Skip to main content
AlexKontorovich/PrimeNumberTheoremAnd
Source indexedtheorem · leanprover/lean4:v4.32.0

Kadiri.kadiri_thm_3_1_q1_eq_11_truncated_limit_of_pointwise_inversion

PrimeNumberTheoremAnd.IEANTN.KadiriEq11Reduction · PrimeNumberTheoremAnd/IEANTN/KadiriEq11Reduction.lean:880 to 914

Source documentation

The faithful truncated-limit form of Kadiri equation (11) (\cite{Kadiri2005}, Théorème 3.1, eq.~(11), arXiv math/0401238): the von Mangoldt sum is the T → ∞ limit of the truncated contour integral kadiri_thm_3_1_q1_I φ a T over the compact segment [1+a-iT, 1+a+iT]. Under the same hypotheses as the stub kadiri_thm_3_1_q1_eq_11, the only remaining analytic input is the single-point truncated-limit Laplace inversion hinv (the displayed identity just before \cite[(11)]{Kadiri2005}); the sum/integral exchange over the compact window and the Tannery domination are discharged by this file, so no full-line integrability is needed. This replaces the (mis-stated) absolute Bochner form ∫ t : ℝ of the stub.

Exact Lean statement

theorem kadiri_thm_3_1_q1_eq_11_truncated_limit_of_pointwise_inversion
    {φ : ℝ → ℂ} (hφ : ContDiff ℝ 1 φ)
    {b : ℝ} (hb : 0 < b)
    (hφ_decay : (fun x : ℝ ↦ φ x * exp ((x : ℂ) / 2))
        =O[Filter.cocompact ℝ] fun x : ℝ ↦ Real.exp (-(1/2 + b) * |x|))
    (hφ'_decay : (fun x : ℝ ↦ deriv φ x * exp ((x : ℂ) / 2))
        =O[Filter.cocompact ℝ] fun x : ℝ ↦ Real.exp (-(1/2 + b) * |x|))
    {a : ℝ} (ha : 0 < a) (hab : a < b) (ha1 : a < 1)
    (hinv : ∀ n : Nat, 1 ≤ n →
      let Φ : ℂ → ℂ := fun s ↦ ∫ y, φ y * exp (-s * (y : ℂ)) ∂volume
      Tendsto
        (fun T : ℝ =>
          (1 / (2 * (Real.pi : ℂ))) *
            ∫ t in (-T)..T,
              Φ ((-(1 + a : ℝ) : ℂ) + (t : ℂ) * I) *
                (n : ℂ) ^ ((-(1 + a : ℝ) : ℂ) + (t : ℂ) * I))
        atTop (𝓝 (φ (Real.log n)))) :
    Tendsto (fun T : ℝ => kadiri_thm_3_1_q1_I φ a T) atTop
      (𝓝 (∑' n : ℕ, (Λ n : ℂ) * φ (Real.log n)))

Complete declaration

Lean source

Canonical source
Full Lean sourceLean 4
theorem kadiri_thm_3_1_q1_eq_11_truncated_limit_of_pointwise_inversion    {φ :   ℂ} (hφ : ContDiff  1 φ)    {b : } (hb : 0 < b)    (hφ_decay : (fun x :   φ x * exp ((x : ℂ) / 2))        =O[Filter.cocompact ] fun x :   Real.exp (-(1/2 + b) * |x|))    (hφ'_decay : (fun x :   deriv φ x * exp ((x : ℂ) / 2))        =O[Filter.cocompact ] fun x :   Real.exp (-(1/2 + b) * |x|))    {a : } (ha : 0 < a) (hab : a < b) (ha1 : a < 1)    (hinv :  n : Nat, 1  n       let Φ : ℂ := fun s  ∫ y, φ y * exp (-s * (y : ℂ)) ∂volume      Tendsto        (fun T :  =>          (1 / (2 * (Real.pi : ℂ))) *            ∫ t in (-T)..T,              Φ ((-(1 + a : ) : ℂ) + (t : ℂ) * I) *                (n : ℂ) ^ ((-(1 + a : ) : ℂ) + (t : ℂ) * I))        atTop (𝓝 (φ (Real.log n)))) :    Tendsto (fun T :  => kadiri_thm_3_1_q1_I φ a T) atTop      (𝓝 (∑' n : , (Λ n : ℂ) * φ (Real.log n))) := by  -- The bank reduction gives the limit for the interval-integral form `∫ t in (-T)..T`.  have hbank :=    kadiri_thm_3_1_q1_eq_11_pv_of_pointwise_inversion:= φ) hφ (b := b) hb hφ_decay hφ'_decay (a := a) ha hab ha1 hinv  dsimp only at hbank  -- `kadiri_thm_3_1_q1_I` uses `∫ t in Set.Ioo (-T) T`; the bank uses `∫ t in (-T)..T`.  -- For `T ≥ 0` these set integrals agree (`(-T)..T = Ioc (-T) T`, and `Ioo`/`Ioc`  -- differ only on the null set `{T}`), so the two limits coincide eventually.  refine hbank.congr' ?_  filter_upwards [Filter.eventually_ge_atTop (0 : )] with T hT  have hTle : (-T)  T := by linarith  unfold kadiri_thm_3_1_q1_I  dsimp only  congr 1  rw [intervalIntegral.integral_of_le hTle,    MeasureTheory.integral_Ioc_eq_integral_Ioo]