On characterization of operator monotone functions (Q890568)

A function \(f\) on an interval \(I \subseteq \mathbb{R}\) is called matrix monotone of order \(n\) if \(A \leq B \Longrightarrow f(A) \leq f(B)\) for any pair of \(n \times n\) Hermitian \(A\), \(B\) whose spectra lie in \(I\). The order \(A \leq B\) denotes the positive semidefiniteness of \(B-A\) and \(f(A)\) is understood by means of the functional calculus for Hermitian matrices. \(f\) is called operator monotone if \(f\) is matrix monotone of order \(n\) for all \(n \in \mathbb{N}\). For \(A\), \(B\) positive definite, the geometric mean \(A \sharp B\) is defined by \[ A \sharp B:=A^{1/2}(A^{-1/2}BA^{-1/2})^{1/2}A^{1/2}. \] The arithmetic mean \(A \triangledown B\) is defined by \[ A \triangledown B:=\frac{A+B}{2}. \] Let \(|A|\) denote the positive part of a Hermitian matrix \(A\), with respect to the positive semidefinite order. For arbitrary positive definite matrices \(X\) and \(Y\), let \(Z=\frac{1}{2}X^{1/2}|I-X^{-1/2}YX^{-1/2}|X^{1/2}\). The author shows that, if \(f\) is a nonnegative function on \([0,\infty)\) satisfying the inequality \[ f(X \triangledown Y) \leq f(X \sharp Y + Z), \] then \(f\) is operator monotone on \([0,\infty)\).