Every rational elliptic curve is modular.

That single sentence — the modularity theorem — is one of the most profound results in mathematics. It says that two apparently unrelated mathematical objects — elliptic curves (geometry) and modular forms (analysis/symmetry) — are secretly the same thing, connected through their L-functions.

What it means

An elliptic curve $E$ over $\mathbb{Q}$ produces a sequence of integers $a_2, a_3, a_5, a_7, \ldots$ by counting points modulo each prime: $a_p = p + 1 - |E(\mathbb{F}_p)|$.

A modular form $f$ of weight 2 produces a sequence of Fourier coefficients $a_1, a_2, a_3, \ldots$ from its $q$-expansion: $f(z) = \sum a_n q^n$.

The modularity theorem says: for every rational elliptic curve $E$, there exists a modular form $f$ whose Fourier coefficients match the point counts of $E$. The two sequences are identical (at all but finitely many primes).

Equivalently: $L(E, s) = L(f, s)$. The L-function of the curve equals the L-function of the form.

History

1955: Yutaka Taniyama first suggested a connection between elliptic curves and modular forms at a Tokyo-Nikko conference. Goro Shimura refined the conjecture over the following years. It became known as the Taniyama-Shimura conjecture (sometimes Taniyama-Shimura-Weil, after André Weil contributed further precision).

For decades, most mathematicians considered it either hopelessly hard or possibly false. It asserted a connection between two fields that seemed to have nothing to do with each other.

1985-86: Everything changed when Gerhard Frey observed that a counterexample to Fermat’s Last Theorem would produce an elliptic curve with very strange properties — so strange that it couldn’t be modular. Kenneth Ribet proved this (Ribet’s theorem, 1986), establishing:

If the Taniyama-Shimura conjecture is true for semistable elliptic curves, then Fermat’s Last Theorem is true.

This transformed the Taniyama-Shimura conjecture from an abstract curiosity into the key to a 350-year-old problem.

1993-95: Andrew Wiles, after seven years of secret work, proved the conjecture for semistable elliptic curves (those with squarefree conductor). This was enough to prove Fermat’s Last Theorem. The initial proof had a gap, discovered during peer review, which Wiles and Richard Taylor repaired in 1994. The corrected proof was published in 1995.

2001: Breuil, Conrad, Diamond, and Taylor extended Wiles’ methods to prove the full modularity theorem for all rational elliptic curves.

The Frey curve

If $a^p + b^p = c^p$ were a solution to Fermat’s equation (for prime $p > 2$), consider the Frey curve:

$$y^2 = x(x - a^p)(x + b^p)$$

This is an elliptic curve whose discriminant is proportional to $(abc)^{2p}$ — a perfect $2p$-th power, which is highly unusual. Frey argued that such a curve would be too pathological to be modular.

Ribet proved this rigorously: Frey curves cannot be modular. Therefore, if all semistable elliptic curves are modular (Taniyama-Shimura), Frey curves can’t exist, and Fermat’s equation has no solutions.

The logic chain:

1. Assume $a^p + b^p = c^p$ has a solution
2. Construct the Frey curve from this solution
3. Ribet’s theorem: this curve is not modular
4. Wiles’ theorem: all semistable elliptic curves ARE modular
5. Contradiction → no solution exists → Fermat’s Last Theorem ∎

Why the connection exists

The modularity theorem is not just a coincidence — it reflects a deep structural principle. Both elliptic curves and modular forms are controlled by the same underlying algebraic object: a Galois representation.

An elliptic curve $E$ determines a representation of the absolute Galois group of $\mathbb{Q}$ (via the action on the torsion points of $E$). A modular form $f$ also determines a Galois representation (via Deligne’s theorem). The modularity theorem says these representations match — the symmetries of the number field encoded by the curve are the same symmetries encoded by the form.

This is a glimpse of the Langlands programme: the idea that all “motivic” Galois representations (from geometry) correspond to “automorphic” representations (from analysis). The modularity theorem is the first major instance of this correspondence being proved.