Number Theory
Modular Forms
Цели урока
- Understand the definition of a modular form of weight k and level N
- Know Hecke operators and why eigenforms have multiplicative Fourier coefficients
- Understand the modularity theorem and its role in proving Fermat's Last Theorem
- Compute dimensions of spaces of modular forms via Riemann-Roch
Предварительные знания
- Elliptic Curves
- Complex analysis on the upper half-plane
- Basic algebraic geometry
How can functions on the complex upper half-plane, satisfying a symmetry condition under 2x2 integer matrices, encode the number of points on elliptic curves over every finite field simultaneously?
- Cryptographic verification: modularity theorem enables verified pairing-based cryptography on BN-256 and BLS-12-381 curves
- Zero-knowledge proofs: zk-SNARKs use elliptic curves whose L-functions are explicitly known via modular forms
- Number theory software: SAGE, Magma, PARI/GP compute modular forms to verify BSD numerically
- String theory: the j-invariant j(tau) = 1/q + 744 + 196884q + ... appears in monstrous moonshine connecting modular forms to the Monster group
A conjecture for 40 years: Taniyama to Wiles
Yutaka Taniyama in 1955 made a vague and imprecise conjecture at a mathematics symposium in Tokyo: that all elliptic curves over Q might be connected to modular forms. Goro Shimura helped clarify the conjecture. Andre Weil gave it a precise form in 1967. For 30 years it seemed impossibly hard. In 1986 Gerhard Frey connected it to Fermat's Last Theorem. In 1990 Ken Ribet proved the connection rigorously. In 1994 Andrew Wiles announced a proof, then discovered a gap. In September 1995, Wiles and Taylor published the complete proof. A 350-year-old problem was solved by proving a 40-year-old conjecture about modular forms.
Definition of Modular Forms
In 1994, Andrew Wiles proved that every semistable elliptic curve over Q is modular, establishing Fermat's Last Theorem. Weight-2 modular forms of level N correspond to differential forms on modular curves, and the space S_2(Gamma_0(N)) has dimension computable by the Riemann-Roch formula.
The space M_k(SL_2(Z)) is spanned by Eisenstein series E_k for k != 12. At weight 12, there is also Delta. Dimension formula: dim M_k = floor(k/12) + epsilon where epsilon depends on k mod 12. For S_k (cusp forms): dim S_k = dim M_k - 1.
What does holomorphicity of a modular form at the cusps require?
Holomorphicity at infinity: the q-expansion f = sum_{n>=0} a_n q^n has no negative powers. For a general modular form, a_0 can be nonzero. Cusp forms have a_0 = 0 at every cusp.
Hecke Eigenforms and Their L-functions
The key property of Hecke eigenforms is that their L-functions factor as Euler products. This is not a coincidence: it follows from the commutativity of the Hecke algebra and the multiplicativity of Fourier coefficients. The Euler product connects modular forms to counting points on elliptic curves over finite fields.
The Hecke algebra acting on S_k(Gamma_0(N)) is commutative and semisimple (for weight k >= 2). This means there is a basis of simultaneous Hecke eigenforms. Each eigenform defines a number field Q(a_1, a_2, ...) over Q generated by its Fourier coefficients.
The unique weight-12 form and Ramanujan
Delta and its multiplicative structure
The space S_12(SL_2(Z)) has dimension 1, spanned by Delta(tau) = sum tau(n) q^n. Since it is one-dimensional, Delta is automatically a Hecke eigenform. The Hecke eigenvalues are tau(p) for prime p. Ramanujan conjectured |tau(p)| is at most 2p^{11/2} - the Ramanujan bound. Deligne proved this in 1974 via the Weil conjectures.
Why do Hecke eigenforms have Euler product L-functions?
Multiplicativity: T_m T_n = T_{mn} when gcd(m,n) = 1. An eigenform of all T_n has multiplicative eigenvalues: a_{mn} = a_m a_n. This is exactly the condition for the Dirichlet series sum a_n n^{-s} to factor as an Euler product.
Modularity Theorem and Fermat's Last Theorem
The Taniyama-Shimura-Wiles modularity theorem is the complete case of the Langlands correspondence for GL_2 over Q. It says: every elliptic curve over Q is a quotient of a modular curve. This single theorem implies Fermat's Last Theorem and has transformed number theory.
| Year | Result | Author(s) |
|---|---|---|
| 1955 | Conjecture: all elliptic curves over Q are modular | Taniyama, Shimura |
| 1986 | Frey curve: if Fermat fails, there is a weird elliptic curve | Frey |
| 1990 | Frey curve cannot be modular (epsilon conjecture) | Ribet |
| 1995 | Semistable elliptic curves are modular | Wiles, Taylor-Wiles |
| 2001 | All elliptic curves over Q are modular | Breuil-Conrad-Diamond-Taylor |
Wiles's proof does not directly prove Fermat's Last Theorem from scratch. It proves modularity for semistable curves, which includes the Frey curve. Ribet's theorem then closes the gap. The full proof is a chain of three deep results over 15 years.
What is the Taniyama-Shimura-Wiles modularity theorem?
Modularity theorem: for every E/Q with conductor N_E, there exists f in S_2(Gamma_0(N_E)) with a_p(E) = a_p(f) for all but finitely many primes p. This gives L(E,s) = L(f,s), connecting the arithmetic of E to the analytic theory of modular forms.
Dimension Formulas and the Riemann-Roch Theorem
The dimension of the space M_k(Gamma) of modular forms of weight k for congruence subgroup Gamma is computed by the Riemann-Roch theorem on the modular curve. This gives explicit formulas in terms of k, the genus, and data at the cusps and elliptic points.
X_0(11): the simplest modular elliptic curve
Level 11, genus 1
The modular curve X_0(11) has genus 1, so dim S_2(Gamma_0(11)) = 1. The unique (up to scalar) cusp form is f = q - 2q^2 - q^3 + 2q^4 + q^5 + 2q^6 - 2q^7 + ... The associated elliptic curve is E: y^2 + y = x^3 - x^2 - 10x - 10, with conductor 11. For every prime p not equal to 11, a_p(E) = a_p(f) = p+1 - #E(F_p).
What is dim S_2(Gamma_0(N)) and what does it count?
dim S_2(Gamma_0(N)) = g(X_0(N)) = genus of the modular curve. At level N=11, g=1, giving one weight-2 form corresponding to the elliptic curve X_0(11) itself.
Connections to other topics
Modular forms connect complex analysis, number theory, and representation theory.
- Algebraic geometry — Related topic
- Representation theory — Related topic
- Physics — Related topic
Итоги
- Modular form of weight k: f((a*tau+b)/(c*tau+d)) = (c*tau+d)^k f(tau) for matrices in Gamma
- q-expansion: f(tau) = sum a_n q^n with q = e^{2*pi*i*tau}; holomorphic at cusps means no negative powers
- Hecke operators T_p: eigenfunctions have multiplicative a_{mn} = a_m*a_n for gcd(m,n) = 1
- dim S_2(Gamma_0(N)) = g(X_0(N)) - connects analytic theory to algebraic geometry
- Modularity theorem (Wiles 1995): every elliptic curve over Q has L(E,s) = L(f,s) for some f in S_2(Gamma_0(N_E))
Вопросы для размышления
- Why does the space M_k(SL_2(Z)) have dimension 0 for weight k = 2 but positive dimension for all even k >= 4?
- How does the Fourier expansion coefficient a_p of a Hecke eigenform encode the number of points on an elliptic curve over F_p?
- In the proof of Fermat's Last Theorem, what exactly does Ribet's theorem say, and why is it logically separate from Wiles's modularity theorem?