Note on the iterative methods for computing the generalized inverse over Banach spaces. (Q2918588)

The author defines iterative methods for approximating the generalized inverse \(A^{(2)}_{T,S}\) of the operator \(A\) over Banach spaces. These iterative methods are of the form NEWLINE\[NEWLINE X_k=\beta \sum _{i=0}^{t-1}(I_x-\beta YA)^iY+(I_x-\beta YA)^tX_{k-1},\; t\geq 1,\; k=1,2,\dots NEWLINE\]NEWLINE or in the dual form NEWLINE\[NEWLINE X_k=\beta Y\sum _{i=0}^{t-1}(I_y-\beta AY)^i+X_{k-1}(I_y-\beta AY)^t,\; t\geq 1,\; k=1,2,\dots NEWLINE\]NEWLINE The conditions for the convergence of defined iterations are defined. Error bounds of these iterative methods are estimated as NEWLINE\[NEWLINE \| A^{(2)}_{T,S}-X_k\| =\frac {| \beta | q^{kt}}{1-q}\| Y\| \| I_y-AX_0\| NEWLINE\]NEWLINE or NEWLINE\[NEWLINE \| A^{(2)}_{T,S}-X_k\| =\frac {| \beta | q^{kt}}{1-q}\| Y\| \| I_x-X_0A\| , NEWLINE\]NEWLINE respectively. Furthermore, analogous iterative methods for computing the generalized Drazin inverses \(a^d\) of the Banach algebra element \(a\) are defined and investigated. A numerical example is presented.NEWLINENEWLINEThe results of the paper are new and mainly represent a significant improvement of the iterative methods defined in the previous paper by \textit{X. Liu, C. Hu} and \textit{Y. Yu} [J. Comput. Appl. Math. 234, No. 3, 684--694 (2010; Zbl 1190.65088)].