Synopsis
A 2-inverse for a matrix A is any matrix G satisfying GAG=G. This book adresses the special, but heretofore ignored, role of 2-inverses in statistics. The relationship among various kinds of generalized inverses and their construction in terms of 2-inverses are explored. The applications discussed include the role of 2-inverses in: canonically representing second-degree polynomials x'Ax+b'x+c, obtaining vectors ß which satisfy K'ß=0 and minimize y'y-Xß(squared), and relating the hypotheses tested by various analytical approaches in generalized linear models. For each topic, the simplification of both the concepts and the notation which results from a 2-inverse approach is stressed and illustrated with examples.
"synopsis" may belong to another edition of this title.
