Read · 2 min

Here, we proved Minkowski's theorem. Here's the statement again:

Minkowski's Convex Body Theorem In an nn-dimensional lattice spanned by the nn vectors

ν1=(k(1,1),k(1,2),⋯ ,k(1,n))ν2=(k(2,1),k(2,2),⋯ ,k(2,n))⋮νn=(k(n,1),k(n,2),⋯ ,k(n,n)),\begin{gather} \nu_1=(k_{(1,1)},k_{(1,2)},\cdots,k_{(1,n)})\\ \nu_2=(k_{(2,1)},k_{(2,2)},\cdots,k_{(2,n)})\\ \vdots\\ \nu_n=(k_{(n,1)},k_{(n,2)},\cdots,k_{(n,n)})\\ \end{gather}, let Δ=∣k(1,1)k(1,2)⋯k(1,n)k(2,1)k(2,2)⋯k(2,n)⋮⋮⋱⋮k(n,1)k(n,2)⋯k(n,n)∣\Delta=\left\lvert\begin{matrix} k_{(1,1)}&k_{(1,2)}&\cdots&k_{(1,n)}\\ k_{(2,1)}&k_{(2,2)}&\cdots&k_{(2,n)}\\ \vdots&\vdots&\ddots&\vdots\\ k_{(n,1)}&k_{(n,2)}&\cdots&k_{(n,n)}\\ \end{matrix}\right\rvert.

Then, if we have a convex region RR symmetric about the origin with volume greater than 2nΔ2^n \Delta, then there exists a non-zero point of the lattice in RR.

It turns out, that we can show that any integer is ths sum of four squares using this!

First, consider the following lemma:


Lemma For an odd prime pp, there exists a,ba,b such that a2≡−1−b2(modp)a^2\equiv -1-b^2\pmod{p}.

For the sake of contradiction, suppose that 1−b21-b^2 is never a quadratic residue modulo pp. Note that there are p−12\frac{p-1}{2} non-zero square modulo pp, so there are p+12\frac{p+1}{2} squares modulo pp. Let them be {s1,s2,⋯ ,sp+12}\left\{s_1,s_2,\cdots,s_{\frac{p+1}{2}}\right\}. Let's subtract them all from 1. We know that since all sis_i are distinct, all −1−si-1-s_i are distinct, and consequently there are p+12\frac{p+1}{2} distinct values for 1−b21-b^2. However, there are only p−12\frac{p-1}{2} quadratic non-residues, so the values sis_i cannot be all distinct, which is a contradiction. Thus, there exists a,ba,b such that a2≡1−b2(modp)a^2\equiv 1-b^2\pmod{p}.

Note: This holds for all n−b2n-b^2, not just −1-1

Theorem Every positive integer is the sum of four squares.

Proof From Euler's four-square identity, if we show that every prime is the sum of four squares, then we can 'construct' all the other integers as the sum of four squares.

By our lemma, there exists x,yx,y such that x2+y2≡−1(modp)x^2+y^2\equiv -1\pmod{p}. Thus, consider the lattice spanned by (p,0,0,0)(p,0,0,0), (0,p,0,0)(0,p,0,0), (x,y,1,0)(x,y,1,0), and (y,−x,0,1)(y,-x,0,1). Note that every point on the lattice is a multiple of pp away from the origin: a(p,0,x,y)+b(0,p,y,−x)+c(0,0,1,0)+d(0,0,0,1)=(ap+cx+dy,bp+cy−dx,c,d)a(p,0,x,y)+b(0,p,y,-x)+c(0,0,1,0)+d(0,0,0,1)=(ap+cx+dy,bp+cy-dx,c,d) ∣(ap+cx+dy,bp+cy−dx,c,d)∣≡(ap+cx+dy)2+(bp+cy−dx)2+(c)2+(d)2≡c(x2+y2+1)+d(x2+y2+1)≡0(modp)\begin{align} |(ap+cx+dy,bp+cy-dx,c,d)|&\equiv (ap+cx+dy)^2+(bp+cy-dx)^2+(c)^2+(d)^2\\ &\equiv c(x^2+y^2+1)+d(x^2+y^2+1)\\ &\equiv 0\pmod{p} \end{align} Now, consider the set of points (w,x,y,z)(w,x,y,z) such that 0≤w2+x2+y2+z2<2p0\leq w^2+x^2+y^2+z^2<2p. This is a convex region symmetric about the origin. Additionally, we have that Δ=∣p0000p00xy10y−x01∣=p2\Delta=\left\lvert\begin{matrix}p&0&0&0\\ 0&p&0&0\\ x&y&1&0\\ y&-x&0&1\end{matrix}\right\rvert=p^2 Since the volume of a 4-dimensional sphere with radius rr is 12π2r4\frac{1}{2}\pi^2r^4, we also know that this is a 4-dimensional sphere with radius 2p\sqrt{2p} and volume 2π2p22\pi^2p^2. Since π2>8\pi^2>8, we have 2π2p2>16p2,2\pi^2p^2>16p^2, so by Minkowski's theorem, there exists (w,x,y,z)∈Z4(w,x,y,z)\in\mathbb{Z}^4 with p∣w2+x2+y2+z2p\mid w^2+x^2+y^2+z^2 and 0<w2+x2+y2+z2<2p0<w^2+x^2+y^2+z^2<2p. Thus, w2+x2+y2+z2=pw^2+x^2+y^2+z^2=p, which is what we wanted to show. □\hspace{1cm}\Box

Type a concept, like “slope” or “photosynthesis”.

↑ ↓ to choose · Enter to openSearch explanations and problems