论文部分内容阅读
一般而言，有限差分法求解点源三维地电场正问题所形成的大型稀疏线性方程组Ａｘ＝ｂ，直接解法的计算效率极低。本文从系数矩阵Ａ的不完全Ｃｈｏｌｅｓｋｙ分解及矩阵特征值的特点等角度，说明了不完全Ｃｈｏｌｅｓｋｙ共轭梯度（ＩＣＣＧ）迭代技术可大大提高电阻率三维正演速度的内在原因。结合矩阵Ａ的稀疏存储模式，使得内存需求也大大减少。
In general, the finite difference method solves the large-scale sparse linear equations Ax = b formed by the positive problem of the three-dimensional electric field of point source. The direct solution method is very inefficient. In this paper, we show that the incomplete Cholesky conjugate gradient (ICCG) iteration technique can greatly improve the intrinsic reason of resistivity 3D forward velocity from the perspective of incomplete Cholesky decomposition of coefficient matrix A and the characteristics of matrix eigenvalues. Combined with the sparse storage matrix A, making the memory requirements are also greatly reduced.