Outer inverses of matrices (Q1601634)

The inverse of an invertible matrix \(A\) is a scalar multiple of the classical adjoint of \(A\). \textit{E. H. Moore} [On the reciprocal of the general algebraic matrix. Bull. Am. Math. Soc., XXVI, 394-395 (1920)] who extended this observation to represent what is now called the Moore-Penrose inverse of a matrix of rank \(r\) as a linear combination of matrices obtained from the adjoint of the \(r{\times}r\) submatrices of \(A\). Basic to the study of generalized inverses of matrices is the equation \(AGA=A\). If this equation holds, then \(A\) is said to be regular and \(G\) is said to be an inner inverse of \(A\). Recently, it was shown that the inner inverses of a regular matrix \(A\) are expressible as the images of \(R\)-module morphisms associated with \(A\) called adjoint mappings. Also, if \(AGA=A\), then \(A\) is said to be an outer inverse of \(G\). In the paper under review the author asks the following question: What characterizes the outer inverses of \(G\)? He gives an answer to this question in terms of adjoint images under \(G\). More precisely, let \(\mathcal{R}\) be a commutative ring with \(1\) and let \(G \in \mathcal{R}^{n{\times}m}\). If \(A \in \mathcal{R}^{m{\times}n}\) is of \({\text{rank}} r \geqslant 1\) with \({\text{tr}} C_r(AG)=1\), then \(AGA=A\) iff \(G_{C_{r}(A)}=A\); that is, \(A\) is an outer inverse of \(G\) iff \(A\) is the adjoint image of \(C_{r}(A)\) under \(G\).