Documentation

Atlas.IntroductionToFunctionalAnalysis.code.LpSpaces

noncomputable def LpSpaces.eLpNorm {α : Type u_1} [MeasurableSpace α] {E : Type u_2} [NormedAddCommGroup E] (f : αE) (p : ENNReal) (μ : MeasureTheory.Measure α := by volume_tac) :

The $L^p$ (extended) norm of a function $f : α → E$ with respect to the measure $μ$. For $1 ≤ p < ∞$, this is $\|f\|_p = \left(\int |f|^p \, dμ\right)^{1/p}$, and for $p = ∞$ it is the essential supremum $\|f\|_∞ = \operatorname{ess\,sup}|f|$. This is a thin wrapper around MeasureTheory.eLpNorm.

Instances For

    For a continuous function $f : α → E$ on a compact space $α$ equipped with a Borel measure of full support, the essential supremum $\|f\|_{L^\infty(μ)}$ coincides with the usual supremum norm $\|f\|_∞$. This is the continuous-case statement of the essential supremum property: on $[a,b]$ with $f \in C([a,b])$, $\|f\|_{L^\infty([a,b])} = \|f\|_\infty$.

    theorem LpSpaces.essSup_properties :
    (∀ {α : Type u_3} [inst : MeasurableSpace α] {μ : MeasureTheory.Measure α} {E : Type u_4} [inst_1 : ENorm E] (f : αE), ∀ᵐ (x : α) μ, f x‖ₑ MeasureTheory.eLpNormEssSup f μ) ∀ {α : Type u_5} [inst : TopologicalSpace α] [inst_1 : CompactSpace α] [inst_2 : MeasurableSpace α] [BorelSpace α] {μ : MeasureTheory.Measure α} [μ.IsOpenPosMeasure] {E : Type u_6} [inst_5 : SeminormedAddCommGroup E] (f : C(α, E)), MeasureTheory.eLpNormEssSup (fun (x : α) => f x) μ = f‖ₑ

    Essential supremum properties: (1) for any measurable function $f : α → E$ we have $|f(x)| ≤ \|f\|_{L^\infty(μ)}$ for almost every $x$, and (2) on a compact Hausdorff space with a Borel measure of full support, the essential supremum of a continuous function $f : C(α, E)$ equals its usual supremum norm.

    theorem LpSpaces.holder_integrable {α : Type u_3} {m : MeasurableSpace α} {μ : MeasureTheory.Measure α} {𝕜 : Type u_4} [NormedField 𝕜] {p q : ENNReal} {f g : α𝕜} (hf : MeasureTheory.MemLp f p μ) (hg : MeasureTheory.MemLp g q μ) [hpq : p.HolderConjugate q] :

    Integrability part of Hölder's inequality: if $f \in L^p(μ)$ and $g \in L^q(μ)$ with $\frac{1}{p} + \frac{1}{q} = 1$, then the pointwise product $f \cdot g$ is integrable.

    theorem LpSpaces.holder_eLpNorm {α : Type u_3} {m : MeasurableSpace α} {μ : MeasureTheory.Measure α} {𝕜 : Type u_4} [NormedField 𝕜] {p q : ENNReal} {f g : α𝕜} (hf : MeasureTheory.MemLp f p μ) (hg : MeasureTheory.MemLp g q μ) [hpq : p.HolderConjugate q] :

    Norm part of Hölder's inequality: if $f \in L^p(μ)$ and $g \in L^q(μ)$ with $\frac{1}{p} + \frac{1}{q} = 1$, then $\|fg\|_{L^1(μ)} ≤ \|f\|_{L^p(μ)} \cdot \|g\|_{L^q(μ)}$.

    theorem LpSpaces.holder_inequality {α : Type u_3} {m : MeasurableSpace α} {μ : MeasureTheory.Measure α} {𝕜 : Type u_4} [NormedField 𝕜] {p q : ENNReal} {f g : α𝕜} (hf : MeasureTheory.MemLp f p μ) (hg : MeasureTheory.MemLp g q μ) [hpq : p.HolderConjugate q] :

    Hölder's inequality for $L^p$ spaces: if $1 ≤ p, q ≤ ∞$ with $\frac{1}{p} + \frac{1}{q} = 1$ and $f, g : α → 𝕜$ are measurable with $f \in L^p(μ)$ and $g \in L^q(μ)$, then $fg$ is integrable and $\int_E |fg| \, dμ ≤ \|f\|_{L^p(μ)} \cdot \|g\|_{L^q(μ)}$.

    Minkowski's inequality for $L^p$ spaces: for $1 ≤ p ≤ ∞$ and measurable functions $f, g : α → E$, $\|f + g\|_{L^p(μ)} ≤ \|f\|_{L^p(μ)} + \|g\|_{L^p(μ)}$. This is the triangle inequality for the $L^p$ norm.

    Riesz-Fischer theorem: for $1 ≤ p ≤ ∞$ and a complete normed space $E$, the space $L^p(μ; E)$ is complete, hence a Banach space.

    def LpSpaces.Lp (E : Type u_3) [NormedAddCommGroup E] (p : ENNReal) {α : Type u_4} [MeasurableSpace α] (μ : MeasureTheory.Measure α) :
    Type (max u_3 u_4)

    The $L^p$ space $L^p(μ; E) = \{f : α → E : f \text{ measurable and } \|f\|_p < ∞\}$, quotiented by the equivalence relation of almost-everywhere equality. Implemented as a thin wrapper around MeasureTheory.Lp.

    Instances For
      theorem LpSpaces.lintegral_eq_iSup_lintegral_Icc_inter (S : Set ) (hS : MeasurableSet S) (g : ENNReal) (hg : Measurable g) :
      ∫⁻ (x : ) in S, g x = ⨆ (n : ), ∫⁻ (x : ) in Set.Icc (-n) n S, g x

      For a measurable set $S \subset ℝ$ and a measurable function $g : ℝ → [0, ∞]$, the integral $\int_S g$ equals the supremum over $n ∈ ℕ$ of the truncated integrals $\int_{[-n,n] \cap S} g$. This is a monotone convergence statement used to characterize $L^p$ membership via restricted integrals.

      theorem LpSpaces.memLp_iff_iSup_lintegral_Icc_inter_lt_top {F : Type u_3} [NormedAddCommGroup F] [MeasurableSpace F] [BorelSpace F] [SecondCountableTopology F] {p : ENNReal} (hp_ne_zero : p 0) (hp_ne_top : p ) (S : Set ) (hS : MeasurableSet S) {f : F} (hf : Measurable f) :

      Characterization of $L^p$ membership in terms of restricted integrals: for $E \subset ℝ$ measurable and $1 ≤ p < ∞$, a measurable function $f$ belongs to $L^p(E)$ if and only if $\lim_{n \to \infty} \int_{[-n, n] \cap E} |f|^p < ∞$.

      @[implicit_reducible]
      noncomputable instance LpSpaces.Lp_instNormedAddCommGroup {α : Type u_1} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {E : Type u_2} [NormedAddCommGroup E] {p : ENNReal} [Fact (1 p)] :

      $L^p(μ; E)$ is a normed additive commutative group for $1 ≤ p ≤ ∞$, with norm $\|\cdot\|_p$. This is part of the statement that $L^p(E)$ is a normed vector space.

      @[implicit_reducible]
      instance LpSpaces.Lp_instNormedSpace {α : Type u_1} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {E : Type u_2} [NormedAddCommGroup E] {𝕜 : Type u_3} [NormedField 𝕜] [NormedSpace 𝕜 E] {p : ENNReal} [Fact (1 p)] :
      NormedSpace 𝕜 (Lp E p μ)

      $L^p(μ; E)$ is a normed vector space over $𝕜$ for $1 ≤ p ≤ ∞$. Together with the additive group instance, this gives the statement that $L^p(E)$ is a normed vector space under $\|\cdot\|_p$.

      theorem LpSpaces.continuous_vanishing_endpoints_dense_in_Lp (a b : ) (hab : a < b) (p : ENNReal) (hp1 : 1 p) (hp_top : p ) (f : ) (hf : MeasureTheory.MemLp f p (MeasureTheory.volume.restrict (Set.Icc a b))) (ε : ENNReal) ( : 0 < ε) :

      Continuous functions vanishing at the endpoints are dense in $L^p([a, b])$: for $a < b$, $1 ≤ p < ∞$, $f \in L^p([a, b])$ and $\varepsilon > 0$, there exists $g \in C([a, b])$ with $g(a) = g(b) = 0$ such that $\|f - g\|_p < \varepsilon$.

      theorem LpSpaces.memLp_of_polynomial_decay {E : Type u_3} [NormedAddCommGroup E] {f : E} (hf_meas : MeasureTheory.AEStronglyMeasurable f MeasureTheory.volume) {C : } (hC : 0 C) {q : } (hq : 1 < q) (hbound : ∀ (x : ), f x C * (1 + |x|) ^ (-q)) {p : ENNReal} (hp : 1 p) :

      Functions with polynomial decay are in $L^p(ℝ)$ for all $p ≥ 1$: if $f : ℝ → E$ is measurable and there exist constants $C ≥ 0$ and $q > 1$ such that $\|f(x)\| ≤ C(1 + |x|)^{-q}$ for almost every $x \in ℝ$, then $f \in L^p(ℝ)$ for every $1 ≤ p ≤ ∞$.