Some operator inequalities concerning generalized inverses (Q809327)

The author extends to the von Neumann-Schatten classes some inequalities concerning generalized inverses established by Penrose for matrices. The author does not cite any papers dealing with generalized inverses of linear operators in Banach or Hilbert spaces. However, in the paragraph preceding the Preliminaries the author states the following: ``After writing this paper I learned of \textit{H. W. Engl} and \textit{M. Z. Nashed}'s work [New extremal characterizations of generalized inverses of linear operators, J. Math. Anal. Appl. 82, 566-586 (1981; Zbl 0492.47012)]. Some of their results are similar to the ones obtained here. Thus, they prove a special case (viz. when \(p=2)\) of Theorem 3.1(b) [Theorem 6.2 in Engl and Nashed]. Their methods are quite different to the ones used in this paper.'' It would have been useful if the author had commented in more detail on the relation between his paper and the cited paper of Engl and Nashed (EN). The similarities between the two papers are striking, although the methods of proofs differ. In this review we describe in a more definite way the relationship between the two papers, and point out that some of the author's results are still of interest. Lemma 1.4 in Mahler's paper follows from Lemma 6.1 in EN. His Theorem 2.1 cannot be obtained from the results of EN since they do not (and, in their approach, cannot) consider the operator norm. Theorem 2.2 essentially follows from Theorem 3.1 in EN; in fact EN consider more general norms and go further by characterizing the Moore-Penrose inverse among all \(\{\) 1,3\(\}\)-inverses. Theorem 3.1a does not follow directly from results in EN; Theorem 3.1b has been proven in EN for \(p=2\), as the author correctly remarks. The last remarks before the references are essentially the same as Corollary 6.3 in EN. Finally as far as Schatten norms are concerned, the results and methods of proof in EN are more general.