论文部分内容阅读
为了提高大地电磁正演模拟的计算效率,开展了基于多重网格算法的Helmholtz方程解算研究;对比Gauss-Seidel、Richardson迭代算法的收敛性,在达到同样的误差限值下,多重网格算法仅需十余次套迭代即可满足误差精度要求,而GaussSeidel和Richardson算法分别需要近500次和5000次迭代;多重网格算法的收敛速度极快,多重网格算法的收敛速度比一般数值算法快近3个数量级;由于其是一种基于套迭代技术,该方法完成一次循环的迭代效率明显比一般数值算法低;多重网格算法极快的收敛速度为将其用于提高大地电磁正演模拟效率提供了可能.
In order to improve the computational efficiency of the forward modeling of the magnetotelluric, a solution to the Helmholtz equation based on the multi-grid algorithm is developed. By comparing with the convergence of the Gauss-Seidel and Richardson iterative algorithms, when the same error limit is reached, the multi-grid algorithm Only need more than ten sets of iterations to meet the requirements of error precision, while GaussSeidel and Richardson algorithms require nearly 500 and 5000 iterations respectively; the multi-grid algorithm converges very fast and the multi-grid algorithm converges faster than the general numerical algorithm Nearly three orders of magnitude faster; because it is based on a set of iterative techniques, the iterative efficiency of the method to complete a loop significantly lower than the general numerical algorithm; multi-grid algorithm convergence rate is very fast for the improvement of the earth magnetism forward Simulation efficiency is possible.