Minkowski's second theorem

In mathematics, Minkowski's second theorem is a result in the geometry of numbers about the values taken by a norm on a lattice and the volume of its fundamental cell. It is named so because it is a strengthening of Minkowski's theorem.

Setting

Let K be a closed convex centrally symmetric body of positive finite volume in n-dimensional Euclidean space ℝn. The gauge or distance Minkowski functional g attached to K is defined by g(x)=inf⁡{λ∈ℝ:x∈λK}.

Conversely, given a norm g on ℝn we define K to be K={x∈ℝn:g(x)≤1}.

Let Γ be a lattice in ℝn. The successive minima of K or g on Γ are defined by setting the k-th successive minimum λk to be the infimum of the numbers λ such that λK contains k linearly-independent vectors of Γ. We have 0 < λ1 ≤ λ2 ≤ ... ≤ λn < ∞.

Statement

The successive minima satisfy 2nn!vol⁡(ℝn/Γ)≤λ1λ2⋯λnvol⁡(K)≤2nvol⁡(ℝn/Γ).

Proof

A basis of linearly independent lattice vectors b1, b2, ..., bn can be defined by g(bj) = λj (warning : it may be not a basis of the lattice, but only a basis of the ambient space).

The lower bound is proved by considering the convex polytope 2n with vertices at ±bj/ λj, which has an interior enclosed by K and a volume which is 2n / (n! λ1λ2...λn) times an integer multiple of a primitive cell of the lattice (as seen by scaling the polytope by λj along each basis vector to obtain 2n n-simplices with lattice point vectors), the integer being the index of the ℤ-span of the family (bi)i as a subgroup of the lattice.

To prove the upper bound, consider functions fj(x) sending points x in K to the centroid of the subset of points in K that can be written as x+∑i=1j−1aibi for some real numbers ai. Then the coordinate transform x′=h(x)=∑i=1n(λi−λi−1)fi(x)/2 has a Jacobian determinant J=λ1λ2…λn/2n. If p and q are in the interior of K and p−q=∑i=1kaibi(with ak≠0) then (h(p)−h(q))=∑i=0kcibi∈λkK with ck=λkak/2, where the inclusion in λkK (specifically the interior of λkK) is due to convexity and symmetry. But lattice points in the interior of λkK are, by definition of λk, always expressible as a linear combination of b1,b2,…bk−1, so any two distinct points of K′=h(K)={x′∣h(x)=x′} cannot be separated by a lattice vector.

The entire space ℝn is tiled by translated copies of a primitive cell ℝn=⋃v∈Γ((ℝn/Γ)+v). The intersections K'v:=K′∩((ℝn/Γ)+v) with K′ have a total volume in sum that is less than the volume of a single primite cell ℝn/Γ and therefore vol⁡(K)⋅J=vol⁡(K′)≤vol⁡(ℝn/Γ).To see this, consider two intersections K'u and K'v with u,v∈Γ and their translated versions K'u−u and K'v−v that are both contained in the primitive cell located at the origin. If they would intersect at some point y∈(ℝn/Γ), then the two original points y+u∈K'u⊆K′ and y+v∈K'v⊆K′ would be separated by a lattice vector, which can be seen by considering the difference (y+u)−(y+v)=u−v∈Γ. This cannot be the case and therefore, translating every of the intersections into the primitive cell located at the origin does not lead to any intersections, which shows the claim.