A family of iterative methods for computing the approximate inverse of a square matrix and inner inverse of a non-square matrix (Q846447)

If \(V_0\) is an approximate inverse of a nonsingular square matrix \(A\) such that \(\|I-AV_0\|<1\), the authors show that the iterative formula \[ V_{q+1}=V_q\left[kI-(k(k-1)/2)AV_q+\dots+(-1)^{k-1}(AV_q)^{k-1}\right],\quad k=2,3,\dots, \] converges to \(A^{-1}\) with order of convergence equal to \(k\). An easy method is presented to find an initial approximate inverse \(V_0\). A matrix \(V\) is called a generalized inner inverse of a rectangular matrix \(A\) if \(AVA=A\). The above iterative formula is shown to converge weakly to an inner inverse of \(A\) under appropriate conditions since the inner inverse is not unique. Numerical examples illustrate the efficiency of the new iterative method with the proposed new initial approximation.