Two Conditions for a Function to be Convex
Journal of convex analysis, Tome 20 (2013) no. 4, pp. 1189-1201
Voir la notice de l'article provenant de la source Heldermann Verlag
We present two sufficient conditions in order that a real function on a finite-dimensional normed space be convex (Theorems 1 and 2) and show some consequences of them. In particular, it comes out that a real function $f$ on a finite-dimensional Hilbert space $X$ is convex, provided that $f$ has the property that for each point $y \in X$ and each $\lambda > 0$ the real function $X \ni x \to \lambda f(x) + \|x-y\|^2$ has a unique global minimum.
@article{JCA_2013_20_4_JCA_2013_20_4_a16,
author = {A. O. Caruso and A. Villani},
title = {Two {Conditions} for a {Function} to be {Convex}},
journal = {Journal of convex analysis},
pages = {1189--1201},
publisher = {mathdoc},
volume = {20},
number = {4},
year = {2013},
url = {http://geodesic.mathdoc.fr/item/JCA_2013_20_4_JCA_2013_20_4_a16/}
}
A. O. Caruso; A. Villani. Two Conditions for a Function to be Convex. Journal of convex analysis, Tome 20 (2013) no. 4, pp. 1189-1201. http://geodesic.mathdoc.fr/item/JCA_2013_20_4_JCA_2013_20_4_a16/