ConceptComplete

Elliptic Curves

An elliptic curve is a smooth projective curve of genus 11 equipped with a distinguished rational point. They are simultaneously the simplest abelian varieties and among the richest objects in arithmetic and algebraic geometry.


Definition and Weierstrass form

Definition4.16Elliptic curve

An elliptic curve over a field kk is a pair (E,O)(E, O) where EE is a smooth projective curve of genus 11 over kk and O∈E(k)O \in E(k) is a specified kk-rational point (the origin).

Every elliptic curve admits a Weierstrass equation: an embedding Eβ†ͺP2E \hookrightarrow \mathbb{P}^2 as a cubic curve

y2+a1xy+a3y=x3+a2x2+a4x+a6,ai∈k,y^2 + a_1 xy + a_3 y = x^3 + a_2 x^2 + a_4 x + a_6, \quad a_i \in k,

with O=[0:1:0]O = [0:1:0] (the point at infinity on the yy-axis).

If char⁑(k)β‰ 2,3\operatorname{char}(k) \neq 2, 3, we can complete the square and cube to get the short Weierstrass form:

y2=x3+Ax+B,A,B∈k,4A3+27B2β‰ 0.y^2 = x^3 + Ax + B, \quad A, B \in k, \quad 4A^3 + 27B^2 \neq 0.

The condition 4A3+27B2β‰ 04A^3 + 27B^2 \neq 0 ensures smoothness (i.e., the discriminant Ξ”=βˆ’16(4A3+27B2)β‰ 0\Delta = -16(4A^3 + 27B^2) \neq 0).

ExampleConcrete elliptic curves
  • E:y2=x3βˆ’xE: y^2 = x^3 - x over Q\mathbb{Q}: Here A=βˆ’1A = -1, B=0B = 0, Ξ”=64β‰ 0\Delta = 64 \neq 0. The curve has complex multiplication by Z[i]\mathbb{Z}[i] (the Gaussian integers), since [i]:(x,y)↦(βˆ’x,iy)[i]: (x, y) \mapsto (-x, iy) is an endomorphism.

  • E:y2=x3βˆ’1E: y^2 = x^3 - 1 over Q\mathbb{Q}: Here A=0A = 0, B=βˆ’1B = -1, Ξ”=βˆ’432β‰ 0\Delta = -432 \neq 0. Has CM by Z[Ο‰]\mathbb{Z}[\omega] where Ο‰=e2Ο€i/3\omega = e^{2\pi i/3}, via [Ο‰]:(x,y)↦(Ο‰x,y)[\omega]: (x,y) \mapsto (\omega x, y).

  • E:y2=x3+x+1E: y^2 = x^3 + x + 1 over F5\mathbb{F}_5: The points are O,(0,Β±1),(2,Β±1),(3,Β±1),(4,Β±2)O, (0, \pm 1), (2, \pm 1), (3, \pm 1), (4, \pm 2), so ∣E(F5)∣=9|E(\mathbb{F}_5)| = 9.

  • E:y2=x(xβˆ’1)(xβˆ’Ξ»)E: y^2 = x(x-1)(x-\lambda) for Ξ»β‰ 0,1\lambda \neq 0, 1: the Legendre family, a universal family of elliptic curves with full level-2 structure. The jj-invariant is j=256(Ξ»2βˆ’Ξ»+1)3Ξ»2(Ξ»βˆ’1)2j = 256 \frac{(\lambda^2 - \lambda + 1)^3}{\lambda^2(\lambda - 1)^2}.


The group law

Definition4.17Group law on an elliptic curve

The set E(k)E(k) of kk-rational points carries a natural abelian group structure with identity OO:

Geometric description (chord-tangent): Given P,Q∈E(k)P, Q \in E(k):

  1. Draw the line LL through PP and QQ (tangent line if P=QP = Q).
  2. LL meets EE in a third point RR (by Bezout: a line meets a cubic in 3 points).
  3. Define P+QP + Q as the reflection of RR over the xx-axis: if R=(x0,y0)R = (x_0, y_0), then P+Q=(x0,βˆ’y0βˆ’a1x0βˆ’a3)P + Q = (x_0, -y_0 - a_1 x_0 - a_3).

Divisor-theoretic description: Under the Abel–Jacobi isomorphism E(k)β†’βˆΌPic⁑0(E)E(k) \xrightarrow{\sim} \operatorname{Pic}^0(E), P↦[Pβˆ’O]P \mapsto [P - O]:

P+Q=Rβ€…β€ŠβŸΊβ€…β€Š[Pβˆ’O]+[Qβˆ’O]=[Rβˆ’O]∈Pic⁑0(E).P + Q = R \iff [P - O] + [Q - O] = [R - O] \in \operatorname{Pic}^0(E).

Equivalently, P+Q+Rβ€²βˆΌ3OP + Q + R' \sim 3O where Rβ€²R' is the third intersection and P+Q=P + Q = reflection of Rβ€²R'.

ExampleExplicit addition formulas

For E:y2=x3+Ax+BE: y^2 = x^3 + Ax + B in short Weierstrass form, P=(x1,y1)P = (x_1, y_1), Q=(x2,y2)Q = (x_2, y_2):

Case Pβ‰ QP \neq Q: The slope of PQPQ is Ξ»=y2βˆ’y1x2βˆ’x1\lambda = \frac{y_2 - y_1}{x_2 - x_1}. Then:

x3=Ξ»2βˆ’x1βˆ’x2,y3=Ξ»(x1βˆ’x3)βˆ’y1.x_3 = \lambda^2 - x_1 - x_2, \quad y_3 = \lambda(x_1 - x_3) - y_1.

Case P=QP = Q (doubling): Ξ»=3x12+A2y1\lambda = \frac{3x_1^2 + A}{2y_1}. Same formulas for x3,y3x_3, y_3.

Negation: βˆ’P=(x1,βˆ’y1)-P = (x_1, -y_1).

Numerical example on y2=x3βˆ’5x+8y^2 = x^3 - 5x + 8: Let P=(1,2)P = (1, 2) and Q=(βˆ’1,βˆ’23)Q = (-1, -2\sqrt{3})... Actually, let's use P=(1,2)P = (1, 2) (check: 4=1βˆ’5+8=44 = 1 - 5 + 8 = 4 βœ“). Doubling: Ξ»=3βˆ’54=βˆ’12\lambda = \frac{3 - 5}{4} = -\frac{1}{2}, so x3=14βˆ’2=βˆ’74x_3 = \frac{1}{4} - 2 = -\frac{7}{4}, y3=βˆ’12(1+74)βˆ’2=βˆ’12β‹…114βˆ’2=βˆ’278y_3 = -\frac{1}{2}(1 + \frac{7}{4}) - 2 = -\frac{1}{2} \cdot \frac{11}{4} - 2 = -\frac{27}{8}.

Check: (βˆ’278)2=72964\left(-\frac{27}{8}\right)^2 = \frac{729}{64} and (βˆ’74)3βˆ’5(βˆ’74)+8=βˆ’34364+354+8=βˆ’34364+56064+51264=72964\left(-\frac{7}{4}\right)^3 - 5\left(-\frac{7}{4}\right) + 8 = -\frac{343}{64} + \frac{35}{4} + 8 = -\frac{343}{64} + \frac{560}{64} + \frac{512}{64} = \frac{729}{64} βœ“

ExampleTorsion points

On E:y2=x3βˆ’xE: y^2 = x^3 - x over Q\mathbb{Q}:

  • 2-torsion (2P=O2P = O): these are points with y=0y = 0, i.e., (0,0)(0,0), (1,0)(1,0), (βˆ’1,0)(-1,0). So E[2](Q)β‰…(Z/2)2E[2](\mathbb{Q}) \cong (\mathbb{Z}/2)^2.

  • Over QΛ‰\bar{\mathbb{Q}}: E[n]β‰…(Z/n)2E[n] \cong (\mathbb{Z}/n)^2 for all nβ‰₯1n \geq 1. This is a free Z/n\mathbb{Z}/n-module of rank 22.

  • Mordell–Weil theorem: E(Q)β‰…ZrβŠ•E(Q)torsE(\mathbb{Q}) \cong \mathbb{Z}^r \oplus E(\mathbb{Q})_{\text{tors}}. For this curve: E(Q)β‰…(Z/2)2E(\mathbb{Q}) \cong (\mathbb{Z}/2)^2 (rank 00, torsion only).

  • Mazur's theorem: For E/QE/\mathbb{Q}, the torsion subgroup E(Q)torsE(\mathbb{Q})_{\text{tors}} is isomorphic to Z/n\mathbb{Z}/n for 1≀n≀101 \leq n \leq 10 or n=12n = 12, or Z/2Γ—Z/2m\mathbb{Z}/2 \times \mathbb{Z}/2m for 1≀m≀41 \leq m \leq 4.


The jj-invariant and isomorphism classes

Definition4.18j-invariant

The jj-invariant of E:y2=x3+Ax+BE: y^2 = x^3 + Ax + B is

j(E)=1728β‹…4A34A3+27B2=βˆ’1728β‹…(4A)3Ξ”.j(E) = 1728 \cdot \frac{4A^3}{4A^3 + 27B^2} = -1728 \cdot \frac{(4A)^3}{\Delta}.

Two elliptic curves over an algebraically closed field kˉ\bar{k} are isomorphic if and only if they have the same jj-invariant.

The jj-invariant defines a map j:M1,1β†’A1j: \mathcal{M}_{1,1} \to \mathbb{A}^1 which is an isomorphism of coarse moduli spaces. Every j0∈kΛ‰j_0 \in \bar{k} is realized by some elliptic curve.

ExampleNotable j-invariants
  • j=0j = 0: E:y2=x3+1E: y^2 = x^3 + 1 (A=0A = 0). Has automorphism group Z/6\mathbb{Z}/6 (generated by (x,y)↦(Ο‰x,βˆ’y)(x,y) \mapsto (\omega x, -y), Ο‰3=1\omega^3 = 1). CM by Z[Ο‰]\mathbb{Z}[\omega], Ο‰=e2Ο€i/3\omega = e^{2\pi i/3}.

  • j=1728j = 1728: E:y2=x3+xE: y^2 = x^3 + x (B=0B = 0). Has Aut⁑(E)β‰…Z/4\operatorname{Aut}(E) \cong \mathbb{Z}/4 (generated by (x,y)↦(βˆ’x,iy)(x,y) \mapsto (-x, iy)). CM by Z[i]\mathbb{Z}[i].

  • jβ‰ 0,1728j \neq 0, 1728: Aut⁑(E)={Β±1}β‰…Z/2\operatorname{Aut}(E) = \{\pm 1\} \cong \mathbb{Z}/2 (the generic case).

  • j=βˆ’215β‹…3β‹…53j = -2^{15} \cdot 3 \cdot 5^3: corresponds to the curve X0(11):y2+y=x3βˆ’x2βˆ’10xβˆ’20X_0(11): y^2 + y = x^3 - x^2 - 10x - 20, the modular elliptic curve of conductor 1111.

  • Over Fp\mathbb{F}_p: the supersingular elliptic curves form a finite set. For pβ‰₯5p \geq 5, EE is supersingular iff ∣E(Fp)∣=p+1|E(\mathbb{F}_p)| = p + 1 iff the Hasse invariant vanishes. There are approximately p/12p/12 supersingular jj-invariants.


Elliptic curves over C\mathbb{C}: lattices and uniformization

Definition4.19Complex uniformization

Over C\mathbb{C}, every elliptic curve is isomorphic to C/Ξ›\mathbb{C}/\Lambda for a lattice Ξ›=ZΟ‰1+ZΟ‰2βŠ‚C\Lambda = \mathbb{Z}\omega_1 + \mathbb{Z}\omega_2 \subset \mathbb{C} (with Ο‰1/Ο‰2βˆ‰R\omega_1/\omega_2 \notin \mathbb{R}).

The isomorphism is given by the Weierstrass β„˜\wp-function:

β„˜(z)=1z2+βˆ‘Ο‰βˆˆΞ›βˆ–0(1(zβˆ’Ο‰)2βˆ’1Ο‰2)\wp(z) = \frac{1}{z^2} + \sum_{\omega \in \Lambda \setminus 0} \left(\frac{1}{(z - \omega)^2} - \frac{1}{\omega^2}\right)

via z↦[β„˜(z):β„˜β€²(z):1]z \mapsto [\wp(z) : \wp'(z) : 1]. This satisfies (β„˜β€²)2=4β„˜3βˆ’g2β„˜βˆ’g3(\wp')^2 = 4\wp^3 - g_2 \wp - g_3 where g2=60G4g_2 = 60 G_4, g3=140G6g_3 = 140 G_6 are Eisenstein series.

The moduli space: Ο„=Ο‰1/Ο‰2∈H\tau = \omega_1/\omega_2 \in \mathfrak{H} (upper half-plane), and C/Λτ≅C/Λτ′\mathbb{C}/\Lambda_\tau \cong \mathbb{C}/\Lambda_{\tau'} iff Ο„β€²=aΟ„+bcΟ„+d\tau' = \frac{a\tau + b}{c\tau + d} for some (abcd)∈SL⁑2(Z)\begin{pmatrix} a & b \\ c & d \end{pmatrix} \in \operatorname{SL}_2(\mathbb{Z}). So the moduli space is H/SL⁑2(Z)\mathfrak{H}/\operatorname{SL}_2(\mathbb{Z}).

ExampleLattices and elliptic curves
  • Square lattice Ξ›=Z+Zi\Lambda = \mathbb{Z} + \mathbb{Z}i: gives Ο„=i\tau = i, j=1728j = 1728. The curve y2=x3+xy^2 = x^3 + x. The 4-fold symmetry z↦izz \mapsto iz of the lattice corresponds to (x,y)↦(βˆ’x,iy)(x,y) \mapsto (-x, iy).

  • Hexagonal lattice Ξ›=Z+ZΟ‰\Lambda = \mathbb{Z} + \mathbb{Z}\omega, Ο‰=e2Ο€i/3\omega = e^{2\pi i/3}: gives Ο„=Ο‰\tau = \omega, j=0j = 0. The curve y2=x3+1y^2 = x^3 + 1. The 6-fold symmetry z↦ωzz \mapsto \omega z gives (x,y)↦(Ο‰x,y)(x,y) \mapsto (\omega x, y)... but the Weierstrass map mixes things up; the 6-fold symmetry descends to 3-fold on xx and sign flip on yy.

  • Ο„=2i\tau = 2i: a "tall" rectangular lattice. No extra symmetry, so jβ‰ 0,1728j \neq 0, 1728. The jj-invariant can be computed from q=e2Ο€iβ‹…2i=eβˆ’4Ο€q = e^{2\pi i \cdot 2i} = e^{-4\pi} via the qq-expansion j(Ο„)=qβˆ’1+744+196884q+β‹―j(\tau) = q^{-1} + 744 + 196884q + \cdots.

  • Degeneration: as Ο„β†’i∞\tau \to i\infty (lattice becomes "infinitely tall"), qβ†’0q \to 0 and jβ†’βˆžj \to \infty. The curve degenerates to a nodal cubic y2=x3+x2y^2 = x^3 + x^2 (a "Neron 1-gon").


Isogenies

Definition4.20Isogeny

An isogeny Ο•:E1β†’E2\phi: E_1 \to E_2 is a morphism of elliptic curves that sends O1↦O2O_1 \mapsto O_2. Equivalently, it is a surjective group homomorphism with finite kernel.

Key properties:

  • Every isogeny has finite kernel ker⁑(Ο•)βŠ‚E1(kΛ‰)\ker(\phi) \subset E_1(\bar{k}).
  • The degree of Ο•\phi equals ∣ker⁑(Ο•)∣|\ker(\phi)| (for separable isogenies; in general, deg⁑ϕ=∣kerβ‘Ο•βˆ£β‹…pi\deg \phi = |\ker \phi| \cdot p^i for some inseparable part).
  • The dual isogeny Ο•^:E2β†’E1\hat{\phi}: E_2 \to E_1 satisfies Ο•^βˆ˜Ο•=[n]\hat{\phi} \circ \phi = [n] (multiplication by n=deg⁑ϕn = \deg \phi) on E1E_1.
  • End⁑(E)=Hom⁑(E,E)\operatorname{End}(E) = \operatorname{Hom}(E, E) is the endomorphism ring: either Z\mathbb{Z}, an order in an imaginary quadratic field (CM case), or a maximal order in a quaternion algebra (supersingular case in char pp).
ExampleIsogenies
  • Multiplication by nn: [n]:Eβ†’E[n]: E \to E is an isogeny of degree n2n^2, with ker⁑([n])=E[n]β‰…(Z/n)2\ker([n]) = E[n] \cong (\mathbb{Z}/n)^2 (if (n,char⁑k)=1(n, \operatorname{char} k) = 1).

  • Frobenius: Over Fq\mathbb{F}_q, the Frobenius Ο•q:(x,y)↦(xq,yq)\phi_q: (x,y) \mapsto (x^q, y^q) is an isogeny of degree qq. The trace aq=q+1βˆ’βˆ£E(Fq)∣a_q = q + 1 - |E(\mathbb{F}_q)| satisfies Ο•q2βˆ’aqΟ•q+q=0\phi_q^2 - a_q \phi_q + q = 0 in End⁑(E)\operatorname{End}(E).

  • 2-isogenies from 2-torsion: If P0=(e,0)∈E[2]P_0 = (e, 0) \in E[2], there is a 2-isogeny Ο•:Eβ†’Eβ€²=E/⟨P0⟩\phi: E \to E' = E/\langle P_0 \rangle given by Velu's formulas. For E:y2=(xβˆ’e)(x2+ex+f)E: y^2 = (x - e)(x^2 + ex + f):

Ο•(x,y)=(x+fβˆ’e2xβˆ’e,β€…β€Šyβ‹…(xβˆ’e)2βˆ’(fβˆ’e2)(xβˆ’e)2).\phi(x, y) = \left(x + \frac{f - e^2}{x - e},\; y \cdot \frac{(x-e)^2 - (f - e^2)}{(x-e)^2}\right).

  • End⁑(E)\operatorname{End}(E) for CM curves: For E:y2=x3βˆ’xE: y^2 = x^3 - x, End⁑(E)β‰…Z[i]\operatorname{End}(E) \cong \mathbb{Z}[i] with i:(x,y)↦(βˆ’x,iy)i: (x,y) \mapsto (-x, iy). The element 1+i∈Z[i]1 + i \in \mathbb{Z}[i] gives a (1+i)(1+i)-isogeny of degree ∣1+i∣2=2|1+i|^2 = 2.

Arithmetic of elliptic curves

TheoremHasse bound

For an elliptic curve EE over Fq\mathbb{F}_q:

∣E(Fq)∣=q+1βˆ’aq,∣aqβˆ£β‰€2q.|E(\mathbb{F}_q)| = q + 1 - a_q, \quad |a_q| \leq 2\sqrt{q}.

This is the "Riemann hypothesis for elliptic curves": the zeta function Z(E/Fq,T)=1βˆ’aqT+qT2(1βˆ’T)(1βˆ’qT)Z(E/\mathbb{F}_q, T) = \frac{1 - a_q T + qT^2}{(1-T)(1-qT)} has its roots on the circle ∣T∣=qβˆ’1/2|T| = q^{-1/2}.

ExamplePoint counting
  • E:y2=x3+x+1E: y^2 = x^3 + x + 1 over F5\mathbb{F}_5: We computed ∣E(F5)∣=9|E(\mathbb{F}_5)| = 9. So a5=5+1βˆ’9=βˆ’3a_5 = 5 + 1 - 9 = -3. Check: βˆ£βˆ’3∣=3≀25β‰ˆ4.47|-3| = 3 \leq 2\sqrt{5} \approx 4.47 βœ“.

  • Over F7\mathbb{F}_7: for y2=x3+1y^2 = x^3 + 1, direct counting gives ∣E(F7)∣=7|E(\mathbb{F}_7)| = 7 (note: 7≑1(mod3)7 \equiv 1 \pmod{3}). So a7=1a_7 = 1. The curve is ordinary.

  • Supersingular over Fp\mathbb{F}_p (pβ‰₯5p \geq 5): ap=0a_p = 0, so ∣E(Fp)∣=p+1|E(\mathbb{F}_p)| = p + 1. Example: y2=x3+1y^2 = x^3 + 1 over F5\mathbb{F}_5: points are OO, (0,Β±1)(0, \pm 1), (2,Β±2)(2, \pm 2), (3,0)(3, 0). That's 6=5+16 = 5 + 1 βœ“, so this is supersingular over F5\mathbb{F}_5.

ExampleThe Birch and Swinnerton-Dyer conjecture

The BSD conjecture (one of the Clay Millennium Problems) relates the arithmetic and analytic data of E/QE/\mathbb{Q}:

ord⁑s=1L(E,s)=rank⁑E(Q),\operatorname{ord}_{s=1} L(E, s) = \operatorname{rank} E(\mathbb{Q}),

where L(E,s)=∏pLp(E,s)L(E, s) = \prod_p L_p(E, s) is the Hasse–Weil LL-function. The leading coefficient involves the regulator, the Tate–Shafarevich group III(E/Q)\text{III}(E/\mathbb{Q}), and other invariants.

Known cases:

  • If ord⁑s=1L(E,s)=0\operatorname{ord}_{s=1} L(E,s) = 0: then rank⁑E(Q)=0\operatorname{rank} E(\mathbb{Q}) = 0 (Kolyvagin, using Gross–Zagier).
  • If ord⁑s=1L(E,s)=1\operatorname{ord}_{s=1} L(E,s) = 1: then rank⁑E(Q)=1\operatorname{rank} E(\mathbb{Q}) = 1 (Gross–Zagier + Kolyvagin).
  • ord⁑s=1L(E,s)β‰₯2\operatorname{ord}_{s=1} L(E,s) \geq 2: wide open.

Elliptic curves and moduli

Definition4.21Modular curves

The modular curve Y(1)=H/SL⁑2(Z)Y(1) = \mathfrak{H}/\operatorname{SL}_2(\mathbb{Z}) parametrizes isomorphism classes of elliptic curves over C\mathbb{C} via τ↦C/(Z+ZΟ„)\tau \mapsto \mathbb{C}/(\mathbb{Z} + \mathbb{Z}\tau).

Its compactification X(1)=Y(1)βˆͺ{cusp}β‰…P1X(1) = Y(1) \cup \{\text{cusp}\} \cong \mathbb{P}^1 via the jj-invariant.

With level structure:

  • Y0(N)Y_0(N) parametrizes pairs (E,C)(E, C) where CβŠ‚E[N]C \subset E[N] is a cyclic subgroup of order NN.
  • Y1(N)Y_1(N) parametrizes pairs (E,P)(E, P) where P∈E[N]P \in E[N] is a point of exact order NN.
  • Y(N)Y(N) parametrizes (E,Ξ±)(E, \alpha) where Ξ±:(Z/N)2β†’βˆΌE[N]\alpha: (\mathbb{Z}/N)^2 \xrightarrow{\sim} E[N] is a full level-NN structure.
ExampleModular curves and modularity
  • X0(11)X_0(11): The first modular curve of genus 11. It is the elliptic curve y2+y=x3βˆ’x2βˆ’10xβˆ’20y^2 + y = x^3 - x^2 - 10x - 20 of conductor 1111.

  • X0(N)X_0(N) for small NN: genus 00 for N≀10N \leq 10 and N=12,13,16,18,25N = 12, 13, 16, 18, 25. First genus-11 case: N=11N = 11. First genus-22 case: N=23N = 23.

  • Modularity theorem (Wiles, Taylor–Wiles, Breuil–Conrad–Diamond–Taylor): Every elliptic curve E/QE/\mathbb{Q} of conductor NN corresponds to a weight-2 newform f∈S2(Ξ“0(N))f \in S_2(\Gamma_0(N)), i.e., there exists a surjection X0(N)β†’EX_0(N) \to E.

  • Fermat's Last Theorem follows: Frey showed a solution ap+bp=cpa^p + b^p = c^p gives a semistable curve E:y2=x(xβˆ’ap)(x+bp)E: y^2 = x(x - a^p)(x + b^p) that cannot be modular by Ribet's theorem. Wiles proved all semistable curves are modular. Contradiction.


Summary: the central role of elliptic curves

RemarkConnections and applications

Elliptic curves sit at a remarkable crossroads:

  • Number theory: Mordell–Weil, BSD, Fermat's Last Theorem, modular forms.
  • Algebraic geometry: simplest abelian varieties, starting point for curves of higher genus.
  • Complex geometry: C/Ξ›\mathbb{C}/\Lambda, moduli of lattices, modular forms.
  • Cryptography: ECDSA, pairing-based cryptography, isogeny-based post-quantum crypto (SIKE/SIDH).
  • Physics: mirror symmetry, string compactifications on elliptic fibrations.