Elliptic Functions — Weierstrass p-function

The Weierstrass ℘-function is the canonical doubly-periodic meromorphic function on the complex plane, with a double pole at each point of a lattice $Λ = Z ω_{1} + Z ω_{2}$ . It is the fundamental building block of elliptic function theory: every elliptic function with period lattice $Λ$ is a rational function of $℘$ and its derivative $℘^{'}$ . Used in algebraic geometry (elliptic curves), number theory, integrable systems, and mathematical physics. Introduced by Karl Weierstrass in lectures of the 1860s–1880s.

Given a lattice $Λ \subset C$ , an elliptic function is a meromorphic function $f$ with $f (z + ω) = f (z)$ for all $ω \in Λ$ . Liouville's theorems force such functions to have poles; the simplest non-constant one is the Weierstrass $℘$ -function, built by summing $1/ (z - ω)^{2}$ over the lattice with convergence-forcing counterterms. Its importance is structural: the field of elliptic functions for $Λ$ is exactly $C (℘, ℘^{'})$ , and $℘$ satisfies a first-order ODE whose cubic right-hand side exhibits the lattice as an elliptic curve $y^{2} = 4 x^{3} - g_{2} x - g_{3}$ . The surprising fact: this purely analytic construction produces an algebraic group law, turning the torus $C /Λ$ into an abelian variety and connecting analysis, geometry, and arithmetic in a single object.

The Core Insight

A non-constant doubly-periodic meromorphic function cannot be holomorphic everywhere (it would be bounded, hence constant by Liouville). The minimal-pole option, a single double pole per fundamental cell, is realized by $℘$ . Because $℘$ is even and $℘^{'}$ is odd, the pair $(℘, ℘^{'})$ parametrizes a plane cubic, and the addition of points on $C /Λ$ becomes the chord-and-tangent group law on that cubic. Periodicity + minimal poles = elliptic curve.

Formal Statement

Lattice and Elliptic Functions

A lattice is $Λ = Z ω_{1} + Z ω_{2}$ where $ω_{1}, ω_{2} \in C$ are $R$ -linearly independent (equivalently $τ = ω_{2} / ω_{1} \in / R$ ; WLOG $Im τ > 0$ ). A meromorphic function $f : C \to C \cup {\infty}$ is elliptic (doubly periodic) with respect to $Λ$ if

f (z + ω) = f (z) \forall ω \in Λ, \forall z .

⟳it suffices to require invariance under the two generators

ω_{1}, ω_{2}

, which then extends to all integer combinations.

A fundamental parallelogram (period cell) is $P = {a ω_{1} + b ω_{2} : a, b \in [0, 1)}$ ; it tiles $C$ under translation by $Λ$ .

The Weierstrass ℘-function

For a lattice $Λ$ , define

℘ (z; Λ) = \frac{1}{z ^{2}} + ω \in Λ ω \neq = 0 \sum (\frac{1}{( z - ω ) ^{2}} - \frac{1}{ω ^{2}}) .

⟳the bare sum

\sum (z - ω)^{- 2}

diverges like

\sum ∣ ω ∣^{- 2}

; subtracting

ω^{- 2}

makes each term

O (∣ ω ∣^{- 3})

for fixed

z

, restoring convergence.

Here $z \in C ∖ Λ$ . The function $℘$ is meromorphic on $C$ with a double pole (no residue) at every lattice point and is holomorphic elsewhere.

Fundamental Properties of ℘

Let $℘ = ℘ (\cdot; Λ)$ . Then:

$℘$ is even: $℘ (- z) = ℘ (z)$ .
$℘$ is $Λ$ -periodic and meromorphic, with order 2 (two poles per cell, counted with multiplicity).
Its derivative is

℘^{'} (z) = - 2 ω \in Λ \sum \frac{1}{( z - ω ) ^{3}},

⟳differentiate [2] term by term; the now-cubic terms converge absolutely so no counterterms are needed, and the

ω = 0

term recombines naturally.

which is odd and

Λ

-periodic of order 3.

$℘$ satisfies the algebraic differential equation

(℘^{'} (z))^{2} = 4 ℘ (z)^{3} - g_{2} ℘ (z) - g_{3},

⟳form

℘^{'2} - 4 ℘^{3} + g_{2} ℘ + g_{3}

; choose

g_{2}, g_{3}

so the principal parts at

z = 0

cancel; the result is elliptic, holomorphic, hence constant

= 0

by Liouville.

where the Eisenstein invariants are

g_{2} = 60 ω \neq = 0 \sum \frac{1}{ω ^{4}} = 60 G_{4}, g_{3} = 140 ω \neq = 0 \sum \frac{1}{ω ^{6}} = 140 G_{6} .

⟳expand

℘

in a Laurent series about

z = 0

; the coefficients are the Eisenstein series

G_{2 k} = \sum_{ω \neq = 0} ω^{- 2 k}

, and matching cubic and linear poles fixes

g_{2}, g_{3}

The Laurent expansion underlying these statements is

℘ (z) = \frac{1}{z ^{2}} + k = 1 \sum \infty (2 k + 1) G_{2 k + 2} z^{2 k} = \frac{1}{z ^{2}} + 3 G_{4} z^{2} + 5 G_{6} z^{4} + \dots .

⟳expand each

(z - ω)^{- 2} - ω^{- 2}

as a geometric series in

z / ω

; odd powers vanish by the symmetry

ω \to - ω

of the lattice.

Why It Holds / Derivation

Convergence, periodicity, and the ODE

Step 1 — Convergence of [2]. For

∣ z ∣ \leq R

and

∣ ω ∣ > 2 R

\frac{1}{( z - ω ) ^{2}} - \frac{1}{ω ^{2}} = \frac{z ( 2 ω - z )}{ω ^{2} ( z - ω ) ^{2}} \leq \frac{C ∣ z ∣}{∣ ω ∣ ^{3}} .

⟳combine fractions over a common denominator and bound

∣2 ω - z ∣ \leq 3∣ ω ∣

∣ z - ω ∣ \geq ∣ ω ∣/2

The lattice sum $\sum_{ω \neq = 0} ∣ ω ∣^{- s}$ converges iff $s > 2$ (comparison with $\int r^{- s} r d r d θ$ ). Hence the series converges absolutely and uniformly on compact sets avoiding $Λ$ , defining a meromorphic function with double poles exactly on $Λ$ .

Step 2 — Evenness. Replacing

z \to - z

and reindexing

ω \to - ω

(a bijection of

Λ

) leaves [2] invariant, so

℘ (- z) = ℘ (z)

. Step 3 — Periodicity. From [3],

℘^{'}

is manifestly

Λ

-periodic since

ω \mapsto ω - ω_{j}

permutes

Λ

. Thus

℘ (z + ω_{j}) - ℘ (z) = c_{j}

is constant. Set

z = - ω_{j} /2

: by evenness

℘ (ω_{j} /2) = ℘ (- ω_{j} /2)

, giving

c_{j} = 0

. So

℘

Λ

-periodic. Step 4 — The differential equation. Using [6],

℘ (z)^{3} = z^{- 6} + 9 G_{4} z^{- 2} + 15 G_{6} + O (z^{2}), ℘^{'} (z)^{2} = 4 z^{- 6} - 24 G_{4} z^{- 2} - 80 G_{6} + O (z^{2}) .

⟳cube [6] and square its term-by-term derivative

℘^{'} (z) = - 2 z^{- 3} + 6 G_{4} z + 20 G_{6} z^{3} + \dots

, keeping terms up to constants.

Form $F (z) = ℘^{'2} - 4 ℘^{3} + 60 G_{4} ℘ + 140 G_{6}$ . By [8] all negative powers and the constant term cancel, so $F$ is holomorphic at $z = 0$ with $F (0) = 0$ . Being elliptic with no poles, $F$ is bounded, hence constant by Liouville, hence $F \equiv 0$ . This is [4] with $g_{2} = 60 G_{4}$ , $g_{3} = 140 G_{6}$ . $■$

Liouville's Theorems for Elliptic Functions

Let $f$ be elliptic with period cell $P$ :

If $f$ is holomorphic on all of $C$ , then $f$ is constant.
$\sum_{poles in P} Res f = 0$ .
The number of zeros equals the number of poles in $P$ (the order of $f \geq 2$ if non-constant).
$\sum (zeros) \equiv \sum (poles) (mod Λ)$ .

Sketch of Liouville's theorems

(1) A holomorphic elliptic function is continuous on the compact closure

\overset{ˉ}{P}

, hence bounded, hence bounded on

C

by periodicity; Liouville's theorem (entire + bounded ⇒ constant) finishes it. (2) Integrate

f

over the boundary

\partial P

. Opposite sides cancel by periodicity, so

\oint_{\partial P} f d z = 0 = 2 π i \sum Res

. (Shift

P

slightly so no pole sits on

\partial P

.) (3) Apply (2) to

f^{'} / f

, whose residues count zeros minus poles. (4) Apply the same boundary argument to

z f^{'} (z) / f (z)

; the non-cancelling boundary terms produce a lattice vector.

■

Karl Weierstrass | 1815–1897

Weierstrass, the "father of modern analysis," rebuilt the theory of elliptic functions from his Berlin lectures (1860s onward) on a rigorous, systematic footing, replacing Jacobi's theta-quotient formalism with the single privileged function $℘$ and its sibling sigma and zeta functions. His insistence on $ε$ – $δ$ rigor, uniform convergence, and power-series foundations shaped the entire subsequent development of complex analysis. Though he published little of this material himself, it spread through students like Sofia Kovalevskaya, Hermann Schwarz, and Adolf Hurwitz, and through the lecture notes that became canonical.

Building Intuition

Think of $℘$ as the "electrostatic potential" of a doubly-periodic array of identical charges. Each lattice point hosts a singularity ( $1/ (z - ω)^{2}$ ); periodicity demands an infinite array. A naive infinite sum of equal contributions diverges, so we subtract the constant background $1/ ω^{2}$ — exactly like renormalizing a periodic potential against a uniform charge sheet. The result is the unique even, order-2 elliptic function regular at the origin up to its prescribed double pole.

The map $z \mapsto (℘ (z), ℘^{'} (z))$ wraps the torus $C /Λ$ onto the cubic curve [4]. The torus has genus 1; so does the curve. The two "missing" points ( $z = 0$ , the pole) both map to the single point at infinity of the projective cubic — the wrap is a bijection of $C /Λ$ onto the curve. Addition of complex numbers mod $Λ$ becomes the group law on the curve, and that is the deep payoff: a transcendental uniformization of an algebraic group.

The lattice Λ, fundamental cell, half-periods, and the ℘-map to the cubic

<svg viewBox="0 0 760 360" xmlns="http://www.w3.org/2000/svg">
<rect width="760" height="360" fill="#0a0e17"/>
<!-- left: lattice -->
<g stroke="#94a3b8" stroke-width="1" fill="none">
  <line x1="40" y1="40" x2="40" y2="320"/>
  <line x1="120" y1="40" x2="120" y2="320"/>
  <line x1="200" y1="40" x2="200" y2="320"/>
  <line x1="280" y1="40" x2="280" y2="320"/>
  <line x1="40" y1="60" x2="280" y2="60"/>
  <line x1="40" y1="140" x2="280" y2="140"/>
  <line x1="40" y1="220" x2="280" y2="220"/>
  <line x1="40" y1="300" x2="280" y2="300"/>
</g>
<!-- lattice points -->
<g fill="#94a3b8">
  <circle cx="40" cy="60" r="3"/><circle cx="120" cy="60" r="3"/><circle cx="200" cy="60" r="3"/><circle cx="280" cy="60" r="3"/>
  <circle cx="40" cy="140" r="3"/><circle cx="120" cy="140" r="3"/><circle cx="200" cy="140" r="3"/><circle cx="280" cy="140" r="3"/>
  <circle cx="40" cy="220" r="3"/><circle cx="120" cy="220" r="3"/><circle cx="200" cy="220" r="3"/><circle cx="280" cy="220" r="3"/>
  <circle cx="40" cy="300" r="3"/><circle cx="120" cy="300" r="3"/><circle cx="200" cy="300" r="3"/><circle cx="280" cy="300" r="3"/>
</g>
<!-- fundamental cell -->
<rect x="120" y="140" width="80" height="80" fill="#d4a853" fill-opacity="0.12" stroke="#d4a853" stroke-width="2"/>
<!-- period vectors -->
<line x1="120" y1="220" x2="200" y2="220" stroke="#d4a853" stroke-width="2" marker-end="url(#a)"/>
<line x1="120" y1="220" x2="120" y2="140" stroke="#d4a853" stroke-width="2" marker-end="url(#a)"/>
<!-- half period -->
<circle cx="160" cy="220" r="3" fill="#e2e8f0"/>
<circle cx="120" cy="180" r="3" fill="#e2e8f0"/>
<circle cx="160" cy="180" r="3" fill="#e2e8f0"/>
<text x="190" y="245" fill="#e2e8f0" font-size="12">ω₁</text>
<text x="92" y="185" fill="#e2e8f0" font-size="12">ω₂</text>
<text x="148" y="245" fill="#e2e8f0" font-size="11">ω₁/2</text>
<text x="60" y="335" fill="#e2e8f0" font-size="12">Lattice Λ ⊂ ℂ,  torus ℂ/Λ</text>
<!-- arrow map -->
<line x1="300" y1="180" x2="430" y2="180" stroke="#94a3b8" stroke-width="2" marker-end="url(#a)"/>
<text x="320" y="170" fill="#d4a853" font-size="13">z ↦ (℘, ℘′)</text>
<!-- right: cubic curve -->
<g>
  <line x1="470" y1="180" x2="730" y2="180" stroke="#94a3b8" stroke-width="1"/>
  <line x1="560" y1="60" x2="560" y2="320" stroke="#94a3b8" stroke-width="1"/>
  <path d="M 560 60 C 530 120, 525 180, 560 180 C 600 180, 650 120, 720 70" fill="none" stroke="#d4a853" stroke-width="2.5"/>
  <path d="M 560 300 C 530 240, 525 180, 560 180 C 600 180, 650 240, 720 290" fill="none" stroke="#d4a853" stroke-width="2.5"/>
  <text x="725" y="175" fill="#e2e8f0" font-size="12">x=℘</text>
  <text x="565" y="72" fill="#e2e8f0" font-size="12">y=℘′</text>
  <text x="490" y="335" fill="#e2e8f0" font-size="12">y² = 4x³ − g₂x − g₃</text>
</g>
<defs>
  <marker id="a" markerWidth="9" markerHeight="9" refX="7" refY="4" orient="auto">
    <path d="M0,0 L8,4 L0,8 z" fill="#d4a853"/>
  </marker>
</defs>
</svg>

Worked Examples

The lemniscatic (square) lattice: half-period values

Given: The lattice

Λ = Z + Z i

(square lattice,

τ = i

). By symmetry under

z \mapsto i z

(which permutes

Λ

), the Eisenstein series

G_{6}

over a square lattice vanishes: rotating by

i

multiplies

G_{6}

i^{- 6} = - 1

, forcing

G_{6} = 0

, hence

g_{3} = 0

. Goal: Find

℘

at the half-periods

e_{1} = ℘ (1/2)

e_{2} = ℘ ((1 + i) /2)

e_{3} = ℘ (i /2)

. Steps:

The cubic [4] factors as $℘^{'2} = 4 (℘ - e_{1}) (℘ - e_{2}) (℘ - e_{3})$ where $e_{i}$ are the values at the three half-periods (these are exactly the points where $℘^{'} = 0$ , since $℘^{'}$ is odd and periodic).
With $g_{3} = 0$ , the cubic is $4 x^{3} - g_{2} x = 4 x (x^{2} - g_{2} /4)$ . Its roots are

e_{2} = 0, e_{1} = + \frac{1}{2} g_{2}, e_{3} = - \frac{1}{2} g_{2} .

⟳solve

4 e^{3} - g_{2} e = 0

; the central root is

0

, the others are symmetric.

For $Λ = Z + Z i$ , numerically $g_{2} = 60 G_{4} \approx 189.07$ , so $g_{2} /2 \approx 6.875$ .

Answer:

e_{2} = 0, e_{1} \approx + 6.875, e_{3} \approx - 6.875

(dimensionless;

e_{1} + e_{2} + e_{3} = 0

always, since the

x^{2}

coefficient of

4 x^{3} - g_{2} x - g_{3}

vanishes).

Addition formula: computing ℘(2u)

Given: A point

u \in C /Λ

with known

℘ (u) = p

and

℘^{'} (u) = q

satisfying

q^{2} = 4 p^{3} - g_{2} p - g_{3}

. We want the duplication value

℘ (2 u)

. Steps:

The addition theorem states

℘ (u + v) = \frac{1}{4} (\frac{℘ ^{'} ( u ) - ℘ ^{'} ( v )}{℘ ( u ) - ℘ ( v )})^{2} - ℘ (u) - ℘ (v) .

⟳

℘ (u + v) + ℘ (u) + ℘ (v)

equals the square of the slope of the line through

(℘ (u), ℘^{'} (u))

and

(℘ (v), ℘^{'} (v))

, divided by 4 — the chord-and-tangent group law on the cubic.

Taking the limit $v \to u$ , the slope becomes the tangent slope $℘^{''} (u) / ℘^{'} (u)$ , giving the duplication formula

℘ (2 u) = \frac{1}{4} (\frac{℘ ^{''} ( u )}{℘ ^{'} ( u )})^{2} - 2 ℘ (u) .

⟳differentiate [4] to get

2 ℘^{'} ℘^{''} = 12 ℘^{2} - g_{2}

, so

℘^{''} = 6 ℘^{2} - g_{2} /2

; substitute.

Therefore, in terms of $p, q$ with $℘^{''} = 6 p^{2} - g_{2} /2$ :

℘ (2 u) = \frac{( 6 p ^{2} - g _{2} /2 ) ^{2}}{4 q ^{2}} - 2 p = \frac{( 12 p ^{2} - g _{2} ) ^{2}}{16 ( 4 p ^{3} - g _{2} p - g _{3} )} - 2 p .

⟳substitute

℘^{''} = 6 p^{2} - g_{2} /2

and

q^{2} = 4 p^{3} - g_{2} p - g_{3}

into [11] and simplify.

Numerical instance: Take

g_{2} = 4

g_{3} = 0

, and

p = ℘ (u) = 1

. Then

q^{2} = 4 (1) - 4 (1) - 0 = 0

— this is a half-period (where

℘^{'} = 0

), so we use [11] differently; instead take

p = 2

q^{2} = 4 \cdot 8 - 4 \cdot 2 = 24

. Then

12 p^{2} - g_{2} = 48 - 4 = 44

, and

℘ (2 u) = \frac{4 4 ^{2}}{16 \cdot 24} - 4 = \frac{1936}{384} - 4 = 5.0417 - 4.

Answer:

℘ (2 u) \approx 1.0417

(dimensionless), illustrating how doubling on the torus is a rational map on the cubic.

Common Misconceptions

$℘$ does not have a simple pole. Its poles are double poles with zero residue. A non-constant elliptic function of order 1 (a single simple pole per cell) is impossible: by Liouville's second theorem the residues in a cell sum to zero, so a single simple pole would need residue 0, contradicting simplicity. The minimal non-constant elliptic functions therefore have order 2 — either one double pole (like

℘

) or two distinct simple poles (like Jacobi's functions in their own normalization). Eisenstein series $G_{2}$ is special. The series

\sum_{ω \neq = 0} ω^{- 2}

is only conditionally convergent and is not modular of weight 2; it transforms with an anomalous additive term. This is why

℘

uses the absolutely convergent counterterm construction [2] rather than naively

\sum (z - ω)^{- 2}

, and why

g_{2}, g_{3}

start at weights 4 and 6. The would-be "

G_{2}

" reappears as the Weierstrass

η

quasi-period and in the non-holomorphic

E_{2}^{*}

One might guess that any even meromorphic doubly-periodic function with one double pole equals $℘$ up to scaling. False without normalizing the principal part. The function $℘ (z) + c$ for any constant $c$ is also even, $Λ$ -periodic, with a single double pole of leading coefficient $1$ — so it is not unique among such functions; uniqueness requires fixing the full principal part $z^{- 2}$ at the lattice point and the value (or the absence of a constant term in the Laurent expansion, as in [6]). Concretely, $℘ (z) + 7$ and $℘ (z)$ satisfy different ODEs ([4] with shifted $g_{3}$ is not even valid — shifting $℘$ does not preserve the cubic form), so "looks like $℘$ " is not enough.

Connections

The $℘$ -function sits at the crossroads of several theories:

Elliptic Curves: The map $z \mapsto [℘ (z) : ℘^{'} (z) : 1]$ is an analytic isomorphism $C /Λ \sim E (C)$ where $E : y^{2} = 4 x^{3} - g_{2} x - g_{3}$ . Group addition on $C /Λ$ matches the chord–tangent law.
Modular Forms: $g_{2}$ and $g_{3}$ are modular forms of weights 4 and 6; the discriminant $Δ = g_{2}^{3} - 27 g_{3}^{2}$ is a weight-12 cusp form, and the $j$ -invariant $j = 1728 g_{2}^{3} /Δ$ classifies lattices up to homothety.
Theta Functions: $℘$ can be written via Jacobi theta quotients; the Weierstrass $σ$ -function is essentially $θ_{1}$ , with $℘ = - (lo g σ)^{''}$ .
Jacobi Elliptic Functions: $sn, cn, dn$ are related to $℘$ by Möbius changes of variable; both describe the same tori in different normalizations.
Riemann Surfaces: $C /Λ$ is the universal example of a compact genus-1 Riemann surface, and $℘$ provides its embedding into $P^{2}$ .

Real-World Applications

Cryptography (ECC): Elliptic-curve cryptography (used in TLS, Bitcoin's secp256k1, signal protocols) relies on the group law that $℘$ uniformizes; the discrete-log problem on $E (F_{q})$ underpins security.
The pendulum and spinning top: The exact (large-amplitude) pendulum angle and the Euler–Poinsot rigid-body rotation are solved by $℘$ and Jacobi functions; energy conservation gives a cubic/quartic under the square root.
Soliton theory (KdV): Periodic traveling-wave ("cnoidal" and elliptic) solutions of the Korteweg–de Vries equation are $℘$ -functions of $x - c t$ ; the algebro-geometric (finite-gap) integration of KdV is built on Weierstrass and theta functions.
General relativity: Geodesics in Schwarzschild and Kerr spacetimes satisfy equations of the form $(\overset{r}{˙})^{2} = cubic/quartic (r)$ , integrated exactly by $℘$ ; light-bending and perihelion precession use this.
Statistical mechanics: The Ising model's correlation functions and Baxter's solution of the eight-vertex model employ elliptic parametrizations governed by $℘$ and theta functions.
Conformal mapping & potential theory: $℘$ maps fundamental domains to slit planes, used in computing electrostatic and fluid-flow fields with periodic boundary conditions.
Number theory algorithms: Computing torsion, isogenies, and the arithmetic of $j$ -invariants (e.g., in proving Fermat's Last Theorem via modularity) routinely uses the $℘$ -parametrization.

Historical Development

1750s–1786 — Euler & Lagrange: Addition theorems for elliptic integrals discovered; Euler's addition formula for $\int d x / quartic$ foreshadows the group law.
1820s — Abel & Jacobi: Independently invert elliptic integrals to obtain doubly-periodic functions; Jacobi's Fundamenta Nova (1829) introduces theta functions and $sn, cn, dn$ .
1844–1847 — Eisenstein & Liouville: Eisenstein studies the lattice series $\sum 1/ (z - ω)^{n}$ ; Liouville proves the foundational theorems on elliptic functions (boundedness ⇒ constant, residue sum zero).
1850s — Hermite, Briot & Bouquet: Systematize the field-theoretic structure of elliptic functions.
1862–1863 — Weierstrass: Lectures in Berlin introduce $℘$ , $σ$ , $ζ$ , the invariants $g_{2}, g_{3}$ , and the canonical ODE; rigor and the single privileged function replace theta quotients.
1882 — Schwarz: Publishes lecture notes (Formeln und Lehrsätze) disseminating Weierstrass's theory widely.
1893 — Halphen: Traité des fonctions elliptiques gives a comprehensive synthesis with applications to mechanics.
1920s–1930s — Hecke, Hurwitz, Klein: Modular interpretation of $g_{2}, g_{3}, Δ, j$ ; the bridge to modular forms is cemented.
1985–present — ECC era: Koblitz and Miller propose elliptic-curve cryptography; the $℘$ -uniformized group law becomes computationally central.

Subtleties & Limits

While $℘$ itself is fully understood, the arithmetic it uncovers fuels open problems:

Birch–Swinnerton-Dyer Conjecture: Relates the rank of the Mordell–Weil group $E (Q)$ — points on the Weierstrass cubic — to the order of vanishing of the $L$ -function $L (E, s)$ at $s = 1$ . A Clay Millennium Problem.
Lehmer-type questions on $Δ$ and $η$ : Non-vanishing and size of Fourier coefficients of the discriminant (Ramanujan's $τ$ -function, $Δ = q \prod (1 - q^{n})^{24}$ ) — whether $τ (n) \neq = 0$ for all $n$ remains open.
Periods and transcendence: $g_{2}, g_{3}$ , the periods $ω_{1}, ω_{2}$ , and quasi-periods $η_{1}, η_{2}$ obey the Legendre relation $η_{1} ω_{2} - η_{2} ω_{1} = 2 π i$ , but the transcendence theory (Chudnovsky's results, conjectures of Grothendieck on periods) is far from complete.
Effective uniformization in higher genus: Generalizing $℘$ to abelian varieties (Klein/Kleinian $℘$ -functions, hyperelliptic sigma functions) is an active area with incomplete algorithmic understanding.