0
$\begingroup$

If $ S \subseteq \mathbb{R}^n$ is finite, show that conv(S) is a closed set. Is the statement still true if S is not finite? Where conv(S) is the convex hull of S.

From what I've read, the convex hull's aren't necessarily closed. Why would finiteness be a sufficient condition for this?

$\endgroup$
1
  • $\begingroup$ By the way, this is true for compact $S$, not just finite; this is usually proven as a corollary of Carathéodory's theorem. $\endgroup$
    – user21467
    Nov 17, 2014 at 22:00

2 Answers 2

0
$\begingroup$

Take $S$ to be an open half-plane. Then $\operatorname{Conv}(S) = S$ isn't closed.

Now assume $S \subseteq \mathbb{R}^n$ is finite, say $S = \{s_1,\dots,s_p\}$. Let $\Delta = \{(x_1,\dots,x_p) \mid x_1, \dots, x_p \geq 0, x_1 + \dots + x_p = 1\}$, and define $f: \mathbb{R}^p \to \mathbb{R}^n$ by $f(x_1,\dots,x_p) = x_1s_1 +\dots + x_ps_p$. Then $\operatorname{Conv}(S) = f(\Delta)$. Since $\Delta$ is compact and $f$ continuous (in fact linear), $\operatorname{Conv}(S)$ must be compact.

$\endgroup$
2
  • $\begingroup$ Thanks! Should $f: \mathbb{R}^p \rightarrow \mathbb{R}$ ? $\endgroup$ Nov 13, 2014 at 2:40
  • $\begingroup$ No, $s_1, \dots, s_p$ are points in $\mathbb{R}^n$. $\endgroup$
    – Mike
    Nov 13, 2014 at 2:48
0
$\begingroup$

For an infinite set whose convex hull is not closed, consider the set of points $$S=\left\{\left(x,e^{-x^2}\right):x\in\Bbb R\right\}$$ in $\Bbb R^2.$ This example can be readily generalized to $\Bbb R^n$ for any $n\ge 2.$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.