大规模方程组参数估计与精度评定的改进算法

AN IMPROVED ALGORITHM OF PARAMETER ESTIMATION AND ACCURACY ASSESSMENT OF LARGE-SCALE NORMAL EQUATIONS

现代大地测量数据处理经常需要解算大规模线性方程组,鉴于传统算法的参数估计与精度评定的效率较低,综合应用矩阵Cholesky分解技术和OpenMP并行库对传统算法进行改进。计算结果表明,改进后的参数估计与精度评定算法的计算结果与传统算法一致,但计算效率得到显著提高,且引入OpenMP并行库的改进算法更适合大规模法方程组的解算。

A large-scale normal equation is often demanded in modem geodetic data processing, whereas the traditional algorithm performs very slowly in parameter estimation and accuracy assessment. This paper presents an improved algorithm of large-scale normal equations by using Cholesky decomposition and the OpenMP parallel li- brary. The simulation experiments show that the improved algorithm can obtain the same results as the traditional algorithm do, but have a better performance in parameter estimation and accuracy assessment compared with traditional algorithm, and the OpenMP parallel library is more suitable for solving large-scale normal equations.