Improved Heinz inequalities via the Jensen functional (Q386399): Difference between revisions

\textit{R. Bhatia} and \textit{C. Davis} [SIAM J. Matrix Anal. Appl. 14, No. 1, 132--136 (1993; Zbl 0767.15012)] proved that the function \(g(p)=|||A^{1+p}XB^{1-p}+A^{1-p}XB^{1+p}|||\) is convex on \([-1,1]\) for all matrices \(A,B,X\) with \(A\geq0\), \(B\geq0\) and all unitarily invariant norms. As a consequence, in the same setting for \(A,B,X\), the matrix inequality \(2|||A^{\frac{1}{2}}XB^{\frac{1}{2}}|||\leq|||A^tXB^{1-t}+A^{1-t}XB^t|||\leq|||AX+XB|||\) which is known as the Heinz norm inequality holds true for every \(t\in[0,1]\). \textit{F. Kittaneh} and \textit{Y. Manasrah} [J. Math. Anal. Appl. 361, No. 1, 262--269 (2010; Zbl 1180.15021)] established some improvements of this inequality. Also, \textit{F. Kittaneh} [Integral Equations Oper. Theory 68, No. 4, 519--527 (2010; Zbl 1230.47026)] obtained several refinements of Heinz norm inequality using the Hermite-Hadamard inequality for convex functions. In the present paper, the authors utilize the properties of the Jensen functional \(\mathfrak{J}(f,x,t)=(1-t)f(x_1)+tf(x_2)-f((1-t)x_1+tx_2)\) applied to the convex function \(f(t)=|||A^tXB^{1-t}+A^{1-t}XB^t|||\) to obtain several improvements of the Heinz norm inequality. They also present some inequalities involving operator means which can be regarded as operator analogues of the Heinz norm inequality.