A majorization inequality for Wright-convex functions revisited (Q443930): Difference between revisions

A function \(f:I\to \mathbb{R}\), where \(I\) is an open interval, is said to be Wright-convex provided \[ f(tx+(1-t)y)+f((1-t)x+ty)\leq f(x)+f(y),\qquad x,y\in I, t\in [0, 1]. \] Moreover, for any \(\bar{x}:=(x_1, \ldots , x_n), \bar{y}:=(y_1, \ldots , y_n)\in {\mathbb{R}}^n\) we write \(\bar{x}\prec \bar{y}\) if, after a possible reordering of the components so that \(x_n \leq \ldots \leq x_1\) and \(y_n \leq \ldots \leq y_1\), we have \[ x_1 +\ldots +x_k \leq y_1 +\ldots +y_k ,\qquad k\in \{1,\ldots , n-1\} \] and \[ x_1 +\ldots +x_n = y_1 +\ldots +y_n . \] In this paper the authors use the fact that \(f:I\to \mathbb{R}\) is Wright-convex if and only if, for any \(x<y\) and \(h>0\) such that \(x, y+h\in I\), we have \[ f(x+h)-f(x)\leq f(y+h)-f(y) \] to prove that \(f\) is Wright-convex if and only if the inequality \[ f(x_1) +\ldots +f(x_n) \leq f(y_1) +\ldots +f(y_n) \] holds for any \(\bar{x}, \bar{y}\in {\mathbb{R}}^n\) such that \(\bar{x}\prec \bar{y}\) and \(x_i , y_i \in I\) for \(i\in \{1,\ldots , n\}\) (see [\textit{C. T. Ng}, General inequalities 5, 5th Int. Conf., Oberwolfach/FRG 1986, ISNM 80, 433--438 (1987; Zbl 0634.39014)] for the original proof of this fact).