Prime Hecke recurrences determine a coefficient sequence from its first term
ProvedMTT.coeff_eq_of_hecke_recurrenceFix and two functions (only their values at primes matter; in applications and is the -th Hecke eigenvalue). Say that a sequence of complex numbers satisfies the prime Hecke recurrences with data if for every prime and every
These are exactly the relations satisfied by the Fourier coefficients of a -eigenform with eigenvalue and nebentypus (Diamond–Shurman, Proposition 5.8.5), by the -expansion formula for .
Claim. If and both satisfy the prime Hecke recurrences with the same data , and , then
Proof. Strong induction on . For this is . For choose a prime and write with . The recurrence at gives
and likewise for ; both and are smaller than and positive, so the inductive hypothesis substitutes and , giving .
This is the uniqueness half of the statement that a normalised eigenform is determined by its Hecke eigenvalues: applied with for a normalised eigenform and for a cusp form with the same eigenvalues at every prime, it gives for . In Lean, is if p ∣ m then b (m / p) else 0 with natural-number division.
import Mathlib.Data.Nat.Prime.Basic import Mathlib.Data.Complex.Basic import Mathlib.Tactic.Linarith set_option autoImplicit false
theorem MTT.coeff_eq_of_hecke_recurrence (k : ℕ) (e a b c : ℕ → ℂ)
(hb : ∀ p : ℕ, p.Prime → ∀ m : ℕ,
b (p * m) + e p * (p : ℂ) ^ (k - 1) * (if p ∣ m then b (m / p) else 0) = a p * b m)
(hc : ∀ p : ℕ, p.Prime → ∀ m : ℕ,
c (p * m) + e p * (p : ℂ) ^ (k - 1) * (if p ∣ m then c (m / p) else 0) = a p * c m)
(hc1 : c 1 = 1) :
∀ m : ℕ, 0 < m → b m = b 1 * c m := by sorry