The Main Theorem (Milla, ch. 9) #
The paper's ch. 9: the differential-equation form thm42 and the
Main Theorem hauptformel,
1/(2π·Im τ) · √(J(τ)/(J(τ)−1))
= ∑ n, ((1−s₂(τ))/6 + n) · (6n)!/((3n)!(n!)³) · (1728·J(τ))⁻ⁿ (Im τ > 1.25)
where √ is the principal branch (here: Complex.cpow (1/2)).
The intermediate propositions thm35 (quasiperiod/period derivative relation) and
thmglg10 (η₁ − (3g₃/2g₂)s₂ = π/Im τ) are stated in lattice language in the paper;
here the whole ch. 9 computation is carried out in modular language, replacing the
quasiperiod calculus by Ramanujan's derivative identities (Ramanujan.lean) applied
to Kummer's identity E₄ = ₂F₁(1/12,5/12;1;1/J)⁴ (Kummer.lean): see the section
"The proof of thm42" below.
The branch of the square root is resolved pointwise in thm42:
J/(J−1) = (G⁶/E₆)² with Re(G⁶/E₆) > 0 everywhere on Region (by explicit
estimates), and the principal square root of w² is w on the right half-plane
(Complex.sq_cpow_two_inv). The connectedness workhorse sq_eq_on_preconnected_eq
originally planned for this step (PLAN A8) is kept for reference/reuse.
The summand of the Main Theorem's series:
((1−s₂(τ))/6 + n) · (6n)!/((3n)!(n!)³) · (1728·J(τ))⁻ⁿ.
Equations
Instances For
The Main Theorem's series converges absolutely on the region Im τ > 1.25
(ratio test: (6n)!/((3n)!(n!)³) ≤ 1728ⁿ and ‖1728·J‖ > 1728 on the region).
The branch-resolution workhorse (PLAN A8): two continuous functions on a preconnected set whose squares agree, the second nonvanishing, agreeing at one point, agree everywhere.
The region Im τ > 1.25 is preconnected.
The reduction hauptformel ← thm42 #
Milla's derivation of the Main Theorem from thm42 is the term-by-term differentiation
of the series darst (hyp2F1_sq_eq_tsum): with cₙ = (6n)!/((3n)!(n!)³·1728ⁿ) and
w = 1/J (so ‖w‖ < 1 on the region),
(1−s₂)/6·G(w) + w·G′(w) = ∑ ((1−s₂)/6 + n)·cₙ·wⁿ = ∑ mainSummand.
The proof of thm42 #
Following the modular-forms reformulation of the paper's ch. 9 computation: with
G = ₂F₁(1/12,5/12;1;1/J) (so E₄ = G⁴ by Kummer's Thm. omegastrich), Ramanujan's
identity D E₄ = (E₂E₄−E₆)/3 and the derivative D J = −J·E₆/E₄ (from D E₄, D E₆)
combine via the chain rule to E₂ = E₆/E₄ + 12G³G′E₆/(E₄²J). Substituting this and
E₄ = G⁴ into the right-hand side of thm42 collapses it to G⁶/(2πE₆·Im τ); and the
left-hand side equals the same value because J/(J−1) = E₄³/E₆² = (G⁶/E₆)² and the
principal square root of w² is w whenever Re w > 0 — which holds here by the
explicit estimates ‖G²−1‖ ≤ 0.15 (from the Gsq power series, ‖1/J‖ < 1/1.096)
and ‖E₆−1‖ ≤ 0.199 (from Estimates.lean). This replaces the paper's
continuity/connectedness branch argument by a pointwise right-half-plane one.
Milla's Prop. thm42: the differential equation
1/(2π·Im τ)·√(J/(J−1)) = (1−s₂)/6 · G(1/J) + (1/J)·G′(1/J)
for Im τ > 1.25, with the principal branch of the square root. The branch is
resolved pointwise: J/(J−1) = (G⁶/E₆)² with Re(G⁶/E₆) > 0 on all of Region
(by the explicit estimates), and the principal square root of w² is w on the
right half-plane.
The Main Theorem (Milla, Thm. hauptformel; Chudnovsky–Chudnovsky 1988,
Eq. (1.4)): for all τ with Im τ > 1.25,
1/(2π·Im τ)·√(J/(J−1)) = ∑ n, ((1−s₂)/6 + n)·(6n)!/((3n)!(n!)³)·(1728·J)⁻ⁿ,
with the principal branch of the square root.