14.02.2013 Views

Gruber P. Convex and Discrete Geometry

Gruber P. Convex and Discrete Geometry

Gruber P. Convex and Discrete Geometry

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

44 <strong>Convex</strong> Bodies<br />

The next result is a simple consequence of Carathéodory’s theorem.<br />

Corollary 3.1. Let A ⊆ E d be compact. Then conv A is compact.<br />

Proof. The set<br />

� (λ1,...,λd+1, x1,...,xd+1) : λi ≥ 0,λ1 +···+λd+1 = 1, x j ∈ A �<br />

is a compact subset of Ed+1+(d+1)d = E (d+1)2 . Hence its image under the continuous<br />

mapping<br />

of E (d+1)2<br />

(λ1,...,λd+1, x1,...,xd+1) → λ1x1 +···+λd+1xd+1<br />

into E d is also compact. By Carathéodory’s theorem this image is conv A.<br />

⊓⊔<br />

Closure <strong>and</strong> Interior<br />

In the following some useful minor results on closure <strong>and</strong> interior of convex sets <strong>and</strong><br />

on convex hulls of subsets of E d are presented. Let C ⊆ E d be convex. By relint C<br />

we mean the interior of C relative to the affine hull aff C of C, i.e. the smallest flat in<br />

E d containing C.<br />

Proposition 3.1. Let C ⊆ E d be convex. Then the following statements hold:<br />

(i) cl C is convex.<br />

(ii) relint C is convex.<br />

(iii) C ⊆ cl relint C.<br />

Let B d denote the solid Euclidean unit ball in E d .<br />

Proof. (i) To show that cl C is convex, let x, y ∈ cl C <strong>and</strong> 0 ≤ λ ≤ 1. Choose<br />

sequences (xn), (yn) in C such that xn → x, yn → y as n →∞. The convexity of<br />

C implies that (1 − λ)xn + λyn ∈ C. Since (1 − λ)xn + λyn → (1 − λ)x + λy, it<br />

follows that (1 − λ)x + λy ∈ cl C.<br />

(ii) It is sufficient to consider the case where int C �= ∅<strong>and</strong> to show that int C is<br />

convex. Let x, y ∈ int C <strong>and</strong> 0 ≤ λ ≤ 1. Choose δ>0 such that x +δB d , y +δB d ⊆<br />

int C ⊆ C. Then<br />

(1 − λ)x + λy + δB d = (1 − λ)(x + δB d ) + λ(y + δB d ) ⊆ C<br />

by the convexity of B d <strong>and</strong> C. Thus (1 − λ)x + λy ∈ int C.<br />

(iii) It is sufficient to consider the case where int C �= ∅<strong>and</strong> to prove that C ⊆<br />

cl int C.Letx ∈ C <strong>and</strong> choose y ∈ int C. Then there is a δ>0 such that y + δB d ⊆<br />

int C ⊆ C. The convexity of C now implies that<br />

(1 − λ)x + λy + λδB d = (1 − λ)x + λ(y + δB d ) ⊆ C.<br />

Thus (1 − λ)x + λy ∈ int C for 0

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!