Integral inequalities of the Heinz means as convex functions (Q2790776)

Let \(M_{n}\) be the set of all \(n\)-by-\(n\) matrices, and \(M_{n}^{+}\) be a set of all positive definite matrices in \(M_{n}\). The author proves monotonicity and convexity of the function NEWLINE\[NEWLINE f(r)=|\!|\!| A^{p+r}XB^{q-r}+A^{q-r}XB^{p+r}|\!|\!| NEWLINE\]NEWLINENEWLINENEWLINEfor \(0\leq r\leq q\), where \(X\in M_{n}\), \(A,B\in M_{n}^{+}\), \(0<p\leq q\), and \(|\!|\!|\cdot |\!|\!|\) is any unitarily invariant norm. Next, the author defines a Heinz function to be convex on \([0,1]\), continuous on \([0,1]\), decreasing on \([0,\frac{1}{2}]\), increasing on \([\frac{1}{2},1]\) and symmetric about \(\frac{1}{2}\). The above function \(f(r)\) is then a Heinz function. Using the Hermite-Hadamard inequality, the author obtains several inequalities for Heinz functions.