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.

Proof. Consider the linear forms<br />

l1(u) = u1 + t1u2 +···+t d−1<br />

1<br />

ud<br />

l2(u) = u1 + t2u2 +···+t d−1<br />

2<br />

ud<br />

................................<br />

ld(u) = u1 + tdu2 +···+t d−1<br />

d ud,<br />

23 Successive Minima 375<br />

where t1,...,td are the roots of p. Foru ∈ Z d \{o}, the linear forms li(u),<br />

i = 1,...,d, all are different from 0, for, if li(u) = 0, then ti is a root of the<br />

polynomial li(u) of degree at most d − 1inti with integer coefficients. This contradicts<br />

the irreducibility of p. Therefore l1(u) ···ld(u) �= 0 <strong>and</strong>, since this product is<br />

a symmetric polynomial in t1,...,td with integer coefficients, it must be an integer.<br />

Thus<br />

|l1(u) ···ld(u)| ≥1foru ∈ Z d \{o}.<br />

The lattice L ={l = (l1(u),...,ld(u)) : u ∈ Z d } is thus admissible for the star set<br />

{x :|x1 ···xd| ≤1}.<br />

By the inequality of the geometric <strong>and</strong> arithmetic mean, this star set contains the<br />

o-symmetric cross-polytope<br />

O = � x : 1 �<br />

|x1|+···+|xd|<br />

d<br />

� ≤ 1 � ,<br />

of volume 2 d d d /d !. Thus L is also admissible for O. The fundamental theorem then<br />

shows that d(L) ≥ d d /d !. Noting that D = d(L) 2 , the proof is complete. ⊓⊔<br />

23 Successive Minima<br />

Successive minima of star bodies or convex bodies with respect to lattices were first<br />

defined <strong>and</strong> investigated by Minkowski in the context of the geometry of numbers.<br />

Minkowski put them to use in algebraic number theory. A hundred years later, successive<br />

minima still play a role in the geometry of numbers <strong>and</strong> in algebraic <strong>and</strong><br />

transcendental number theory. See, e.g. Bertr<strong>and</strong> [102], Chen [204] <strong>and</strong> Matveev<br />

[697], but they are also important in Diophantine Approximation, see, e.g. Schmidt<br />

[896], <strong>and</strong> in computational geometry, compare Lagarias, Lenstra <strong>and</strong> Schnorr [626],<br />

Schnorr [913] <strong>and</strong> Blömer [134]. There is a surprising link to Nevanlinna’s value distribution<br />

theory, a branch of complex analysis, see Wong [1028] <strong>and</strong> Hyuga [534].<br />

Relations to lattice polytopes <strong>and</strong> roots of Ehrhart polynomials were studied by<br />

Stanley, Henk, Schürmann <strong>and</strong> Wills, see the references in Sect. 19.1 <strong>and</strong> the survey<br />

of Henk <strong>and</strong> Wills [493].<br />

In this section, we first present Minkowski’s theorem on successive minima <strong>and</strong><br />

prove a result of Mahler relating successive minima of a convex body with respect<br />

to a lattice <strong>and</strong> successive minima of the polar body with respect to the polar lattice.<br />

Then Jarník’s transference theorem is proved. It connects lattice packing <strong>and</strong><br />

lattice covering of a given convex body. Finally, we give a result of Perron <strong>and</strong><br />

Khintchine on Diophantine approximation.

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

Saved successfully!

Ooh no, something went wrong!