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.

The induction is complete, concluding the proof of the identity (8).<br />

The second identity is as follows.<br />

7 Valuations 115<br />

(9) Let C1,...,Cm, D1,...,Dn ∈ S. Then<br />

φ � (C1 ∪···∪Cm) ∪ (D1 ∪···∪ Dn) �<br />

= �<br />

(−1)<br />

I<br />

|I |−1φ(CI ) + �<br />

(−1)<br />

I,J<br />

|I |+|J|−1φ(CI ∩ DJ ) + �<br />

(−1)<br />

J<br />

|J|−1φ(DJ )<br />

= φ(C1 ∪···∪Cm) + �<br />

(−1) |I |+|J|−1φ(CI ∩ DJ ) + φ(D1 ∪···∪ Dn).<br />

I,J<br />

Now add the identities (8) <strong>and</strong> (9) to get φ(C ∪ D) + φ(C ∩ D) = φ(C) + φ(D)<br />

for C, D ∈ L (S), concluding the proof of (6).<br />

Thus we have obtained a valuation on L(S) which extends the given valuation<br />

on S. Since each valuation on the lattice L(S), which extends the given valuation on<br />

S, satisfies (5), the extension is unique. ⊓⊔<br />

Remark. For far-reaching related results, see Groemer [402].<br />

Extension of Valuations on Boxes <strong>and</strong> <strong>Convex</strong> Polytopes<br />

A box in E d is a set of the form {x : αi ≤ xi ≤ βi}. Itsedge-lengths are βi −<br />

αi, i = 1,...,d. Let B = B(E d ) be the space of boxes in E d . For the definition<br />

of convex polytopes, see Sects. 6.2, 14.1 <strong>and</strong> let P = P(E d ) be the space of convex<br />

polytopes in E d .BothB <strong>and</strong> P are intersectional families of sets. The following<br />

polytope extension theorem is due to Voll<strong>and</strong> [1011].<br />

Theorem 7.2. Let φ be a valuation on B or P. Then φ satisfies the inclusion–<br />

exclusion formula on B, respectively, P <strong>and</strong> thus can be extended uniquely to a<br />

valuation on L(B), the space of polyboxes, respectively, L(P), the space of polyconvex<br />

polytopes by Theorem 7.1.<br />

Proof. We consider only the case of convex polytopes, the case of boxes being analogous.<br />

By Voll<strong>and</strong>’s first extension theorem, it is sufficient to show that φ satisfies<br />

the inclusion–exclusion formula on P:<br />

(10) Let P, P1,...,Pm ∈ P(Ed ) such that P = P1 ∪···∪ Pm. Then<br />

φ(P) = �<br />

I<br />

(−1) |I |−1 φ(PI ).<br />

The proof of (10) is by double induction on d <strong>and</strong> m. (10) holds trivially for d =<br />

0, 1,...,<strong>and</strong> m = 1 <strong>and</strong> for d = 0 <strong>and</strong> m = 1, 2,... Assume now that d > 0 <strong>and</strong><br />

m > 1 <strong>and</strong> that (10) holds for dimensions 0, 1,...,d − 1 <strong>and</strong> in dimension d for<br />

1, 2,...,m − 1 polytopes. We have to establish it for d <strong>and</strong> m.<br />

In the first step of the proof, we show that (10) holds in the following special<br />

cases:<br />

(i) One of the polytopes P1,...,Pm has dimension less than d, sayPm.<br />

(ii) One of the polytopes P1,...,Pm coincides with P, sayP1.

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

Saved successfully!

Ooh no, something went wrong!