Documentation

TauCeti.NumberTheory.ModularForms.AtkinLehner.DoubleCoset

Atkin–Lehner matrices and the Γ₀(N) double cosets #

An Atkin–Lehner matrix W for a divisor Q of N, read in GL(2, ℚ), normalizes the image of Γ₀(N). This file shows that conjugation by W moreover fixes every double coset Γ₀(N) α Γ₀(N) with α ∈ Δ₀(N) of determinant coprime to Q:

W⁻¹ α W ∈ Γ₀(N) α Γ₀(N).

Writing W = !![Q a, b; N c, Q d] and α = !![p, q; N r, s], the conjugate is the integral matrix B with W B = α W. Its lower-left entry is divisible by N, and its upper-left entry is congruent to s modulo Q and to p modulo N / Q, hence a unit modulo N (s is a unit modulo Q because det α ≡ p s is); so W⁻¹ α W ∈ Δ₀(N). It has the determinant of α, and the same common divisors of entries (a common divisor of either matrix is coprime to Q, and Q • B = adj W · α · W), so the two lie in the same Γ₀(N)-double coset (HeckeRing.GL2.mem_doubleCoset_of_det_eq_of_dvd_iff). No coprimality with N / Q is needed, so this covers the U_p double cosets at the primes p ∣ N / Q.

These are the two hypotheses under which the slash by W commutes with the Hecke operator of Γ₀(N) α Γ₀(N) (HeckeRing.GL2.heckeSlashSum_slash_of_mem_normalizer).

Main results #

References #

An Atkin–Lehner matrix normalizes Γ₀(N) in GL(2, ℚ).

Conjugation by an Atkin–Lehner matrix fixes a Γ₀(N) double coset of determinant coprime to Q. If W is an Atkin–Lehner matrix for Q ∣ N and α ∈ Δ₀(N) has an integral matrix A with determinant coprime to Q, then W⁻¹ α W lies in the double coset Γ₀(N) α Γ₀(N).