Documentation

RequestProject.HadamardCn3LocalGapBridge

Local-Gap Bridge Layer #

This file contains the shared local-gap support used by both the residual and fixed-n modules.

The first public theorems a reader should inspect are:

Most other declarations here are implementation detail for those bridge statements or for downstream residual/fixed-n estimates.

The Gaussian normalizer factors exactly as the tex prefactor times gaussianF.

theorem coreMass_le_gaussianF (n : ) (t : ) (ht : 0 < t) :

The inner-core Gaussian mass is bounded above by the full Gaussian mass.

theorem one_le_dim_of_three_le (n : ) (hn : 3 n) :
1 dim n

For n ≥ 3, the edge-space dimension is at least one.

theorem exp_neg_dim_over_eight_le_exp_neg_nsq_div_thirty_two {n : } (hn : 2 n) :
Real.exp (-((dim n) / 8)) Real.exp (-(n ^ 2 / 32))

The standard annulus factor exp(-dim/8) is at most exp(-n^2/32) once n ≥ 2.

theorem exp_neg_c_mul_dim_le_exp_neg_c_div_four_mul_nsq {n : } (hn : 2 n) {c : } (hc : 0 c) :
Real.exp (-(c * (dim n))) Real.exp (-(c / 4 * n ^ 2))

A generic conversion from exponential decay in dim n to exponential decay in , using dim n ≥ n² / 4.

theorem dim_div_t_le_inv_C (n t : ) {C : } (hC_pos : 0 < C) (hC_one : 1 C) (hn_pos : 0 < n) (ht : t C * n ^ 3) :
(dim n) / t 1 / C

At the t \ge C n^3 scale, dim / t is O(1/C).

theorem n_mul_sq_dim_div_t_le_inv_C (n t : ) {C : } (hC_pos : 0 < C) (hC_one : 1 C) (hn_pos : 0 < n) (ht : t C * n ^ 3) :
n * ((dim n) / t) ^ 2 1 / C

At the t \ge C n^3 scale, the quartic n (dim/t)^2 parameter is O(1/C).

theorem n_cube_div_t_le_inv_C (n t : ) {C : } (hC_pos : 0 < C) (hn_pos : 0 < n) (ht : t C * n ^ 3) :
n ^ 3 / t 1 / C

At the t \ge C n^3 scale, the cubic bridge parameter n^3/t is O(1/C).

noncomputable def gaussianEvenMomentEnvelope (K : ) :

A finite envelope for the even Gaussian moment coefficients up to degree 2K.

Equations
Instances For
    theorem quarticCorr_pointwise_bound_uniform :
    ∃ (C : ), 0 < C ∀ (n : ) (lam : Fin nFin n), |quarticCorr n lam| C * sNorm n lam ^ 2

    The quartic correction admits a uniform O(s^2) bound.

    theorem cubicT_pointwise_bound_uniform :
    ∃ (C : ), 0 < C ∀ (n : ) (lam : Fin nFin n), |cubicT n lam| C * sNorm n lam ^ (3 / 2)

    Dimension-free cubic-form bound.

    theorem psi_pow_sub_correctedCoreIntegrand_norm_le_uniform :
    ∃ (c₂ : ) (C₂ : ), 0 < c₂ 0 < C₂ ∀ (n t : ) (mu : Cn3Torus.Edge n), Cn3Torus.sqNormEdge n mu c₂4 * t * C₂ * Cn3Torus.sqNormEdge n mu ^ 3 1Cn3Torus.psi n mu ^ (4 * t) - correctedCoreIntegrand n t mu 8 * t * C₂ * Cn3Torus.sqNormEdge n mu ^ 3 * Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu + 4 * t * quarticCorr n (matrixOfEdge n mu))
    theorem cubicCoreIntegrand_re (n t : ) (mu : Cn3Torus.Edge n) :
    (cubicCoreIntegrand n t mu).re = Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu) * Real.cos (4 * t * cubicT n (matrixOfEdge n mu))

    The real part of the cubic core integrand is the Gaussian-cosine model from cubic_core_lower_bound.

    theorem edgeEuclidBall_subset_edgeBox (n : ) {r delta : } (hr_nonneg : 0 r) (hr : r delta) :

    A Euclidean ball in edge coordinates sits inside the coordinate box of the same radius.

    theorem cs_helper {α : Type u_1} [MeasurableSpace α] (μ : MeasureTheory.Measure α) {f g : α} (hf_ae : MeasureTheory.AEStronglyMeasurable f μ) (hg_ae : MeasureTheory.AEStronglyMeasurable g μ) (hf_nonneg : 0 ≤ᵐ[μ] f) (hg_nonneg : 0 ≤ᵐ[μ] g) (hf_int : MeasureTheory.Integrable (fun (x : α) => f x ^ 2) μ) (hg_int : MeasureTheory.Integrable (fun (x : α) => g x ^ 2) μ) :
    (x : α), f x * g x μ ( (x : α), f x ^ 2 μ) ^ (1 / 2) * ( (x : α), g x ^ 2 μ) ^ (1 / 2)

    Real Cauchy-Schwarz on an arbitrary measure space, specialized to the controls used in the four-bridge argument.

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For
      theorem cubicT_sq_full_gaussian_correction_exact (n : ) (t : ) (ht : 0 < t) :
      8 * t ^ 2 * (mu : Cn3Torus.Edge n), cubicT n (matrixOfEdge n mu) ^ 2 * Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu) = (n.choose 3) / (8 * t) * gaussianF (dim n) t
      Equations
      • One or more equations did not get rendered due to their size.
      theorem cubicT_fourth_full_gaussian_bound_explicit (n : ) (t : ) (ht : 1 t) :
      (mu : Cn3Torus.Edge n), cubicT n (matrixOfEdge n mu) ^ 4 * Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu) 7 * 6 ^ 12 * gaussianEvenMomentEnvelope 6 ^ 6 * n ^ 6 / t ^ 6 * gaussianF (dim n) t
      theorem cubic_core_second_order_gap_uniform :
      ∃ (C : ), 0 < C ∀ (n : ) (t : ), 1 t|( (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu) * Real.cos (4 * t * cubicT n (matrixOfEdge n mu))) - (coreMass (dim n) t - 8 * t ^ 2 * (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, cubicT n (matrixOfEdge n mu) ^ 2 * Real.exp (-2 * t * Cn3Torus.sqNormEdge n mu))| C * n ^ 6 / t ^ 2 * gaussianF (dim n) t

      On the dynamic core, the cubic oscillatory factor admits the explicit second-order expansion with a fourth-moment remainder. This is the first quantitative theorem that uses the new T^4 machinery.

      Removing the quintic phase costs only the size of the real quintic phase itself.

      theorem small_ball_gaussian_decay_uniform :
      ∃ (r₀ : ), 0 < r₀ r₀ < Real.pi / 4 ∀ (n : ) (mu : Cn3Torus.Edge n), Cn3Torus.sqNormEdge n mu r₀ ^ 2∀ (t : ), 1 tCn3Torus.psi n mu ^ (4 * t) Real.exp (-(3 * t * Cn3Torus.sqNormEdge n mu / 2))

      A uniform small-ball decay bound for the edge characteristic function.

      theorem edgePrimaryLocalAnnulus_integral_bound_coreMass_uniform :
      ∃ (r₁ : ) (C₆ : ), 0 < r₁ 0 < C₆ r₁ < Real.pi / 4 ∀ (n : ), 1 dim n∀ {r delta : }, 0 < rr r₁∀ (t : ), 1 t (mu : Cn3Torus.Edge n) in (edgeBox n delta edgeEuclidBall n r) \ edgeCoreRegion n t, Cn3Torus.psi n mu ^ (4 * t) C₆ * Real.exp (-((dim n) / 8)) * coreMass (dim n) t

      The primary annulus contribution is uniformly exponentially small relative to coreMass.

      The tex prefactor is always nonnegative.

      Real parts of complex set integrals are controlled by the integral of the complex norm difference.

      theorem exp_neg_mul_t_le_exp_neg_nsq_mul_gaussianScale_uniform_ge_three (a : ) (ha : 0 < a) :
      ∃ (c : ) (C : ), 0 < c 0 < C ∀ (n : ), 3 n∀ (t : ), t C * n ^ 3Real.exp (-a * t) Real.exp (-(c * n ^ 2)) * gaussianScale n t

      A quantitative version of exp_neg_mul_t_le_eps_gaussianScale_uniform_ge_three with an explicit exp(-c n^2) factor on the Gaussian scale.

      theorem qpow_le_exp_neg_nsq_mul_gaussianScale_uniform_ge_three (q : ) (hq_pos : 0 < q) (hq_lt : q < 1) :
      ∃ (c : ) (C : ), 0 < c 0 < C ∀ (n : ), 3 n∀ (t : ), t C * n ^ 3q ^ (4 * t) Real.exp (-(c * n ^ 2)) * gaussianScale n t

      A strict geometric decay on the far shell is bounded by an explicit exp(-c n^2) multiple of the Gaussian scale at the n^3 threshold.

      theorem exp_neg_mul_t_div_n_two_thirds_le_exp_neg_nsq_mul_gaussianScale_uniform_ge_three (a : ) (ha : 0 < a) :
      ∃ (c : ) (C : ), 0 < c 0 < C ∀ (n : ), 3 n∀ (t : ), t C * n ^ 3Real.exp (-(a * t / n ^ (2 / 3))) Real.exp (-(c * n ^ 2)) * gaussianScale n t

      A quantitative version of exp_neg_mul_t_div_n_two_thirds_le_eps_gaussianScale_uniform_ge_three with an explicit exp(-c n^2) factor on the Gaussian scale.

      theorem coreMass_gap_le_exp_gaussianF :
      ∃ (c : ), 0 < c ∀ (n : ) (t : ), 1 t|coreMass (dim n) t - gaussianF (dim n) t| Real.exp (-(c * (dim n))) * gaussianF (dim n) t

      The truncated Gaussian core differs from the full Gaussian mass by an exponentially small factor in the dimension.

      The even far shell is measurable.

      Continuity of the repaired core integrand in edge coordinates.

      Continuity of the quartic core integrand in edge coordinates.

      Continuity of the cubic core integrand in edge coordinates.

      theorem integralOn_mono_of_nonneg {α : Type u_1} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {s t : Set α} {f : α} (hsubset : s t) (hs : MeasurableSet s) (ht : MeasurableSet t) (hf_int : MeasureTheory.IntegrableOn f t μ) (hf_nonneg : 0 ≤ᵐ[μ.restrict t] f) :
      (x : α) in s, f x μ (x : α) in t, f x μ

      Monotonicity of set integrals for nonnegative real functions.

      theorem localIntegral_sub_core_abs_le_annulus_plus_far (n t : ) (hbox : (dim n) / t Cn3Torus.delta ^ 2) (r : ) :
      have box := edgeBox n Cn3Torus.delta; have core := edgeCoreRegion n t; have A := (box edgeEuclidBall n r) \ core; have B := (box \ edgeEuclidBall n r) \ core; |Cn3Torus.localIntegral n t - (mu : Cn3Torus.Edge n) in core, (Cn3Torus.psi n mu ^ (4 * t)).re| ( (mu : Cn3Torus.Edge n) in A, Cn3Torus.psi n mu ^ (4 * t)) + (mu : Cn3Torus.Edge n) in B, Cn3Torus.psi n mu ^ (4 * t)

      Splitting the local box into the dynamic core plus the two shell pieces used in the final argument.

      theorem exactCore_correctedCore_gap_abs_le_gaussianF_short :
      ∃ (c : ) (K : ), 0 < c 0 < K ∀ (n : ), 3 n∀ (t : ), 1 t(dim n) / t cn ^ 6 / t ^ 2 c|( (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (Cn3Torus.psi n mu ^ (4 * t)).re) - (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (correctedCoreIntegrand n t mu).re| K * (n ^ 6 / t ^ 2) * gaussianF (dim n) t

      Public bridge 1: the exact core integral is close to the corrected core model on the Gaussian scale with n^6 / t^2 loss.

      theorem correctedCore_quarticCore_gap_abs_le_gaussianF_short :
      ∃ (c : ) (K : ), 0 < c 0 < K ∀ (n : ), 2 n∀ (t : ), 1 t(dim n) / t cn * ((dim n) / t) ^ 2 c|( (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (correctedCoreIntegrand n t mu).re) - (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (quarticCoreIntegrand n t mu).re| K * (n ^ (5 / 2) / t ^ (3 / 2)) * gaussianF (dim n) t

      Public bridge 2: the corrected-core to quartic-core bridge with the natural t^{-3/2} decay.

      theorem quarticCore_cubicCore_gap_abs_le_gaussianF_short :
      ∃ (c : ) (K : ), 0 < c 0 < K ∀ (n t : ), 1 t(dim n) / t cn * ((dim n) / t) ^ 2 c|( (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (quarticCoreIntegrand n t mu).re) - (mu : Cn3Torus.Edge n) in edgeCoreRegion n t, (cubicCoreIntegrand n t mu).re| K * (n ^ 2 / t) * gaussianF (dim n) t

      Public bridge 3: stripping the quartic correction costs only n^2 / t on the Gaussian scale.

      theorem abs_sub_le_sum_abs_sub_six (x₀ x₁ x₂ x₃ x₄ x₅ x₆ : ) :
      |x₀ - x₆| |x₀ - x₁| + |x₁ - x₂| + |x₂ - x₃| + |x₃ - x₄| + |x₄ - x₅| + |x₅ - x₆|

      A six-step telescoping chain for absolute differences.

      theorem div_le_eps_six_of_scaled_bound {K C ε : } (hC : 0 < C) ( : 0 < ε) (hscaled : 6 * K / ε C) :
      K / C ε / 6

      Convert the bookkeeping constraint 6 K / ε ≤ C into the useful bound K / C ≤ ε / 6.