OAI/Analysis/AffineBernstein/ActualLogMeasureFurther.lean

GitHub source file

19 declarations, 0 contain sorry, 0 declared as axioms

  1. theorem OAI.AffineBernstein.no_eventual_polynomial_exponential_growth line 33

    Source comment

    The quantifier order in bounds.tex: every finite box has its own eventual index. No common threshold for infinitely many boxes is needed.
    theorem no_eventual_polynomial_exponential_growth
        (M : ℕ → ℕ → ℝ) (k l₀ : ℕ) {c q C : ℝ}
        (hc : 0 < c) (hq : 1 < q)
        (hpos : ∀ᶠ j in atTop, c ≤ M j l₀)
        (hstep : ∀ l, l₀ ≤ l → ∀ᶠ j in atTop, q * M j l ≤ M j (l+1))
        (hupper : ∀ L, l₀ ≤ L → ∀ᶠ j in atTop, M j L ≤ C * (L+1:ℝ)^k) : False := by
    …
  2. theorem OAI.AffineBernstein.no_eventual_polynomial_shell_growth line 68

    Source comment

    The source shell inequality has precisely the exponential recurrence required above. Integrals may vary with the approximating index.
    theorem no_eventual_polynomial_shell_growth
        (M : ℕ → ℕ → ℝ) (k l₀ : ℕ) {c A C : ℝ}
        (hc : 0 < c) (hA : 0 < A)
        (hpos : ∀ᶠ j in atTop, c ≤ M j l₀)
        (hshell : ∀ l, l₀ ≤ l → ∀ᶠ j in atTop,
          M j l ≤ A * (M j (l+1)-M j l))
        (hupper : ∀ L, l₀ ≤ L → ∀ᶠ j in atTop, M j L ≤ C * (L+1:ℝ)^k) : False := by
    …
  3. theorem OAI.AffineBernstein.lipschitzWith_mul_unit line 85

    Source comment

    A product of two unit-bounded Lipschitz functions has the sum of their Lipschitz constants. Here the multiplication is the ring multiplication on ℝ.
    theorem lipschitzWith_mul_unit {X : Type*} [PseudoMetricSpace X]
        {f g : X → ℝ} {K L : NNReal} (hf : LipschitzWith K f) (hg : LipschitzWith L g)
        (hfb : ∀ x, ‖f x‖ ≤ 1) (hgb : ∀ x, ‖g x‖ ≤ 1) :
        LipschitzWith (K+L) (fun x => f x*g x) := by
    …
  4. theorem OAI.AffineBernstein.lipschitzWith_prod_unit line 104
  5. theorem OAI.AffineBernstein.smoothTransition_exists_lipschitz line 123

    Source comment

    The fixed smooth transition has some finite global Lipschitz constant; constancy off [0,1] makes the derivative bound compact.
    theorem smoothTransition_exists_lipschitz : ∃ K : NNReal, LipschitzWith K Real.smoothTransition := by
    …
  6. def OAI.AffineBernstein.logarithmicCutoff line 149

    Source comment

    Closed boxes in the literal logarithmic base coordinates. A fixed-width smooth box cutoff; its derivatives have no dependence on L apart from coordinate translations.
    def logarithmicCutoff (k : ℕ) (L : ℝ) (x : Space k) : ℝ :=
    …

    Mentions: OAI.AffineBernstein.Space

  7. theorem OAI.AffineBernstein.logarithmicCutoff_mem_Icc line 152
  8. theorem OAI.AffineBernstein.logarithmicCutoff_one line 160
  9. theorem OAI.AffineBernstein.logarithmicCutoff_zero line 169
  10. theorem OAI.AffineBernstein.logarithmicCutoff_smooth line 180
  11. theorem OAI.AffineBernstein.logarithmicCutoff_uniform_lipschitz line 190

    Source comment

    One constant controls every translated logarithmic cutoff.
    theorem logarithmicCutoff_uniform_lipschitz (k : ℕ) :
        ∃ K : NNReal, 0 < K ∧ ∀ L : ℝ, LipschitzWith K (logarithmicCutoff k L) := by
    …

    Mentions: OAI.AffineBernstein.logarithmicCutoff

  12. theorem OAI.AffineBernstein.logarithmicBox_mono line 238
  13. theorem OAI.AffineBernstein.logarithmicBox_isClosed line 241
  14. theorem OAI.AffineBernstein.logarithmicBox_isCompact line 247
  15. theorem OAI.AffineBernstein.logarithmicCutoff_hasCompactSupport line 258
  16. theorem OAI.AffineBernstein.logarithmicCutoff_fderiv_eq_zero_of_mem line 267

    Source comment

    The derivative vanishes on the inner closed box, including its faces.
    theorem logarithmicCutoff_fderiv_eq_zero_of_mem {k : ℕ} {L : ℝ} {x : Space k}
        (hx : x ∈ logarithmicBox k L) : fderiv ℝ (logarithmicCutoff k L) x = 0 := by
    …

    Mentions: OAI.AffineBernstein.Space, OAI.AffineBernstein.logarithmicBox, OAI.AffineBernstein.logarithmicCutoff

  17. theorem OAI.AffineBernstein.logarithmicCutoff_fderiv_eq_zero_of_outside line 277

    Source comment

    The same holds at and outside the outer faces; no zero-boundary-mass assumption is needed in the ensuing shell estimate.
    theorem logarithmicCutoff_fderiv_eq_zero_of_outside {k : ℕ} {L : ℝ} {x : Space k}
        (hx : ∃ i, L+1 ≤ |x i|) : fderiv ℝ (logarithmicCutoff k L) x = 0 := by
    …

    Mentions: OAI.AffineBernstein.Space, OAI.AffineBernstein.logarithmicCutoff

  18. theorem OAI.AffineBernstein.logarithmic_shell_growth line 287

    Source comment

    The logarithmic Poincaré inequality turns the genuine smooth cutoff into mass growth in a fixed-width shell. Measures need only be finite on this box.
    theorem logarithmic_shell_growth {k : ℕ} {L : ℝ} (μ : Measure (Space k))
        {K : NNReal} (hK : LipschitzWith K (logarithmicCutoff k L)) (hfin : μ (logarithmicBox k (L+1)) ≠ ⊤)
        {A : ℝ} (hA : 0 ≤ A)
        (hP : (∫ x, (logarithmicCutoff k L x)^2 ∂μ) ≤
          A * ∫ x, ‖fderiv ℝ (logarithmicCutoff k L) x‖^2 ∂μ) :
        μ.real (logarithmicBox k L) ≤ A*(K:ℝ)^2 *
          (μ.real (logarithmicBox k (L+1))-μ.real (logarithmicBox k L)) := by
    …

    Mentions: OAI.AffineBernstein.Space, OAI.AffineBernstein.logarithmicCutoff, OAI.AffineBernstein.logarithmicBox

  19. theorem OAI.AffineBernstein.no_polynomial_logarithmic_poincare line 364

    Source comment

    The growth contradiction, for actual Borel measures and actual smooth cutoffs, with only eventual availability on each finite logarithmic box.
    theorem no_polynomial_logarithmic_poincare {k l₀ : ℕ} (μ : ℕ → Measure (Space k))
        {c A C : ℝ} (hc : 0 < c) (hA : 0 < A)
        (hpos : ∀ᶠ j in atTop, c ≤ (μ j).real (logarithmicBox k l₀))
        (hfin : ∀ L : ℕ, l₀ ≤ L → ∀ᶠ j in atTop, μ j (logarithmicBox k L) ≠ ⊤)
        (hupper : ∀ L : ℕ, l₀ ≤ L → ∀ᶠ j in atTop,
          (μ j).real (logarithmicBox k L) ≤ C*(L+1:ℝ)^k)
        (hP : ∀ L : ℕ, l₀ ≤ L → ∀ᶠ j in atTop,
          (∫ x, (logarithmicCutoff k L x)^2 ∂μ j) ≤
            A * ∫ x, ‖fderiv ℝ (logarithmicCutoff k L) x‖^2 ∂μ j) : False := by
    …

    Mentions: OAI.AffineBernstein.Space, OAI.AffineBernstein.logarithmicBox, OAI.AffineBernstein.logarithmicCutoff