InformationTheory

InformationTheory.Shannon.FisherConvBound

source

Stam convolution Fisher information bound J(pX ∗ g_s) ≤ 1/s #

For any probability density pX and Gaussian kernel g_s of variance s, the Fisher information of the convolution pX ∗ g_s satisfies J(pX ∗ g_s) ≤ 1/s.

Main statements #

Implementation notes #

The proof follows a pointwise Cauchy-Schwarz route: the derivative formula convDensityAdd_deriv1_gaussian_eq gives (deriv p_s x) as an integral, Hölder's inequality with p = q = 2 bounds (logDeriv p_s x)² · p_s x pointwise, and Tonelli's theorem plus the Gaussian second moment ∫ u² g_s u du = s closes the bound.

theorem

InformationTheory.Shannon.FisherInfo.integral_sq_mul_gaussianPDFReal

source
{s : } (hs : 0 < s) :

∫ u, u ^ 2 * gaussianPDFReal 0 ⟨s, _⟩ u ∂volume = s. @audit:ok

Used by
    theorem

    InformationTheory.Shannon.FisherInfo.integrable_sq_mul_gaussianPDFReal

    source

    u ↦ u ^ 2 * gaussianPDFReal 0 ⟨s, _⟩ u is integrable over volume. @audit:ok

    Used by
      theorem

      InformationTheory.Shannon.FisherInfo.convSecondMoment_integrand_integrable

      source
      (pX : ) (_hpX_meas : Measurable pX) (hpX_int : MeasureTheory.Integrable pX MeasureTheory.volume) {s : } (hs : 0 < s) (x : ) :

      y ↦ (x - y) ^ 2 * (pX y * g_s(x - y)) is integrable over volume for each x. @audit:ok

      Used by
        theorem

        InformationTheory.Shannon.FisherInfo.convScore_sq_le_pointwise

        source
        (pX : ) (hpX_nn : ∀ (x : ), 0 pX x) (hpX_meas : Measurable pX) (hpX_int : MeasureTheory.Integrable pX MeasureTheory.volume) {s : } (hs : 0 < s) (x : ) :

        Pointwise Cauchy-Schwarz: (∫ pX y (x-y) g_s(x-y))² ≤ p_s(x) · ∫ pX y (x-y)² g_s(x-y). @audit:ok

        Used by
          theorem

          InformationTheory.Shannon.FisherInfo.convDensityAdd_deriv_eq

          source
          (pX : ) (hpX_meas : Measurable pX) (hpX_int : MeasureTheory.Integrable pX MeasureTheory.volume) {s : } (hs : 0 < s) :

          deriv (convDensityAdd pX g_s) x = ∫ y, pX y * g_s(x-y) * (-(x-y)/s) ∂volume. @audit:ok

          Used by
            theorem

            InformationTheory.Shannon.FisherInfo.convLogDeriv_sq_mul_le

            source
            (pX : ) (hpX_nn : ∀ (x : ), 0 pX x) (hpX_meas : Measurable pX) (hpX_int : MeasureTheory.Integrable pX MeasureTheory.volume) (hpX_mass : (y : ), pX y = 1) {s : } (hs : 0 < s) (x : ) :

            (logDeriv p_s x) ^ 2 * p_s x ≤ (1/s²) * ∫ (x-y)² * (pX y * g_s(x-y)) ∂volume. @audit:ok

            Used by
              theorem

              InformationTheory.Shannon.FisherInfo.gaussianConv_fisher_le_inv_var

              source
              (pX : ) (hpX_nn : ∀ (x : ), 0 pX x) (hpX_meas : Measurable pX) (hpX_int : MeasureTheory.Integrable pX MeasureTheory.volume) (hpX_mass : (y : ), pX y = 1) {s : } (hs : 0 < s) :

              Stam convolution Fisher information bound: J(pX ∗ g_s) ≤ 1/s for any probability density pX and Gaussian kernel g_s of variance s > 0. @audit:ok

              Used by