Per-modulus monomial piece bounded by the squared divisor count
Provedgpy_monomialPairSumAtL_abs_leLet be a tuple of shifts, let be a level, let be monomial indices, let be a coordinate, and let be a modulus. Consider the tuples of positive integers with each entry at most and total product at most . A pair is good at when some witness satisfies the joint divisibility system for all and is coprime to the combined modulus . The monomial piece at exact modulus is the sum over good pairs with of . Then
where counts ordered factorizations of into positive factors. This is the error-side control for the monomial evaluation: every surviving term has Möbius factor at most in absolute value and log-ratio at most , and the number of surviving good pairs at modulus is at most by the Selberg fiber bound. Formalization Note: the support set, the good-pair condition, and the divisor count are all unfolded into explicit finite sums, filters, and cardinalities.
import Mathlib open Classical
theorem gpy_monomialPairSumAtL_abs_le {k : ℕ} (h : Fin k → ℕ) (R : ℕ)
(a b : ℕ) (i : Fin k) (L : ℕ) (hR : 2 ≤ R) :
abs (∑ d ∈ (Fintype.piFinset (fun _ => Finset.Icc 1 R)).filter (fun d => ∏ i, d i ≤ R),
∑ e ∈ (Fintype.piFinset (fun _ => Finset.Icc 1 R)).filter (fun d => ∏ i, d i ≤ R),
if (Finset.lcm Finset.univ (fun j => Nat.lcm (d j) (e j)) = L
∧ (∃ w, (∀ j, Nat.lcm (d j) (e j) ∣ w + h j) ∧
Nat.Coprime (w + h i) (Finset.lcm Finset.univ (fun j => Nat.lcm (d j) (e j)))))
then (ArithmeticFunction.moebius (∏ j, d j) : ℝ)
* (ArithmeticFunction.moebius (∏ j, e j) : ℝ)
* (-(Real.log ((∏ j, d j : ℕ) : ℝ) / Real.log R)) ^ a
* (-(Real.log ((∏ j, e j : ℕ) : ℝ) / Real.log R)) ^ b
else 0)
≤ (((Fintype.piFinset (fun _ : Fin (k + 1) => Finset.Icc 1 L)).filter
(fun d => ∏ i, d i = L)).card : ℝ) ^ 2 := by sorry