research40 min studyLesson 599 in Mathematics

Complex uniformization and lattices

Build a complex elliptic curve from a rank-two lattice and interpret complex-linear maps through their lattice indices.

On this page

What you’ll understand

  • State the Weierstrass ℘ normalization and its relation to a short cubic.
  • Distinguish a basis change of a lattice from scaling its complex homothety class.
  • Calculate the degree and kernel of a complex-linear isogeny using lattice inclusion.

Over the complex numbers, the elliptic group can be described as the plane modulo two independent periods. This turns addition into addition of complex numbers modulo a lattice, but the torus still has a precise algebraic equation. The Weierstrass function and its derivative recover that equation and its differential. Uniformization thereby explains torsion and isogeny degree geometrically, while preserving the distinction between a real-linear picture of a parallelogram and the complex-linear maps that are actually holomorphic.

A rank-two lattice gives a compact complex group

A lattice is Z omega-one plus Z omega-two with the ratio of the periods nonreal. The quotient C/Λ is a compact one-dimensional complex Lie group with identity the zero class. Choosing a different integral basis with determinant one and the same orientation does not change Λ; it changes the displayed fundamental parallelogram. Choosing a nonzero complex scalar u produces a homothetic lattice uΛ and a holomorphically isomorphic torus via z mapped to uz.

The Weierstrass function is an even meromorphic lattice-periodic function with a double pole at each lattice point. Its derivative is odd with triple poles. The defining sum is regularized by subtracting the constant inverse-square contribution for each nonzero lattice point, and converges normally away from the lattice. Summing unmodified inverse squares in an arbitrary order would not be the same justified construction. The associated absolutely convergent Eisenstein sums define the cubic's coefficients.

℘Λ(z)=1z2+∑0≠λ∈Λ(1(z−λ)2−1λ2),g2=60∑λ≠0λ−4,g3=140∑λ≠0λ−6\wp_\Lambda(z)=\frac1{z^2}+\sum_{0\ne\lambda\in\Lambda}\left(\frac1{(z-\lambda)^2}-\frac1{\lambda^2}\right),\qquad g_2=60\sum_{\lambda\ne0}\lambda^{-4},\quad g_3=140\sum_{\lambda\ne0}\lambda^{-6}
The sums define meromorphic periodic data on C/Λ. The ratio of two lattice generators must be nonreal; a rank-one period group would not give a compact elliptic torus.↔ Scroll wide equations horizontally.

The derivative normalization fixes the cubic

The identity for the square of the derivative is wp-prime squared equal to four wp cubed minus g₂ wp minus g₃. Therefore the short Weierstrass coordinates are x=wp and y=wp-prime divided by two. Their equation has coefficients A=−g₂/4 and B=−g₃/4, and the zero class maps to O through the poles. The discriminant is nonzero for a genuine lattice. Using wp-prime itself as y without changing the cubic would introduce a factor-of-four error.

The invariant differential dx/(2y) pulls back to dz, and the map respects addition. Conversely, every elliptic curve over C admits such a uniformization as a complex analytic group. This theorem links analytic periods to algebraic geometry; it does not assert that arbitrary real periods recover a curve defined over Q. A rational model can have transcendental periods, and the lattice description alone does not decide which of its complex points are rational over a smaller field.

℘′(z)2=4℘(z)3−g2℘(z)−g3,z⟼(℘(z),℘′(z)2),dx2y=dz\wp'(z)^2=4\wp(z)^3-g_2\wp(z)-g_3,\qquad z\longmapsto\left(\wp(z),\frac{\wp'(z)}2\right),\qquad \frac{dx}{2y}=dz
These conventions yield y²=x³−(g₂/4)x−g₃/4. The lattice origin is sent to the projective point at infinity.↔ Scroll wide equations horizontally.

Complex linearity and finite index describe isogenies

A holomorphic group homomorphism between two such tori lifts to z mapped to alpha z with alphaΛ contained in Λ-prime. If alpha is nonzero, it is an isogeny and its degree is the finite index of alphaΛ inside Λ-prime. The kernel is alpha-inverse Λ-prime modulo Λ. In characteristic zero these geometric kernel points already account for the degree, matching the algebraic separability statement. Multiplication by n has index n² because it scales both lattice directions.

A real-linear map on the underlying two-dimensional plane need not commute with multiplication by i, and therefore need not define a holomorphic elliptic-curve morphism. Likewise, a matrix changing a chosen lattice basis is not automatically an additional complex endomorphism. Extra endomorphisms require a complex number alpha preserving the lattice beyond the integer scalars. The square lattice provides an example through multiplication by i; a generic lattice has only integer endomorphisms.

An extra endomorphism of the square lattice

Let Λ=Z+iZ. Calculate the degree and kernel of multiplication by alpha=1+i on C/Λ.

  1. The images of the basis 1,i are 1+i and −1+i. In that basis the integer matrix is [[1,−1],[1,1]], whose determinant is two.

  2. Thus alphaΛ is a sublattice of Λ of index two, and the induced complex isogeny has degree two.

  3. Besides the zero class, the class z=(1+i)/2 is in the kernel because (1+i)z=i lies in Λ. Its double also lies in Λ, and the index computation proves there are exactly these two kernel classes.

The result

The square-lattice endomorphism has degree two and a cyclic two-element kernel. It is complex-linear, unlike an arbitrary real matrix of determinant two.

Explore the model

Interactive lab

Interactive lab · matrix

Distinguish a lattice basis shear from a sublattice

Pause and reset, then set both controls to a = 1 and b = 1. The selected integer matrix [[1,1],[0,1]] has determinant one and changes the basis of Z² without changing that lattice. For a fresh comparison, pause and reset, then set a = 2 and b = 0; predict determinant two and an index-two image sublattice. These real maps illustrate index bookkeeping. The second map scales only one direction and is not complex-linear on the square torus, so determinant alone does not make it an elliptic isogeny.

Distinguish a lattice basis shear from a sublatticeThe selected matrix is A = [[a,b],[0,1]]. Playback interpolates geometry between the identity and A; readouts always describe A. Its determinant is a and its singular values come from AᵀA. Positive a makes every supported matrix invertible. This two-dimensional family is not a general eigensolver or a learned network. Current readings: Area scale · det A: 1.4; Singular values · σmax, σmin: 1.597, 0.876; Condition number · κ₂(A): 1.823. All parameters are adjustable with the labeled controls below.Matrix A with first row 1.4, 0.6 and second row zero, one; display interpolates between identity and A while final determinant remains 1.4.−4−2024−2−1012x · equal geometric scaleyA = [ 1.4 0.6 ; 0 1 ]Display blend: 0% · final values below
Interpolated geometryOriginal unit squareSelected final transform
Area scale · det A
1.4
Singular values · σmax, σmin
1.597, 0.876
Condition number · κ₂(A)
1.823
t = 0 s

The selected matrix is A = [[a,b],[0,1]]. Playback interpolates geometry between the identity and A; readouts always describe A. Its determinant is a and its singular values come from AᵀA. Positive a makes every supported matrix invertible. This two-dimensional family is not a general eigensolver or a learned network. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forA lattice index gives an isogeny degree only after the required complex-linear map has been established.

Check your understanding

Which condition makes z↦alpha z descend to a nonzero isogeny C/Λ→C/Λ′?

Try it yourself

For an arbitrary rank-two lattice Λ, compute the degree and geometric kernel of multiplication by two. Give representatives using a basis omega-one, omega-two. Separately state how g₂ and g₃ change when Λ is replaced by uΛ.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Both integer lattice directions contribute a factor two to the index. In the Eisenstein sums, replace each lattice vector by u times that vector.

Compare with the solution

The degree is four. The kernel classes are zero, omega-one/2, omega-two/2 and (omega-one+omega-two)/2, giving two cyclic factors of order two. Scaling the lattice gives g₂(uΛ)=u to the minus four times g₂(Λ), and g₃(uΛ)=u to the minus six times g₃(Λ). The corresponding Weierstrass coordinates scale by u to the minus two and minus three. These homotheties preserve the j-invariant and analytic isomorphism class, whereas a mere change of basis leaves the lattice itself unchanged.

Take it with you

  • The compact quotient C/Λ has a cubic model through ℘ and ℘′/2.
  • Uniformization respects addition and pulls the invariant differential back to dz.
  • A complex-linear lattice inclusion gives degree by index; an arbitrary real matrix does not.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.