ProofComplete

Quadratic Residues and Reciprocity - Key Proof

We present Gauss's proof of quadratic reciprocity using Gauss's Lemma, one of the most elegant among the many known proofs. This approach reveals the geometric structure underlying the reciprocity law.

ProofGauss's Lemma

Let pp be an odd prime and gcd(a,p)=1\gcd(a, p) = 1. We prove that (ap)=(1)n\left(\frac{a}{p}\right) = (-1)^n, where nn is the number of least positive residues of a,2a,,p12aa, 2a, \ldots, \frac{p-1}{2}a that exceed p/2p/2.

By Euler's criterion: (ap)a(p1)/2(modp)\left(\frac{a}{p}\right) \equiv a^{(p-1)/2} \pmod{p}

Consider the product: P=a2a3ap12a=a(p1)/2p12!P = a \cdot 2a \cdot 3a \cdots \frac{p-1}{2}a = a^{(p-1)/2} \cdot \frac{p-1}{2}!

For each kaka where 1k(p1)/21 \leq k \leq (p-1)/2, let rkr_k be its least positive residue modulo pp. These residues can be partitioned into:

  • Those with rk<p/2r_k < p/2 (there are (p1)/2n(p-1)/2 - n of them)
  • Those with rk>p/2r_k > p/2 (there are nn of them)

For the larger residues, prk<p/2p - r_k < p/2. The crucial observation is that the multiset: {rk:rk<p/2}{prk:rk>p/2}\{r_k : r_k < p/2\} \cup \{p - r_k : r_k > p/2\} equals {1,2,,(p1)/2}\{1, 2, \ldots, (p-1)/2\} (possibly after reordering).

Therefore: P(1)np12!(modp)P \equiv (-1)^n \cdot \frac{p-1}{2}! \pmod{p}

Comparing with P=a(p1)/2p12!P = a^{(p-1)/2} \cdot \frac{p-1}{2}! and canceling p12!\frac{p-1}{2}!: a(p1)/2(1)n(modp)a^{(p-1)/2} \equiv (-1)^n \pmod{p}

By Euler's criterion, this gives (ap)=(1)n\left(\frac{a}{p}\right) = (-1)^n.

ProofLaw of Quadratic Reciprocity

Let pp and qq be distinct odd primes. We use Gauss's Lemma to count the relevant residues.

Consider the lattice rectangle R=[0,p/2]×[0,q/2]R = [0, p/2] \times [0, q/2] in R2\mathbb{R}^2. The diagonal line qxpy=0qx - py = 0 passes through (0,0)(0, 0) and (p/2,q/2)(p/2, q/2).

Key insight: The number of lattice points below the diagonal in RR (excluding boundary) equals the number from Gauss's Lemma.

Specifically, let μ\mu be the number of integers kk with 1k<p/21 \leq k < p/2 such that the fractional part of kq/pkq/p exceeds 1/21/2. This counts lattice points in the rectangle above the diagonal.

By geometric considerations and symmetry: μ=(p1)(q1)4(lattice points below diagonal)\mu = \frac{(p-1)(q-1)}{4} - \text{(lattice points below diagonal)}

Similarly, let ν\nu count the corresponding quantity with pp and qq interchanged.

The total number of interior lattice points in RR is: (p1)(q1)4=μ+ν\frac{(p-1)(q-1)}{4} = \mu + \nu

By Gauss's Lemma: (qp)=(1)μ,(pq)=(1)ν\left(\frac{q}{p}\right) = (-1)^\mu, \quad \left(\frac{p}{q}\right) = (-1)^\nu

Therefore: (pq)(qp)=(1)μ+ν=(1)(p1)(q1)/4\left(\frac{p}{q}\right)\left(\frac{q}{p}\right) = (-1)^{\mu + \nu} = (-1)^{(p-1)(q-1)/4}

This completes the proof.

Remark

This geometric proof, though originally due to Gauss in a different form, can be made rigorous by carefully counting lattice points. It reveals that quadratic reciprocity is fundamentally a statement about the geometry of lattices in the plane.

ProofFirst Supplement

We prove (1p)=(1)(p1)/2\left(\frac{-1}{p}\right) = (-1)^{(p-1)/2}.

By Euler's criterion: (1p)(1)(p1)/2(modp)\left(\frac{-1}{p}\right) \equiv (-1)^{(p-1)/2} \pmod{p}

Since both sides are ±1\pm 1, congruence modulo pp implies equality.

Alternatively, 1-1 is a quadratic residue if and only if the equation x21(modp)x^2 \equiv -1 \pmod{p} has a solution. The multiplicative group (Z/pZ)(\mathbb{Z}/p\mathbb{Z})^* is cyclic of order p1p-1. An element of order 44 exists if and only if 4(p1)4 \mid (p-1), i.e., p1(mod4)p \equiv 1 \pmod{4}.

If such an element gg exists, then g2g^2 has order 22, so g21(modp)g^2 \equiv -1 \pmod{p}.

ExampleVerification

For p=13p = 13: (p1)/2=6(p-1)/2 = 6 is even, so (113)=1\left(\frac{-1}{13}\right) = 1. Indeed, 52=251(mod13)5^2 = 25 \equiv -1 \pmod{13}.

For p=11p = 11: (p1)/2=5(p-1)/2 = 5 is odd, so (111)=1\left(\frac{-1}{11}\right) = -1. The equation x21(mod11)x^2 \equiv -1 \pmod{11} has no solution.

These proofs demonstrate the beautiful interplay between algebra, number theory, and geometry that makes quadratic reciprocity so profound.