Is there any characterization of strictly convex functions in terms of convex function, strongly convex function, or the epigraph

convex optimizationconvex-analysisconvex-geometryreal-analysis

The definition of strict convexity is given in terms of the Jensen's inequality.

However, I am seeking alternative definition of strict convexity, either

  1. stated in terms of the epigraph. For example, a function is convex if its epigraph is convex, by analogy, a function is strictly convex if its epigraph is ??

  2. in terms of convex, or strongly convex function. For example, a function is strongly convex if $f(x) – \frac{1}{2}\|x\|_2^2$ is convex, by analogy, a function is strictly convex if ?? is convex/stronglyconvex.

Can we fill in the blank for ?? ?

If not, does there exist any other alternative characterization of strictly convex functions?

Best Answer

As far as I remember, the answer to 1. is also "strictly convex" (meaning that for every point of the boundary on the set there is a hyperplane that has the whole set (except this one point) on one side).

For 2. there is no answer I know of. (The function $x\mapsto \sqrt{x^2+1}$ is strictly convex, but somehow becomes less so for larger $x$.)