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.

222 <strong>Convex</strong> Bodies<br />

To show that (K + x) ∩ (K + y) ⊆ K + x ∨ y, letw ∈ (K + x) ∩ (K + y).<br />

Then w ∈ K + x <strong>and</strong> w ∈ K + y <strong>and</strong> thus x, y � w <strong>and</strong> therefore x ∨ y � w, or<br />

w ∈ K + x ∨ y. To show the reverse inclusion, let w ∈ K + x ∨ y. Then x ∨ y � w<br />

<strong>and</strong> thus x, y � w. This yields w ∈ (K + x) ∩ (K + y). The proof of (4) is complete.<br />

Choose a hyperplane H as in the remarks before the theorem <strong>and</strong> let C = H ∩ K .<br />

Then the following statement holds:<br />

(5) Let λ, µ ≥ 0 <strong>and</strong> x, y ∈ E d with (λC + x) ∩ (µC + y) �= ∅. Then<br />

(λC + x) ∩ (µC + y) = νC + x ∨ y for suitable ν ≥ 0.<br />

Let G be the hyperplane parallel to H which contains λC + x <strong>and</strong> µC + y. Then<br />

<strong>and</strong> thus,<br />

λC + x = (K + x) ∩ G, µC + y = (K + y) ∩ G<br />

(λC + x) ∩ (µC + y) = (K + x) ∩ (K + y) ∩ G<br />

mm = (K + x ∨ y) ∩ G = νC + x ∨ yfor suitableν ≥ 0<br />

by (4). This concludes the proof of (5).<br />

(5) Together with Theorem 12.2 of Choquet <strong>and</strong> Rogers <strong>and</strong> Shephard implies<br />

that C is a simplex. Hence K is a simplicial cone. To see that dim K = d, consider<br />

a basis {b1,...,bd} of E d <strong>and</strong> note that for z = o ∧ b1 ∧···∧bd the cone K + z<br />

contains o, b1,...,bd. Thus dim K = dim(K + z) = d.<br />

(ii)⇒(i) This is easy to prove on noting that after a suitable linear transformation<br />

we may assume that K ={x : 0 ≤ xi}. ⊓⊔<br />

Remark. Choquet actually proved his result in infinite dimensions using Choquet<br />

simplices instead of (finite dimensional) simplices. A generalization to vector spaces<br />

without any topology is due to Kendall [574]. See the surveys by Peressini [789] <strong>and</strong><br />

Rosenthal [857].<br />

12.2 A Characterization of Balls by Their Gravitational Fields<br />

During the twentieth century a multitude of different characterizations of (Euclidean)<br />

balls <strong>and</strong> spheres have been given. Besides elementary characterizations, characterizations<br />

by extremal properties, in particular by properties of isoperimetric type,<br />

<strong>and</strong> other characterizations in the context of convexity, there is a voluminous body<br />

of differential geometric characterizations. Interesting sporadic characterizations of<br />

balls have their origin in other branches of mathematics, for example in potential<br />

theory <strong>and</strong>, even outside of mathematics.<br />

In the sequel we consider a characterization of balls by their Newtonian gravitational<br />

fields. We consider the case d = 3, but the result can be extended easily to any<br />

d ≥ 2, where, for d = 2, logarithmic potentials have to be used.<br />

Surveys of characterizations of balls are due to Bonnesen <strong>and</strong> Fenchel [149],<br />

Giering [377], Burago <strong>and</strong> Zalgaller [178], Bigalke [114] <strong>and</strong> Heil <strong>and</strong> Martini [488].<br />

We add two references, one related to cartography by <strong>Gruber</strong> [426] <strong>and</strong> one to<br />

electrostatics by Mendez <strong>and</strong> Reichel [719]. For a characterization of balls using<br />

topological tools see Montejano [750].

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

Saved successfully!

Ooh no, something went wrong!