Formulas for calculating the maximum and minimum ranks of products of generalized inverses of matrices

Yongge Tian
2015-01-01
Abstract:Abstract. A matrix X is called an {i, . . . , j}-inverse of A, denoted by A, if it satisfies the ith,. . . , jth equations of the four matrix equations (i) AXA = A, (ii) XAX = X, (iii) (AX)∗ = AX, (iv) (XA)∗ = XA. The {i, . . . , j}-inverse of A is not necessarily unique and their general expressions can be written as certain linear or quadratical matrix-valued functions that involve one or more variable matrices. Let A and B be two matrices such that the product AB is defined and let A and B be the {i, . . . , j}-inverses of A and B, respectively. Then the product BA will occur in the reverse-order law (AB) = BA. The product BA is usually not unique and is certain linear or nonlinear matrix-valued function with one or more independent variable matrices. In order to characterize the performance of BA, this paper considers maximum and minimum possible ranks of BA for the eight frequently used {i, . . . , j}-inverses of matrices. We shall establish 126 analytical formulas for calculating the global maximum and minimum values of the ranks of BA, and use the formulas to characterize some fundamental algebraic properties of these products. Especially, we establish necessary and sufficient conditions for rank(BA) = rank(AB) and BA = 0 to hold, respectively.
What problem does this paper attempt to address?