2

Let $A$ be a convex closed subset of $\mathbb R^n$. Is there a convex function $g: \mathbb R^n \rightarrow \mathbb R$ such that

$ A=\{x: g(x) \leq 0\} ? $

user 531
  • 23
  • 2

1 Answers1

1

Assume without loss of generality that $A\neq\varnothing$.

Since $A$ is closed and non-empty $\min_{y\in A}\|x-y\|$ exists for any $x\in\mathbb{R}^n$ and we define $g(x) = \min_{y\in A}\|x-y\|$. Clearly $g(x) = 0$ for $x\in A$ and $g(x)>0$ for $x\notin A$ and it remains to show that $g$ is convex.

Let $x,y\in\mathbb{R}^n$ and let $t\in[0,1]$ then there exist $x^\prime,y^\prime\in A$ such that $\|x-x^\prime\|=g(x)$ and such that $\|y-y^\prime\|=g(y)$.

Since $A$ is convex, $tx^\prime+(1-t)y^\prime\in A$ and $g(tx+(1-t)y)$ is less than $$\|[tx+(1-t)y] - [tx^\prime+(1-t)y^\prime]\| = \|t(x-x^\prime) + (1-t)(y-y^\prime)\|,$$ which by the triangle inequality $$\leq t\|x-x^\prime\| + (1-t)\|y-y^\prime\| = tg(x)+(1-t)g(y).$$ Hence $g$ is indeed convex.

Abel
  • 7,312