for improving the overall speedup.
出处
《微处理机》
2008年第2期152-154,157,共4页
Microprocessors
关键词
雅可比矩阵
显式方法
隐式方法
块边界矩阵
Jacobi matrix
Explicit method
Implicit method
Block bordered matrix
