import OAI.Analysis.Fourier.LocalSmoothing.DyadicAssembly import OAI.Analysis.Fourier.LocalSmoothing.DyadicEstimates open OAI.Analysis.Fourier.Wave open _root_.MeasureTheory open scoped ENNReal noncomputable section namespace OAI.Analysis.Fourier.LocalSmoothing /-- The source's positive-loss summation step, retaining the exact Bessel potential on the right and proving both integrability statements. -/ theorem localSmoothing_sobolev_of_frequency_bounds {n : ℕ} {p : ℝ≥0∞} (hp : 1 ≤ p) (hpt : p ≠ ⊤) (s : ℝ) (A : ℕ) {C : ℝ} (hC : 0 ≤ C) (hfreq : ∀ h : ℝ, 0 < h → h ≤ 1 → ∀ f : SchwartzMap (Space n) ℂ, eLpNorm (fun z : Space n × ℝ ↦ frequencyHalfWave smoothingAnnularCutoff h z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal (C * h^(-s) * (1+Real.log (1/h))^A) * eLpNorm f p volume) {ε : ℝ} (hε : 0 < ε) {C₀ : ℝ} (hC₀ : 0 ≤ C₀) (hlow : ∀ f : SchwartzMap (Space n) ℂ, eLpNorm (fun z : Space n × ℝ ↦ frequencyHalfWave smoothingLowCutoff 1 z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal C₀ * eLpNorm (besselPotential (s+ε) f) p volume) : ∃ D : ℝ, 0 < D ∧ ∀ f : SchwartzMap (Space n) ℂ, MemLp (besselPotential (s+ε) f) p volume ∧ MemLp (fun z : Space n × ℝ ↦ halfWaveIntegral z.2 f z.1) p (observationMeasure n) ∧ eLpNorm (fun z : Space n × ℝ ↦ halfWaveIntegral z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal D * eLpNorm (besselPotential (s+ε) f) p volume := by obtain ⟨C₁,hC₁,hann⟩ := smoothing_dyadic_sobolev_estimate hp hpt s A hC hfreq ε let w : ℕ → ℝ := fun j ↦ smoothingDyadicScale j ^ ε * smoothingDyadicLog j ^ A have hw₀ (j : ℕ) : 0 ≤ w j := mul_nonneg (Real.rpow_nonneg (smoothingDyadicScale_pos j).le _) (pow_nonneg (smoothingDyadicLog_pos j).le _) have hw : Summable w := summable_smoothing_sobolev_weights A hε let D : ℝ := C₀ + C₁ * ∑' j, w j have hD : 0 ≤ D := add_nonneg hC₀ (mul_nonneg hC₁ (tsum_nonneg hw₀)) refine ⟨D+1, by linarith, ?_⟩ intro f have hmem := OAI.Analysis.Fourier.BesselPotential.memLp_besselPotential (s+ε) f p obtain ⟨hsol,hbound⟩ := smoothing_bound_of_dyadic_bounds f hp hmem hw₀ hw hC₀ hC₁ (hlow f) (fun j ↦ hann j f) refine ⟨hmem,hsol,hbound.trans ?_⟩ exact mul_le_mul' (ENNReal.ofReal_le_ofReal (show D ≤ D+1 by linarith)) le_rfl /-- The source's logarithmic summation step: `B>A+1` is exactly the summability condition, and there is no positive differential loss. -/ theorem localSmoothing_log_of_frequency_bounds {n : ℕ} {p : ℝ≥0∞} (hp : 1 ≤ p) (hpt : p ≠ ⊤) (s : ℝ) (A : ℕ) {C : ℝ} (hC : 0 ≤ C) (hfreq : ∀ h : ℝ, 0 < h → h ≤ 1 → ∀ f : SchwartzMap (Space n) ℂ, eLpNorm (fun z : Space n × ℝ ↦ frequencyHalfWave smoothingAnnularCutoff h z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal (C * h^(-s) * (1+Real.log (1/h))^A) * eLpNorm f p volume) {B : ℝ} (hB : (A : ℝ)+1 < B) {C₀ : ℝ} (hC₀ : 0 ≤ C₀) (hlow : ∀ f : SchwartzMap (Space n) ℂ, eLpNorm (fun z : Space n × ℝ ↦ frequencyHalfWave smoothingLowCutoff 1 z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal C₀ * eLpNorm (logBesselPotential s B f) p volume) : ∃ D : ℝ, 0 < D ∧ ∀ f : SchwartzMap (Space n) ℂ, MemLp (logBesselPotential s B f) p volume ∧ MemLp (fun z : Space n × ℝ ↦ halfWaveIntegral z.2 f z.1) p (observationMeasure n) ∧ eLpNorm (fun z : Space n × ℝ ↦ halfWaveIntegral z.2 f z.1) p (observationMeasure n) ≤ ENNReal.ofReal D * eLpNorm (logBesselPotential s B f) p volume := by obtain ⟨C₁,hC₁,hann⟩ := smoothing_dyadic_log_estimate hp hpt s A hC hfreq B let w : ℕ → ℝ := fun j ↦ smoothingDyadicLog j ^ A * smoothingDyadicLog j ^ (-B) have hw₀ (j : ℕ) : 0 ≤ w j := mul_nonneg (pow_nonneg (smoothingDyadicLog_pos j).le _) (Real.rpow_nonneg (smoothingDyadicLog_pos j).le _) have hw : Summable w := summable_smoothing_log_weights A hB let D : ℝ := C₀ + C₁ * ∑' j, w j have hD : 0 ≤ D := add_nonneg hC₀ (mul_nonneg hC₁ (tsum_nonneg hw₀)) refine ⟨D+1, by linarith, ?_⟩ intro f have hmem := OAI.Analysis.Fourier.BesselPotential.memLp_logBesselPotential s B f p obtain ⟨hsol,hbound⟩ := smoothing_bound_of_dyadic_bounds f hp hmem hw₀ hw hC₀ hC₁ (hlow f) (fun j ↦ hann j f) refine ⟨hmem,hsol,hbound.trans ?_⟩ exact mul_le_mul' (ENNReal.ofReal_le_ofReal (show D ≤ D+1 by linarith)) le_rfl end OAI.Analysis.Fourier.LocalSmoothing