最小二乘是测量平差最主要的手段,该方法需要对法方程系数阵求逆。介绍对此类实对称矩阵求逆的Jacobi数值方法,通过正交矩阵变换,将实对称矩阵的非对角元变为零,求得特征根和特征向量,从而使求逆变得非常容易,并且有利于判断矩阵是否奇异以及秩亏数是多少。此方法为数值方法,易于计算机实现。
Least square is general approach in surveying adjustment, where inversion of normal equation is inevitable. The algorithm of Jacobi numerical method for inversing real symmetry matrix is introduced. By orthogonal transformation, the non diagonal elements can be changed to zero and eigenvalues with corresponding eigenvectors are obtained. It makes the inversion of matrix very easy. The ranks of normal equation are the number of non-zero eigenvalues. The approach is numerical so can be realized easily with a computer.