A note on definition of matrix convex functions (Q2496647): Difference between revisions

The author obtains the following Theorem: For a function \(f: S\rightarrow \mathbb R\), (\(S\) is any interval in \(\mathbb R\)) the following conditions are equivalent: (i) \(f\) is a matrix convex function, i.e., \(f(\alpha X+(1-\alpha )Y)\leq \alpha f(X)+(1-\alpha )f(Y)\) for all self-adjoint operators \(X,~Y\) in a finite-dimensional Hilbert space and every \(\alpha \in [0,1]\) provided that the spectra of \(X\) and \(Y\) lie in \(S\); (ii) for any natural \(k\), for all families of positive operators \(\{A_i\}_{i=1}^k\) in a finite-dimensional Hilbert space \(H\), such that \(\sum_{i=1}^k A_i=1_H\), and arbitrary numbers \(x_i \in S\), the inequality \[ f\left( \sum_{i=1}^k x_i A_i \right) \leq \sum_{i=1}^k f(x_i)A_i \] holds true.