116 Functional inequalities for max eigenvalues of nonnegative matrices.pdfVIP

Linear Algebra and its Applications 426 (2007) 290–298 /locate/laa Functional inequalities for max eigenvalues of nonnegative matrices Ludwig Elsner a,∗, Daniel Hershkowitz b a Fakultät für Mathematik, Universität Bielefeld, Postfach 100131, 33501 Bielefeld, Germany b Mathematics Department, Technion – Israel Institute of Technology, Haifa 32000, Israel Received 23 February 2007; accepted 27 April 2007 Available online 22 May 2007 Submitted by R.A. Brualdi Dedicated to Hans Schneider on the occasion of his 80th birthday Abstract Let P+n denote the set of all n× n nonnegative matrices. For a function f : Rm+ → R+ and matrices Ak = (akij )ni,j=1, k = 1, . . . , m, define f (A1, . . . , Am) = (f (a1ij , . . . , amij ))nij=1. For each A ∈ P+n we denote its spectral radius by ρ(A) and its max eigenvalue by μ(A). In a previous paper, all functions f which satisfy ρ(f (A1, . . . , Am))  f (ρ(A1), . . . , ρ(Am)), ∀n ∈ N, ∀A1, . . . , Am ∈ P+n and some functions which satisfy f (ρ(A1), . . . , ρ(Am))  ρ(f (A1, . . . , Am)), ∀n ∈ N, ∀A1, . . . , Am ∈ P+n , were characterized. Here, for an interval I in R+, we characterize those functions f satisfying μ(f (A1, . . . , Am))  f (μ(A1), . . . , μ(Am)), ∀n ∈ N, ∀A1, . . . , Am ∈ Inn ∗ Corresponding author. E-mail addresses: elsner@math.uni-bielefeld.de (L. Elsner), hershkow@techunix.technion.ac.il (D. Hershkowitz). 0024-3795/$ - see front matter ( 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.laa.2007.04.021 L. Elsner, D. Hershkowitz / Linear Algebra and its Applications 426 (2007) 290–298 291 as well as the functions satisfying f (0) = 0 and f (μ(A1), . . . , μ(Am))  μ(f (A1, . . . , Am)), ∀n ∈ N, ∀A1, . . . , Am ∈ Inn. © 2007 Elsevier Inc. All rights reserved. AMS classification: 15A23; 15A33 Keywords: Nonnegative matrix; Spectral radius; Functional inequalities; Max eigenvalue; Max algebra; Max-plus algebra 1. Introduction For a positive integer number n let P+n denote the set of all n× n nonnegative matrices. For a