2
$\begingroup$

I've shown that the following result is valid for $V=\Bbb R^v$:

If $K\subset V$ is compact and convex and $V$ is a $v$-dimensional Banach space, then $K$ has at least one extreme points.

The proof I know uses the parallelogram law. Can we do something similar for general finite dimensional Banach spaces?

Obs.: The proof of Krein-Milman Theorem (for locally convex topological vector spaces) needs the existence of extreme points in general, and the proof of this uses Zorn's Lemma. I was wondering if there is a proof using something different, maybe similar to the parallelogram law...

$\endgroup$
1
$\begingroup$

A point being extreme (or not) has nothing to do with the norm or topology of the space; it's just a concept of linear algebra. Topology enters the picture only when we talk about $K$ being compact. But then again, on a finite-dimensional space all norms induce the same topology, so they choice of a norm does not matter.

Upshot: if $V$ is a finite-dimensional normed vector space and $K$ is its compact subset, consider the Euclidean norm on $V$ instead and take the point of $K$ with maximal norm. That's an extreme point, by the parallelogram law.

$\endgroup$
  • $\begingroup$ So, even when the finite dimensional space is not isometric to some euclidean, if we equip it with the euclidean norm, the parallelogram law is satisfied? $\endgroup$ – Filburt Aug 12 '17 at 20:07
  • $\begingroup$ Yes, that is the case. Although they are not isometric, they are isomorphic, so we can have an inner product (= parallelogram law) $\endgroup$ – Filburt Aug 12 '17 at 20:16
  • $\begingroup$ Simpler way to put it: forget the original norm and use Euclidean one instead. $\endgroup$ – user357151 Aug 12 '17 at 20:25

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.