本文提出了采用穆尔-彭劳兹广义逆A~+最小二乘拟合穆斯堡尔谱的计算机算法,给出了计算A~+的迭代公式和算法程序框图。算法如下:用高斯-牛顿法得到第K次初估值Q~K的法方程组Hδ~K=B。再由矩阵H的穆尔-彭劳兹广义逆A~+得到修正量δ~K的解δ~K=A+B。由一维寻查法找出最优步长因子λ~K,构成第K+1次初估值Q~(K+1)=Q~K+λ~Kδ~K。如此迭代直到满意为止。由于任意矩阵H,其穆尔-彭劳兹广义逆A~+必有唯一解存在。当H非奇异 In this paper, we propose a computer algorithm that uses Moore - Penrose generalized inverse A ~ + least square fitting Mossbauer spectroscopy, and presents an iterative formula for computing A ~ + and a block diagram of the algorithm. The algorithm is as follows: Using the Gauss - Newton method to get the Kth initial estimate Q ~ K of the system of equations Hδ ~ K = B. The solution δ ~ K = A + B of the correction δ ~ K is obtained from the Moore - Penrose singular inverse A + of the matrix H. The optimal step factor λ ~ K is found by one-dimensional searching method, which constitutes the first K + 1 initial estimate Q ~ (K + 1) = Q ~ K + λ ~ Kδ ~ K. So iterative until satisfied. Due to arbitrary matrix H, the Moore-Penrose generalized inverse A ~ + must exist. When H is not singular