Fermat's Last Theorem — Wiles' Strategy

Wiles proved Fermat's Last Theorem (aⁿ + bⁿ = cⁿ has no positive integer solutions for n > 2) by proving the Modularity Theorem for semistable elliptic curves over ℚ
Core strategy: established isomorphism between Hecke algebras and universal deformation rings (the "R=T theorem") linking Galois representations to modular forms
Used Galois deformation theory to study mod p representations, connecting elliptic curve arithmetic to modular form theory—two seemingly separate mathematical areas
Built on Ribet's work connecting Fermat's equation to Frey curves and their modularity to complete the proof chain
The R=T theorem became foundational for the Langlands Program and proving instances of the Fontaine-Mazur Conjecture
Revolutionary 1995 proof ended a 358-year-old conjecture by synthesizing algebraic geometry, representation theory, and arithmetic geometry

Fermat's Last Theorem — Wiles' Strategy

Overview

Wiles' proof of Fermat's Last Theorem represents one of the crowning achievements of 20th-century mathematics, synthesizing deep ideas from algebraic geometry, representation theory, and arithmetic geometry. The theorem, conjectured by Pierre de Fermat in 1637, states that no three positive integers $a$ , $b$ , $c$ satisfy $a^{n} + b^{n} = c^{n}$ for any integer $n > 2$ . What makes Wiles' approach revolutionary is not merely that it settled a 358-year-old problem, but that it did so by proving a substantial portion of the Modularity Theorem (then the Taniyama-Shimura-Weil conjecture), fundamentally connecting two seemingly disparate areas: the arithmetic of Elliptic Curves and the theory of Modular Forms.

The strategy pioneered by Wiles, and completed with crucial assistance from Richard Taylor, relies on establishing a deep relationship between Galois representations arising from elliptic curves and those arising from modular forms. The key innovation was to prove that every semistable elliptic curve over $Q$ is modular, meaning its associated $L$ -function coincides with that of a modular form. This modularity result, when combined with earlier work by Ribet connecting Fermat's equation to the modularity of certain Frey curves, provided the final link needed to prove Fermat's Last Theorem.

The technical heart of Wiles' strategy involves the study of Galois Deformation Theory, particularly deformations of mod $p$ Galois representations, and the construction of an isomorphism between two rings: a Hecke Algebra acting on spaces of modular forms and a universal deformation ring parametrizing Galois representations. This "R=T theorem" remains central to modern developments in the Langlands Program, serving as a prototype for establishing instances of the Fontaine-Mazur Conjecture and pursuing generalizations to higher-dimensional varieties.

The fundamental insight is that modularity of elliptic curves can be proven by showing that certain universal deformation rings are isomorphic to Hecke algebras, with this isomorphism established by analyzing the structure of Selmer Groups and demonstrating both rings are complete intersections.

Fermat's Last Theorem — Wiles' Strategy: The strategy consists of the following components:

The Frey Curve: For a hypothetical solution $(a, b, c)$ to $a^{p} + b^{p} = c^{p}$ with prime $p \geq 5$ , construct the elliptic curve $E : y^{2} = x (x - a^{p}) (x + b^{p})$ .

Level-Lowering (Ribet's Theorem): Show that if the Frey curve exists, it is semistable and has conductor $N = rad (ab c)$ , but the associated mod $p$ Galois representation $\overset{ρ}{ˉ}_{E, p} : Gal (\overline{Q} / Q) \to GL_{2} (F_{p})$ cannot arise from a modular form of level dividing $N$ .

Modularity: Prove that every semistable elliptic curve over $Q$ is modular, i.e., there exists a weight-2 newform $f$ of level $N$ such that $L (E, s) = L (f, s)$ .

The R=T Theorem: For a suitable prime $p$ and residual representation $\overset{ρ}{ˉ}$ , construct an isomorphism between the universal deformation ring $R$ and the Hecke algebra $T$ , both acting on spaces of modular forms.

Numerical Criterion: Use commutative algebra techniques to show $R$ and $T$ are both complete intersections with matching dimensions, forcing them to be isomorphic.

Core Theory

Main Result

Theorem (Wiles-Taylor, 1995) [1]: Every semistable elliptic curve over

Q

is modular. That is, if

E / Q

is an elliptic curve with conductor

N

that is semistable at all primes (meaning

E

has multiplicative reduction at primes dividing

N

), then there exists a weight-2 newform

f \in S_{2} (Γ_{0} (N))

such that for all primes

ℓ ∤ N

a_{ℓ} (E) = a_{ℓ} (f)

where

a_{ℓ} (E) = ℓ + 1 - # E (F_{ℓ})

and

a_{ℓ} (f)

is the

ℓ

-th Fourier coefficient of

f

. Theorem (Fermat's Last Theorem) [2]: For all integers

n \geq 3

, the equation

x^{n} + y^{n} = z^{n}

has no solutions in positive integers

x, y, z

We prove Theorem [2] assuming Theorem [1].

Step 1 (Reduction to prime exponents): It suffices to prove the statement for

n = 4

(done by Fermat via infinite descent) and for prime exponents

p \geq 5

. If

x^{n} + y^{n} = z^{n}

with

n = m p

, then

(x^{m})^{p} + (y^{m})^{p} = (z^{m})^{p}

. Step 2 (Construction of Frey curve): Suppose

a^{p} + b^{p} = c^{p}

is a solution with

a, b, c

pairwise coprime and

p \geq 5

prime. Define the elliptic curve

E : y^{2} = x (x - a^{p}) (x + b^{p})

This is a semistable elliptic curve over

Q

with discriminant

Δ = \frac{1}{16} (ab c)^{2 p}

and minimal conductor

N = rad (ab c)

(the product of distinct primes dividing

ab c

). Step 3 (Properties of the Frey curve): The curve

E

has the following properties:

It is semistable (multiplicative reduction at primes dividing $ab c$ , good reduction elsewhere)
For any prime $ℓ ∤ 2 ab c$ , we have $a_{ℓ} (E) \equiv ℓ + 1 (mod p)$
The mod $p$ representation $\overset{ρ}{ˉ}_{E, p} : G_{Q} \to GL_{2} (F_{p})$ has specific local properties

Step 4 (Ribet's Level-Lowering Theorem): Ribet (1990) proved that if such a Frey curve were modular, the mod

p

representation

\overset{ρ}{ˉ}_{E, p}

would arise from a newform of level

2

. However, the space

S_{2} (Γ_{0} (2))

is trivial (dimension 0), so no such modular form exists. Step 5 (Contradiction via Modularity): By Theorem [1], the Frey curve

E

must be modular since it is semistable. This contradicts Step 4. Therefore, no solution

(a, b, c)

a^{p} + b^{p} = c^{p}

can exist for

p \geq 5

The proof of Theorem [1] itself requires establishing the R=T theorem. We outline the key steps:

Step 6 (Setup for R=T): Fix a prime

p \geq 3

and a continuous representation

\overset{ρ}{ˉ} : G_{Q} \to GL_{2} (F_{p})

arising from a semistable elliptic curve. Let

R

be the universal deformation ring parametrizing deformations of

\overset{ρ}{ˉ}

satisfying certain local conditions at primes dividing

N p

(minimal flat at

p

, minimally ramified at primes dividing

N

). Step 7 (Hecke algebra): Let

T

be the Hecke algebra acting on the space of modular forms

S_{2} (Γ_{0} (N p), O)

localized at the maximal ideal corresponding to

\overset{ρ}{ˉ}

(where

O

is the ring of integers in a finite extension of

Q_{p}

). Step 8 (Natural map): The Eichler-Shimura theory produces a Galois representation

ρ_{T} : G_{Q} \to GL_{2} (T)

, inducing a natural map

ϕ : R \to T

by the universal property of

R

. Step 9 (Surjectivity): Using Ihara's Lemma and analysis of congruences between modular forms, one shows

ϕ

is surjective. Step 10 (Equality of dimensions): The crucial step is showing that the Krull dimensions of

R

and

T

(as complete local

O

-algebras) are equal. This uses:

Lower bounds on $dim T$ from eigenform theory
Upper bounds on $dim R$ from Galois Cohomology calculations
Specifically, $dim R = 1 + [K : Q_{p}]$ where $K$ is the coefficient field

Step 11 (Complete intersections): Using the numerical criterion of Wiles' Numerical Criterion, both

R

and

T

are shown to be complete intersections over

O

. For

R

, this involves deep analysis of Selmer Groups associated to the adjoint representation

ad^{0} (\overset{ρ}{ˉ})

. Step 12 (Isomorphism): A surjective map between complete intersections of the same dimension is an isomorphism. Therefore

R ≅ T

, which implies that every deformation of

\overset{ρ}{ˉ}

satisfying the local conditions arises from a modular form. Step 13 (Conclusion): The Galois representation from the original semistable elliptic curve provides such a deformation, hence must arise from a modular form. □

A critical error is to assume that all elliptic curves over $Q$ were proven modular by Wiles' original work. Wiles proved modularity only for semistable elliptic curves. The full Modularity Theorem for all elliptic curves over $Q$ was completed later by Breuil, Conrad, Diamond, and Taylor (2001). The semistability condition was essential for controlling the local deformation conditions at primes of bad reduction, and removing this hypothesis required fundamentally new techniques in the theory of p-adic Modular Forms.

Andrew Wiles (1995) — Proved modularity of semistable elliptic curves over

Q

using Galois deformation theory and the numerical criterion for Hecke algebras to be isomorphic to universal deformation rings.

Examples and Applications

1

Modularity of $y^{2} = x^{3} - x$ : Consider the elliptic curve

E : y^{2} = x^{3} - x

over

Q

, which has conductor

N = 32

. Step 1: Compute the discriminant:

Δ = - 64

, and verify

E

has multiplicative reduction at

p = 2

only (split multiplicative, Kodaira type I₄*). Step 2: Since

E

has complex multiplication by

Z [i]

, by classical CM theory it is modular. Explicitly, it corresponds to the modular form

f (z) = q n = 1 \prod \infty (1 - q^{n})^{2} (1 - q^{2 n})^{2} = q - 2 q^{2} - q^{3} + 2 q^{4} + q^{5} + \dots

where

q = e^{2 π i z}

, which is a weight-2 newform of level 32. Step 3: Verify coefficient matching. For

ℓ = 3

$a_{3} (E) = 3 + 1 - # E (F_{3}) = 4 - 5 = - 1$ (counting: $E (F_{3})$ has points including point at infinity)
$a_{3} (f) = - 1$ from the $q$ -expansion

The coefficients match for all primes

ℓ ∤ 2

. Step 4: For Wiles' strategy applied to this curve at

p = 3

, the mod 3 representation

\overset{ρ}{ˉ}_{E, 3}

is reducible (due to CM), which requires modifications to the standard deformation theory. Nevertheless, the curve's modularity can be established through CM theory independently.

2

Application to $a^{5} + b^{5} = c^{5}$ : Suppose there exists a non-trivial solution to

a^{5} + b^{5} = c^{5}

with

g cd (a, b, c) = 1

and

2∣ a

. Step 1: The Frey curve is

E : y^{2} = x (x - a^{5}) (x + b^{5})

. Step 2: The conductor is

N = rad (ab c)

since

E

is semistable. At each prime

ℓ ∣ ab c

E

has multiplicative reduction. Step 3: The discriminant is

Δ_{E} = 2^{- 4} (ab c)^{10}

, so the minimal discriminant involves

ab c

to the 10th power. Step 4: Consider the mod 5 representation

\overset{ρ}{ˉ}_{E, 5}

. By careful analysis of the

j

-invariant and discriminant:

j (E) = \frac{2 ^{8} (( a ^{5} ) ^{2} + ( b ^{5} ) ^{2} + ( a ^{5} ) ( b ^{5} ) ) ^{3}}{( ab c ) ^{10}}

Step 5: Ribet's theorem (1990) shows that if

E

were modular, then

\overset{ρ}{ˉ}_{E, 5}

would arise from a modular form of level

2

(since we can arrange

ab c

to be coprime to 2 by scaling). Step 6: But

S_{2} (Γ_{0} (2)) = 0

, yielding a contradiction. Combined with Wiles' proof that

E

is modular (being semistable), we conclude no such solution exists. Step 7: This establishes Fermat's Last Theorem for

n = 5

via Wiles' strategy. Why semistability is essential: Consider the elliptic curve

E : y^{2} + y = x^{3} - x^{2} - 10 x - 20

which has conductor

N = 11 \cdot 121 = 1331 = 1 1^{3}

. This curve has additive reduction at

p = 11

(Kodaira type IV), hence is not semistable.

Wiles' original techniques do not apply directly because:

Local deformation theory breaks down: At primes of additive reduction, there is no canonical "minimal" or "flat" deformation condition. The local deformation ring at 11 has more complicated structure.

Level-lowering fails: Ribet's level-lowering theorem, crucial for the Frey curve argument, requires semistability. For additive reduction, we cannot control how mod $p$ representations factor through level.

The R=T machinery needs modification: The calculation of $dim R$ via Galois cohomology assumes controlled ramification. Additive reduction introduces wild ramification that Wiles' original approach couldn't handle.

This curve is modular (proven by the later work of Breuil-Conrad-Diamond-Taylor using p-adic Families of Modular Forms), but different techniques were required. Removing the semistability hypothesis was a major achievement requiring new ideas about modularity lifting beyond Wiles' original framework.

The geometric picture: Think of Wiles' strategy as building a bridge between two landscapes: Galois Landscape: Here we have symmetries of algebraic numbers. An elliptic curve

E

gives a 2-dimensional "movie" of how its

p

-torsion points

E [p]

transform under

G_{Q}

. This is the representation

ρ_{E, p}

. The challenge: can we characterize which 2-dimensional Galois representations come from geometry? Modular Landscape: Here we have highly symmetric analytic functions on the upper half-plane. Modular forms

f (z)

have beautiful transformation properties under

SL_{2} (Z)

. They encode arithmetic data in their Fourier coefficients

a_{n} (f)

, which (via Eichler-Shimura) also give Galois representations. The Bridge (R=T): Wiles constructs two "continents" of deformations:

Continent R: All possible ways to deform $\overset{ρ}{ˉ}$ while preserving local structure (minimal ramification conditions)
Continent T: The space carved out by Hecke operators acting on modular forms

The shocking fact is these continents have the same dimension (computed via cohomology vs. eigenform theory). The map from R to T is surjective (every Hecke eigenform produces a deformation). A surjection between spaces of equal dimension is an isomorphism!

Physical analogy: It's like discovering that two seemingly different crystals (one from Galois theory, one from analysis) have identical lattice structures. Once you know they have the same dimension and one can be mapped onto the other, they must be the same crystal.

Historical Development

The path to Wiles' proof spans decades of development in arithmetic geometry:

1955-1965: Yutaka Taniyama and Goro Shimura formulated the initial conjecture linking elliptic curves to modular forms, refined by André Weil. The Taniyama-Shimura-Weil conjecture proposed that every elliptic curve over

Q

should be modular. 1985: Gerhard Frey suggested that a counter-example to Fermat's Last Theorem would yield an elliptic curve unlikely to be modular. The Frey curve construction showed explicitly how to build a semistable elliptic curve from a hypothetical solution to

a^{p} + b^{p} = c^{p}

. 1986: Jean-Pierre Serre formulated the epsilon-conjecture, making Frey's intuition precise: the Frey curve, if it existed, could not be modular due to its peculiar Galois representation properties. 1990: Ken Ribet proved Serre's epsilon-conjecture (now Ribet's theorem), establishing that modularity of semistable elliptic curves implies Fermat's Last Theorem. This reduced a 350-year-old problem to proving a specific case of modularity. 1993-1994: Andrew Wiles announced his proof of modularity for semistable elliptic curves, introducing Galois Deformation Theory techniques. A gap was discovered in the original proof related to making certain Hecke rings complete intersections. 1994-1995: Wiles, collaborating with Richard Taylor, developed the "Taylor-Wiles patching argument" to overcome the gap. Their key innovation involved constructing auxiliary primes and using numerical criterion arguments. The complete proof appeared in Wiles (1995, Annals of Mathematics) "Modular elliptic curves and Fermat's Last Theorem" and Taylor-Wiles (1995) "Ring-theoretic properties of certain Hecke algebras." 2001: Breuil, Conrad, Diamond, and Taylor completed the proof of full modularity for all elliptic curves over

Q

using techniques beyond Wiles' original approach, particularly exploiting theory of Crystalline Representations.

Connections to Other Areas

Connection to The Birch and Swinnerton-Dyer Conjecture: Wiles' modularity theorem has profound implications for BSD Conjecture. When an elliptic curve

E / Q

is modular, its

L

-function

L (E, s)

equals

L (f, s)

for a modular form

f

. The analytic properties of modular form

L

-functions are much better understood than general elliptic curve

L

-functions. Specifically:

Analytic continuation: Modularity automatically implies $L (E, s)$ extends to an entire function (via the modularity of $f$ ).

Functional equation: The $L$ -function satisfies a functional equation $Λ (E, s) = w Λ (E, 2 - s)$ where $w = \pm 1$ is the sign in the functional equation.

BSD at $s = 1$ : The order of vanishing of $L (E, s)$ at $s = 1$ relates to the Mordell-Weil rank. Work of Gross-Zagier and Kolyvagin uses modularity crucially to prove BSD (rank part) when $ord_{s = 1} L (E, s) \leq 1$ .

Heegner points: Modularity allows construction of Heegner Points on $E$ via the modular parametrization $X_{0} (N) \to E$ , providing explicit rational points and enabling Kolyvagin's Euler system techniques.

This connection makes modularity one of the most powerful tools for studying rational points on elliptic curves, directly impacting computational number theory and explicit BSD calculations.

Fermat's Last Theorem — Wiles' Strategy