Quadratic coefficient of the truncated eta quotient.
ProvedMoonshineHeadTable.two_mul_coeff_two_etaPartialaether-catalognumbertheory
Quadratic coefficient of the truncated eta quotient. For M ≥ 2 it equals
a 1 (a 1 + 1) / 2 + a 1 + a 2, again independent of the truncation.
theorem MoonshineHeadTable.two_mul_coeff_two_etaPartial(a : ℕ → ℤ) {M : ℕ} (hM : 2 ≤ M) :
2 * coeff 2 ((etaPartial R a M : (R⟦X⟧)ˣ) : R⟦X⟧)
= ((a 1 * (a 1 + 1) + 2 * (a 1 + a 2) : ℤ) : R) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source NumberTheory/MoonshineHeadTable.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from NumberTheory/MoonshineHeadTable.lean
import Mathlib
import Definitions.Def_NumberTheory_MoonshineHeadTable
/-!
# Head coefficients of eta-quotient McKay–Thompson series, from frame shapes
This file is cycle 3 of the research thread begun in `Shared.PoleOrderObstruction`
and continued in `Shared.PoleOrderObstructionDeep`. Those files reduced the
"Monster-sized product" of the `194` McKay–Thompson series to a *finite* piece of
arithmetic: the Laurent coefficients just above the pole are elementary symmetric
expressions in the tabulated numbers `c_g(1)`.
The obvious next question is: **where does the table `c_g(1)` come from?** For a
large family of Monster classes the McKay–Thompson series is (up to an additive
constant) a Dedekind eta quotient
`T_g(τ) = 1 / η_g(τ) + const`, `η_g(τ) = ∏_k η(k τ) ^ (a k)`,
attached to the *frame shape* `∏ k ^ (a k)` of `g` acting on the Leech lattice
(a balanced frame shape: `∑ k * a k = 24`). For such `g` the head coefficient
`c_g(1)` is therefore **not tabulated data at all**: it is computable from the
finitely many integers `a k`.
The main theorem of this file, `MoonshineHeadTable.coeff_two_etaPartial`, computes
the relevant coefficient of the `q`-expansion of `1/η_g` purely formally, and gives
the closed formula
`c_g(1) = a 1 * (a 1 + 3) / 2 + a 2`.
Contents.
* `MoonshineHeadTable.coeff_one_prod_one`, `MoonshineHeadTable.two_mul_coeff_two_prod_one`
— Newton-type identities for the linear and quadratic coefficients of a finite
product of power series with constant term `1`, over an arbitrary commutative
ring (the `ℂ`-versions live in `Shared.PoleOrderObstructionDeep`).
* `MoonshineHeadTable.jet_zpow` — the `2`-jet of an *integer* power `u ^ z` of a
unit power series with constant term `1`. This is what makes eta *quotients*
(negative exponents) accessible.
* `MoonshineHeadTable.etaPartial` — the truncated eta-quotient unit
`∏_{m ≤ M} (1 - qᵐ) ^ (-∑_{k ∣ m} a k)`, i.e. `q · (1/η_g)` truncated after the
`M`-th factor.
* `MoonshineHeadTable.coeff_one_etaPartial`, `MoonshineHeadTable.coeff_two_etaPartial`
— its linear and quadratic coefficients, *stable in `M`*
(`MoonshineHeadTable.coeff_two_etaPartial_stable`), so that they are honest
coefficients of the infinite product.
* `MoonshineHeadTable.headCoeff`, `MoonshineHeadTable.headCoeff_pmFrame` — the
closed formula, and its evaluation on the family of balanced frame shapes
`1^(-e) n^(e)` with `e * (n - 1) = 24`.
* `MoonshineHeadTable.etaHeadTable` — the resulting *derived* (not recalled) table
of eight head coefficients `276, 54, 20, 9, 2, 0, -1, -1`, checked by `decide`,
together with its sum `359`.
Everything is proved from scratch over a general commutative ring and specialized
to `ℤ`, where the resulting statements are decidable.
-/
open MoonshineHeadTable
open PowerSeries Finset
/-! ## 1. Newton identities for low coefficients of products, over any ring -/
variable {R : Type*} [CommRing R] {ι : Type*}
/-! ## 2. The `2`-jet of an integer power of a unit power series -/
/-! ## 3. Truncated eta quotients attached to a frame shape -/
variable (R : Type*) [CommRing R]
variable {R}Formal statement
theorem MoonshineHeadTable.two_mul_coeff_two_etaPartial(a : ℕ → ℤ) {M : ℕ} (hM : 2 ≤ M) :
2 * coeff 2 ((etaPartial R a M : (R⟦X⟧)ˣ) : R⟦X⟧)
= ((a 1 * (a 1 + 1) + 2 * (a 1 + a 2) : ℤ) : R) := by sorrySource