摘要
基于广义 Gauss-Seidel迭代法的思想 ,提出了一种求解最小二乘问题的并行算法 ,并证明了该算法是收敛的。采用该算法计算时 ,由于仅需对 Jacobi矩阵的一个子矩阵进行三角分解 ,而不需要对整个 Jacobi矩阵进行三角分解 ,使计算量大为减少 ,从而大大节省了计算所需的时间。
Based on the thought of Gauss Seidel iteration,a parallel algorithm is proposed for solving the least square problems,the method is proved to be convergent The calculated amount is greatly reduced,thus calculating time is saved extensively when the method is used,because only triangle decomposition is needed for one of the submatrices of Matrix Jacobi and the decomposition is not needed for the whole Jacobi
出处
《江汉石油学院学报》
EI
CSCD
北大核心
2002年第1期92-93,共2页
Journal of Jianghan Petroleum Institute
