The forward order laws for \(\{1,2,3\}\)- and \(\{1,2,4\}\)-inverses of multiple matrix products (Q2279016): Difference between revisions

The Moore-Penrose inverse \(A^{\dagger}\) of a possibly rectangular matrix \(A\) with complex entries is most frequently defined as the solution matrix \(X\) (of appropriate order) of the four equations: \(AXA=A, XAX=X, (AX)^*=AX, (XA)^*=XA\). While all the four equations taken together have a unique solution (viz., \(A^{\dagger}\)), any subset of these equations may have infinitely many solutions, in general. Let \(A\{1,2,3\}\) denote the set of all matrices \(X\) that satisfy the first three equations. The set \(A\{1,2,4\}\) is defined similarly. Given matrices \(A_1, A_2, \dots, A_n\), we say that a matrix \(B \in A_1\{1,2,3\}A_2\{1,2,3\}\dots A_n\{1,2,3\}\) if and only if \(B=B_1B_2 \dots B_n\), where \(B_i \in A_i\{1,2,3\}\) for \(i=1,2,\dots, n\). The set \(A_1\{1,2,4\}A_2\{1,2,4\}\dots A_n\{1,2,4\}\) is defined similarly. The authors, using the generalized Schur complement, prove characterizations for the following set inclusions to hold: \[A_1\{1,2,3\}A_2\{1,2,3\}\dots A_n\{1,2,3\} \subseteq (A_1A_2\dots A_n)\{1,2,3\}\] and \[A_1\{1,2,4\}A_2\{1,2,4\}\dots A_n\{1,2,4\} \subseteq (A_1A_2\dots A_n)\{1,2,4\}.\]