摘要
汪荣江提出一个简单的正交归一化技术来克服经典的Thomson-Haskell传播矩阵方法中存在的数值不稳定问题.为了进一步提高计算效率,给出该方法的2种改进.一种改进方法是将传播矩阵中与频率无关的部分分离出来,对于某一固定的水平慢度,这些矩阵只需计算一次;另一个改进是利用Langer块对角化的技术,将传播矩阵分解为几个稀疏矩阵的乘积.我们将改进之后的算法应用于计算水平分层模型中的广义反射系数.较之原有方案,提出的改进能节省一半计算时间.
Wang proposed a simple orthonormalization technique to overcome the numerical instability in the original Thomson-Haskell propagator algorithm. In this paper, two kinds of improvements to the method for computational efficiency are presented. One is to separate the frequency independent matrices from the propagator, these terms can be computed once for a fixed slowness; the other is to decompose the propagator by the Langer block-diagonalizatlon into a product of several sparse matrices. The new developments are applied to calculate the generalized reflection coefficients for a multi-layered model. It leads nearly half reduction in computation time compared with the original scheme.
出处
《中国科学院研究生院学报》
CAS
CSCD
2008年第1期47-53,共7页
Journal of the Graduate School of the Chinese Academy of Sciences
基金
supported by National Natural Science Foundation of China(40374009 ,40574024)
关键词
传播矩阵方法
分层模型
计算效率
反射矩阵
平面波响应
propagator method, layered model, computational efficiency, reflection matrix, plane-wave response