7 Elliptic Curves over Finite Fields

Lemma 7.1. Assuming that:

  • A an abelian group

  • q:A a positive definit quadratic form

Then
|q(x+y)q(x)q(y)=x,y|2q(x)q(y)x,yA.

Proof. We may assume x0, otherwise the result is clear. So q(x)0.

Let m,n. Then

0q(mx+ny)=12mx+ny,mx+ny=m2q(x)+mnx,y+n2q(y)=q(x)(m+x,y2q(x)n)2+(q(y)x,y24q(x))n2

Take m=x,y, n=2q(x)0 to deduce

q(y)x,y24q(x)0

hence x,y24q(x)q(y) and hence

|x,y|2q(x)q(y).

Theorem 7.2 (Hasse’s Theorem). Assuming that:

Then
|#E(𝔽q)(q+1)|2q.

Proof. Recall Gal(𝔽qr𝔽q) is cyclic of order r and generated by Frobenius xxq.

Let E have Weierstrass equation with coefficients a1,,a6𝔽q (so aiq=ai). Define the Frobenius endomorphism

ϕ:EE(x,y)(xq,yq)

This is an isogeny of degree q.

Then

E(𝔽q)={PE|ϕ(P)=P}=ker(1ϕ).

ϕω=ϕ(dxy)=d(xq)yq=qxq1dxyq=0(q0(modp)) Lemma 6.3 tells us that
(1ϕ)ω=ωϕ(ω)=ω0.

Hence 1ϕ is separable.

By Theorem 2.8 and the fact that 1ϕ is a group homomorphism, we argue as in the proof of Theorem 6.5 that

# ker (1ϕ)=#E(𝔽q)= deg (1ϕ).

deg:Hom(E,E) is a positive definite quadratic form (Theorem 5.7, and positive definiteness is obvious since non-constant morphisms have positive degree).

Lemma 7.1 gives

|deg(1ϕ)1 deg ϕ|2 deg ϕ.

Hence

|#E(𝔽q)1q|2q.

Definition. For ϕ,ψEnd(E)=Hom(E,E), we put ϕ,ψ=deg(ϕ+ψ) deg ϕ deg ψ and tr(ϕ)=ϕ,1.

Corollary 7.3. Assuming that:

Then #E(𝔽q)=q+1tr(ϕ) and |tr(ϕ)|2q.

7.1 Zeta functions

For K a number field, let

ζK(s)=𝔞OK1(N𝔞)s=𝔭OKprime(11(N𝔭)s)1.

For K a function field, i.e. K=𝔽q(C) where C𝔽q is a smooth projective curve,

ζK(s)=x|C|(11(Nx)s)1,

where

|C|={closed points on C}

(closed points are orbits for action of Gal(𝔽q¯𝔽q) on C(𝔽q¯)) and Nx=qdegx, degx is the size of orbit.

We have ζK(s)=F(qs) for some F[[T]],

F(T)=x|C|(1T deg x)1logF(T)=x|C|m=11mTm deg xlog(1x)=m=1xmmTddT(logF(T))=x|C|m=1 deg xTm deg x=n=1(x|C| deg x|n deg x)Tnn=mdegx=n=1#C(𝔽qn)Tn

Therefore

F(T)=exp(n=1#C(𝔽qn)nTn).

Definition (Zeta function). The zeta function ZC(T) of a smooth projective curve C𝔽q is defined by

ZC(T)=exp(n=1#C(𝔽qn)nTn).

Theorem 7.4. Assuming that:

Then
ZE(T)=1aT+qT2(1T)(1qT).

Proof. Let ϕ:EE be the q power Frobenius map. By Corollary 7.3

#E(𝔽q)=q+1tr(ϕ).

Hence tr(ϕ)=a, deg(ϕ)=q.

Example Sheet 2, Q6(iii) implies ϕ2aϕ=q=0, hence ϕn=2aϕn+1+qϕn=0, so

tr(ϕn+2)atr(ϕn+1)+qtr(ϕn)=0.

This second order difference equation with initial conditions tr(1)=2, tr(ϕ)=a has solution

tr(ϕn)=αn+βn,

where α,β are roots of X2aX+q=0.

Again by Corollary 7.3,

#E(𝔽qn)=qn+1tr(ϕn)=1+qnαnβn

Therefore

ZE(T)=expn=1(Tnn+(qT)nn(αT)nn(βT)nn)=(1αT)(1βT)(1T)(1qT)=1aT+qT2(1T)(1qT)

Remark. Hasse’s Theorem tells us that |a|2q. α=β¯, and so |α|=|β|=q ().

Let K=𝔽q(E). ζK(s)=0, so ZE(qs)=0, so qs=α or β. Then qRe(s)=|α| or |β|, so by (), Re(s)=12.

“This is an analog of the Riemann hypothesis.”