Zeta5: rational functionals and Hankel determinant polynomials
DefinitionZeta5_SourceConstructionFor every natural number , define . The rational linear functional on monomials is
Let . Its polynomial-valued pole functional is
Polynomial division by , together with the residues , defines . For , set , , and define
where . Finally put
These are the explicit rational polynomials used throughout the project. The source module is retained verbatim, including its namespace and its total definitions at ; the exponent and endpoint written use natural-number subtraction.
import Mathlib
/-!
# The Hankel determinant construction of the paper "ζ(5) is irrational"
This file *defines* (in Lean) the objects of Section 2.1 of the paper
"ζ(5) is irrational" (A. Fauzan, 17 September 2026):
* the polynomials `D_m(t) = ∏_{j=1}^m (t + j²)`;
* the `ℚ`-linear functional `μ_X`, given on monomials by (2.2) and on the simple poles
`1/(t + j²)` by (2.3), extended to `P / D_K` by polynomial division and partial fractions;
* the Hankel matrix `G_K(X)` of (2.4), its determinant `Δ_K(X)`, the scalar `S_K` of (2.5)
and `F_K = S_K Δ_K`.
Throughout, `K = 40 n`, `N = 3 n`, `h = 37 n` as in (2.1).
**Convention.** Two polynomial rings `ℚ[X]` occur: the paper's variable `t` (the argument of the
rational functions the functional is applied to) and the paper's indeterminate `X` (the
value of the functional is affine in `X`). Both are represented by `Polynomial ℚ`; the
docstrings say which is which.
Nothing is proved about these objects here; see `Apery.MainEstimate` for the main estimate.
-/
open Polynomial Finset
namespace Apery
/-- `D_m(t) = ∏_{j=1}^m (t + j²)`, a monic polynomial in `t`. -/
noncomputable def D (m : ℕ) : ℚ[X] := ∏ j ∈ Icc 1 m, (X + C ((j : ℚ) ^ 2))
/-- The polynomial moments (2.2): `μ(t^e) = (-1)^e B_{2e+2} (2e+3)(2e+4)(2e+5) / 24`.
Mathlib's `bernoulli` uses the convention `B₁ = -1/2`, as does the paper. -/
def μmono (e : ℕ) : ℚ :=
(-1) ^ e * _root_.bernoulli (2 * e + 2) * ((2 * e + 3) * (2 * e + 4) * (2 * e + 5)) / 24
/-- `μ` on polynomials in `t`, extended `ℚ`-linearly from the monomials. -/
noncomputable def μpoly (P : ℚ[X]) : ℚ := P.sum fun e c => c * μmono e
/-- The generalized harmonic number `H_j^{(5)} = ∑_{v=1}^j v⁻⁵`. -/
def H5 (j : ℕ) : ℚ := ∑ v ∈ Icc 1 j, 1 / (v : ℚ) ^ 5
/-- The pole values (2.3), as polynomials in the indeterminate `X`:
`μ_X(1/(t + j²)) = j⁴ (X - H_j^{(5)}) - 1/4 + 1/(2j)`. -/
noncomputable def poleValue (j : ℕ) : ℚ[X] :=
C ((j : ℚ) ^ 4) * (X - C (H5 j)) - C (1 / 4) + C (1 / (2 * (j : ℚ)))
/-- `μ_X(P(t) / D_K(t))` for a polynomial `P` in `t`, as a polynomial in `X`.
Since `D_K` is monic with the simple roots `t = -j²` (`1 ≤ j ≤ K`),
`P / D_K = (P /ₘ D_K) + ∑_j (P(-j²) / D_K'(-j²)) / (t + j²)`, and `μ_X` is applied termwise. -/
noncomputable def μX (K : ℕ) (P : ℚ[X]) : ℚ[X] :=
C (μpoly (P /ₘ D K)) +
∑ j ∈ Icc 1 K,
C (P.eval (-(j : ℚ) ^ 2) / (derivative (D K)).eval (-(j : ℚ) ^ 2)) * poleValue j
/-- The `h × h` Hankel matrix `G_K(X)` of (2.4), whose `(i, j)` entry is
`μ_X (D_N(t)^6 t^(i+j) / D_K(t))`, with `K = 40 n`, `N = 3 n`, `h = 37 n`. -/
noncomputable def G (n : ℕ) : Matrix (Fin (37 * n)) (Fin (37 * n)) ℚ[X] :=
fun i j => μX (40 * n) (D (3 * n) ^ 6 * X ^ ((i : ℕ) + (j : ℕ)))
/-- `Δ_K(X) = det G_K(X)`, a polynomial in `X` (of degree `h`, by (2.9)). -/
noncomputable def Δ (n : ℕ) : ℚ[X] := (G n).det
/-- The scalar `S_K` of (2.5). -/
noncomputable def S (n : ℕ) : ℚ :=
((40 * n).factorial : ℚ) ^ (2 * (37 * n)) * 4 ^ (37 * n - 1) /
(((3 * n).factorial : ℚ) ^ (12 * (37 * n)) *
∏ i ∈ Icc 1 (37 * n - 1), ((2 * i).factorial : ℚ) ^ 2)
/-- `F_K = S_K Δ_K` of (2.5). -/
noncomputable def F (n : ℕ) : ℚ[X] := C (S n) * Δ n
end Apery
Read-back
What the Lean code literally says, in plain math · GPT-6 (Codex independent auditor)
D. For each natural number , including , this defines the rational polynomial , where each natural number is regarded as a rational number inside the polynomial. The product at is empty, so .
μmono. For every natural number , this defines the rational number , where is the rational Bernoulli number with convention and the natural-number factors are interpreted in . The index is included, giving .
μpoly. For every rational polynomial , this defines the rational number , where is the finite set of exponents having nonzero coefficient, is the coefficient of , and are the Bernoulli numbers with . For the zero polynomial the sum is empty and the value is .
H5. For every natural number , this defines the rational number . The indices are positive natural numbers interpreted in ; when the sum is empty and .
poleValue. For each natural number , this defines the rational polynomial in the output variable . There is no hypothesis : at , the sum is empty and division by zero in gives , so the defined polynomial is the constant .
μX. For every natural number and every rational polynomial , this defines the rational polynomial by the following formula. Here for , for , and for , where are the rational Bernoulli numbers with . For and , let be the polynomial quotient on division of by the monic polynomial , and put
The first sum is over the finitely many nonzero coefficients of ; denotes a coefficient and the prime denotes formal differentiation. The variable is the input polynomial variable and is the output polynomial variable. Empty products are , empty sums are , and rational division by zero has value . In particular, , , and . For this gives the zero polynomial. This declaration assigns the displayed value to the pair .
G. For every natural number , this defines a square matrix over with row and column indices and entries . Here for , for , and for , where are the rational Bernoulli numbers with . For and , let be the polynomial quotient on division of by the monic polynomial , and put
The first sum is over the finitely many nonzero coefficients of ; denotes a coefficient and the prime denotes formal differentiation. The variable is the input polynomial variable and is the output polynomial variable. Empty products are , empty sums are , and rational division by zero has value . In particular, , , and . At both index sets are empty, so the matrix has no entries.
Δ. For every natural number , this defines the rational polynomial . Here for , for , and for , where are the rational Bernoulli numbers with . For and , let be the polynomial quotient on division of by the monic polynomial , and put
The first sum is over the finitely many nonzero coefficients of ; denotes a coefficient and the prime denotes formal differentiation. The variable is the input polynomial variable and is the output polynomial variable. Empty products are , empty sums are , and rational division by zero has value . In particular, , , and . When the matrix is empty and its determinant is , so .
S. For every natural number , this defines the rational scalar by the following formula. Put , the truncated natural-number subtraction used in the code, and . All factorials have natural-number arguments and are then embedded in ; the finite product is over natural numbers . In particular , the product at is empty, all zero exponents give , and .
F. For every natural number , this defines the rational polynomial as the product of the constant polynomial with value and the determinant polynomial . The rational polynomial in this assertion is . Put , the truncated natural-number subtraction used in the code, and . Here for , for , and for , where are the rational Bernoulli numbers with . For and , let be the polynomial quotient on division of by the monic polynomial , and put
The first sum is over the finitely many nonzero coefficients of ; denotes a coefficient and the prime denotes formal differentiation. The variable is the input polynomial variable and is the output polynomial variable. Empty products are , empty sums are , and rational division by zero has value . In particular, , , and . At the determinant is that of the empty matrix and equals , and .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.