Nontrivial bound
ProvedCircleMethod.S_boundanalytic-number-theorycircle-methodnumber-theorywaring-problem
For , and every ,
A genuine power saving over the trivial bound . The derivation is pleasingly direct: is a Weyl sum of length at frequency , and that frequency satisfies the approximation hypothesis exactly, so Weyl's inequality applies with no work. With the bracket becomes , and raising to the power gives the saving.
This is weaker than the classical , but it is enough for the purpose: combined with the trivial count of reduced residues it forces the local densities to decay faster than once is large, which is what makes the singular series converge absolutely.
Preamble
import Definitions.Def_CircleMethod_waring import Definitions.Def_CircleMethod_weylsum import Mathlib.Analysis.SpecialFunctions.Pow.Real import Mathlib.Analysis.SpecialFunctions.Integrals.Basic import Mathlib.Analysis.PSeries import Mathlib.Analysis.Normed.Group.InfiniteSum import Mathlib.NumberTheory.Divisors import Mathlib.Data.Fintype.Pi open Finset
Formal statement
namespace CircleMethod
theorem S_bound {k : ℕ} (hk : 2 ≤ k) {ε : ℝ} (hε : 0 < ε) :
∃ C : ℝ, 0 < C ∧ ∀ (q : ℕ) (a : ℤ), 0 < q → Nat.Coprime a.natAbs q →
‖S k q a‖ ≤ C * (q : ℝ) ^ (1 - (2 : ℝ) ^ (1 - (k : ℝ)) + ε) := by sorry
end CircleMethodSource
R. C. Vaughan, The Hardy-Littlewood Method, 2nd ed., Cambridge Tracts in Mathematics 125, Cambridge University Press, 1997, Section 2.6 and Lemma 2.4 (Weyl's inequality applied at N = q); the classical sharper bound is Theorem 4.2.