我们已研制出用共轭梯度松弛法进行快速3D直流电阻率正演模拟和反演的算法。正演网格模拟计算中,一种不完备的Cholcsky分解法(用于预处理)与稀疏矩阵法相结合,形成了快速有效的算法(对50×50×20区块而言.用SUN SPARC—2工作站所需的CPU时间大约为2分钟)。侧边界与底边界条件为标定阻抗条件,它考虑了不同电流源排列在边界处产生的局部电流量。我们用共轭梯度松弛法求解极大似然反演方程。由于共轭梯度法仅需要 We have developed an algorithm for fast 3D DC resistivity forward modeling and inversion using the conjugate gradient relaxation method. In forward grid simulation, an incomplete Cholcsky decomposition (for preprocessing) combined with the sparse matrix method results in a fast and efficient algorithm (for a 50 × 50 × 20 block, the SUN SPARC- 2 workstation CPU time required is about 2 minutes). The side and bottom boundary conditions are calibrated impedance conditions that take into account the amount of local current generated by the different current sources arranged at the boundary. We use the conjugate gradient relaxation method to solve the maximum likelihood inversion equation. Because conjugate gradient method only needs