OAI/Combinatorics/SharpThreshold/SpectralBounds.lean

GitHub source file

10 declarations, 0 contain sorry, 0 declared as axioms

  1. theorem OAI.Problem313.finite_spectral_tail line 12

    Source documentation

    A finite spectral tail is controlled by its first degree moment.

    theorem finite_spectral_tail {ι : Type u} (s : Finset ι)
        (mass degree : ι → ℝ) (cutoff : ℝ) (hcutoff : 0 < cutoff)
        (hmass : ∀ i ∈ s, 0 ≤ mass i)
        (hdegree : ∀ i ∈ s, 0 ≤ degree i) :
        (∑ i ∈ s with cutoff < degree i, mass i) ≤
          (∑ i ∈ s, degree i * mass i) / cutoff := by
    …
  2. theorem OAI.Problem313.finite_spectral_split line 34

    Source documentation

    Splitting at a positive spectral cutoff gives a low-degree term and a tail.

    theorem finite_spectral_split {ι : Type u} (s : Finset ι)
        (mass degree : ι → ℝ) (cutoff : ℝ) (hcutoff : 0 < cutoff)
        (hmass : ∀ i ∈ s, 0 ≤ mass i)
        (hdegree : ∀ i ∈ s, 0 ≤ degree i) :
        (∑ i ∈ s, mass i) ≤
          (∑ i ∈ s with degree i ≤ cutoff, mass i) +
          (∑ i ∈ s, degree i * mass i) / cutoff := by
    …
  3. theorem OAI.Problem313.variance_tail_bias_cancellation line 48

    Source documentation

    The bias cancels exactly in the tail term at the bias-dependent cutoff.

    theorem variance_tail_bias_cancellation (sigma ell : ℝ)
        (hsigma : sigma ≠ 0) (hell : ell ≠ 0) :
        sigma ^ 2 / ((sigma * ell / 64) ^ 2 / 2) = 8192 / ell ^ 2 := by
    …
  4. theorem OAI.Problem313.variance_from_spectral_capture line 55

    Source documentation

    Assemble low-degree capture and the Dirichlet identity into a variance bound.

    theorem variance_from_spectral_capture {ι : Type u} (s : Finset ι)
        (mass degree : ι → ℝ) (sigma ell influence lowConstant : ℝ)
        (hsigma : 0 < sigma) (hell : 0 < ell)
        (hmass : ∀ i ∈ s, 0 ≤ mass i)
        (hdegree : ∀ i ∈ s, 0 ≤ degree i)
        (hcapture : (∑ i ∈ s with degree i ≤ (sigma * ell / 64) ^ 2 / 2, mass i)
          ≤ lowConstant * influence)
        (hdirichlet : (∑ i ∈ s, degree i * mass i) = sigma ^ 2 * influence) :
        (∑ i ∈ s, mass i) ≤ (lowConstant + 8192 / ell ^ 2) * influence := by
    …
  5. theorem OAI.Problem313.variance_constant_assembly line 75

    Source documentation

    The two numerical terms fit inside the advertised universal constant.

    theorem variance_constant_assembly (t ell lowConstant : ℝ)
        (ht : 0 < t) (hell : t / 3 ≤ ell)
        (hlow : lowConstant ≤ 256 / t ^ 2) :
        lowConstant + 8192 / ell ^ 2 ≤ (2 : ℝ) ^ 17 / t ^ 2 := by
    …
  6. theorem OAI.Problem313.variance_constant_assembly_budget line 92

    Source documentation

    A more flexible numerical budget, useful with coarse exponential estimates.

    theorem variance_constant_assembly_budget (t ell lowConstant budget : ℝ)
        (ht : 0 < t) (hell : t / 3 ≤ ell)
        (hlow : lowConstant ≤ budget / t ^ 2)
        (hbudget : budget + 73728 ≤ (2 : ℝ) ^ 17) :
        lowConstant + 8192 / ell ^ 2 ≤ (2 : ℝ) ^ 17 / t ^ 2 := by
    …
  7. theorem OAI.Problem313.variance_small_log_assembly line 109

    Source documentation

    A small logarithm only needs the elementary Poincaré bound.

    theorem variance_small_log_assembly (t variance influence : ℝ)
        (ht : 0 < t) (htsmall : t ≤ 16) (hinfluence : 0 ≤ influence)
        (hpoincare : variance ≤ influence / 4) :
        variance ≤ ((2 : ℝ) ^ 17 / t ^ 2) * influence := by
    …
  8. theorem OAI.Problem313.variance_large_log_assembly line 123

    Source documentation

    Final algebraic combination, leaving only the graph-specific spectral input.

    theorem variance_large_log_assembly (t ell variance influence lowConstant : ℝ)
        (ht : 0 < t) (hell : t / 3 ≤ ell) (hinfluence : 0 ≤ influence)
        (hlow : lowConstant ≤ 256 / t ^ 2)
        (hvariance : variance ≤ (lowConstant + 8192 / ell ^ 2) * influence) :
        variance ≤ ((2 : ℝ) ^ 17 / t ^ 2) * influence :=
    …
  9. theorem OAI.Problem313.variance_from_fourier_capture line 132

    Source documentation

    The spectral variance bound for Fourier coefficients indexed by finite supports.

    theorem variance_from_fourier_capture {ι : Type u} [Fintype ι] [DecidableEq ι]
        (coeff : Finset ι → ℝ) (variance sigma ell influence lowConstant : ℝ)
        (hsigma : 0 < sigma) (hell : 0 < ell)
        (hvariance : variance = ∑ S : Finset ι with S.Nonempty, coeff S ^ 2)
        (hdirichlet : (∑ S : Finset ι, (S.card : ℝ) * coeff S ^ 2) = sigma ^ 2 * influence)
        (hcapture : (∑ S : Finset ι with S.Nonempty ∧
            (S.card : ℝ) ≤ (sigma * ell / 64) ^ 2 / 2, coeff S ^ 2)
          ≤ lowConstant * influence) :
        variance ≤ (lowConstant + 8192 / ell ^ 2) * influence := by
    …
  10. theorem OAI.Problem313.variance_le_dirichlet_of_fourier line 158

    Source documentation

    Parseval and the first spectral moment immediately imply Poincaré.

    theorem variance_le_dirichlet_of_fourier {ι : Type u} [Fintype ι] [DecidableEq ι]
        (coeff : Finset ι → ℝ) (variance dirichlet : ℝ)
        (hvariance : variance = ∑ S : Finset ι with S.Nonempty, coeff S ^ 2)
        (hdirichlet : (∑ S : Finset ι, (S.card : ℝ) * coeff S ^ 2) = dirichlet) :
        variance ≤ dirichlet := by
    …