Documentation

SHSLib.Gardiner.Ito.Calculus.Integral

The Itô integral of a general nonanticipating integrand (Gardiner §4.2.1, §4.2.4, §4.2.6) #

For an Itô integrand G on [0, T] (Gardiner.Ito.IsItoIntegrand) and a Wiener process W with respect to 𝓕 (Gardiner.Ito.IsFWiener), Gardiner.Ito.itoInt W 𝓕 P T G is the Itô integral ∫₀ᵀ G(t') dW(t'): the extension by continuity of the elementary integral, which is an isometry from elementary integrands (norm ⟨∫₀ᵀ G² dt⟩^{1/2}) into L²(P), to their closure, which by Gardiner.Ito.exists_elem_approx contains every Itô integrand. Concretely it is the conditional expectation onto 𝓕 T of the mean-square limit of the elementary integrals of a fixed approximating sequence, so that it is exactly (not only a.e.) 𝓕 T-measurable.

def Gardiner.Ito.approxSeq {Ω : Type u_1} [mΩ : MeasurableSpace Ω] (𝓕 : MeasureTheory.Filtration ℝ mΩ) (P : MeasureTheory.Measure Ω) (T : ℝ) (G : ℝ → Ω → ℝ) (k : ℕ) :
(n : ℕ) × ElemIntegrand 𝓕 P (Fin n)

A fixed sequence of elementary integrands approximating G on [0, T] to within 1/(k+1) (in squared norm).

Equations
Instances For
    theorem Gardiner.Ito.approxSeq_spec {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} [hP : MeasureTheory.IsFiniteMeasure P] {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (k : ℕ) :
    (approxSeq 𝓕 P T G k).snd.SupportedIn T ∧ ∫ (p : ℝ × Ω), (Function.uncurry (approxSeq 𝓕 P T G k).snd.process p - Function.uncurry G p) ^ 2 ∂tsMeasure P T < 1 / (↑k + 1)
    def Gardiner.Ito.itoInt {Ω : Type u_1} [mΩ : MeasurableSpace Ω] (W : ℝ → Ω → ℝ) (𝓕 : MeasureTheory.Filtration ℝ mΩ) (P : MeasureTheory.Measure Ω) (T : ℝ) (G : ℝ → Ω → ℝ) :
    Ω → ℝ

    The Itô integral ∫₀ᵀ G(t') dW(t') of an Itô integrand (Gardiner (4.2.8)–(4.2.10), §4.2.6).

    Equations
    Instances For
      theorem Gardiner.Ito.ElemIntegrand.integral_sm {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {ι : Type u_2} [Fintype ι] {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) (H : ElemIntegrand 𝓕 P ι) {T : ℝ} (hH : H.SupportedIn T) :

      The elementary integral of an integrand supported in (0, T] is 𝓕 T-measurable.

      theorem Gardiner.Ito.tendsto_of_lt_one_div {f : ℕ → ℝ} (h0 : ∀ (k : ℕ), 0 ≤ f k) (h : ∀ (k : ℕ), f k < 1 / (↑k + 1)) :
      theorem Gardiner.Ito.ae_eq_zero_of_integral_sq {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {P : MeasureTheory.Measure Ω} {f : Ω → ℝ} (hf : MeasureTheory.MemLp f 2 P) (h : ∫ (ω : Ω), f ω ^ 2 ∂P = 0) :
      f =ᵐ[P] 0
      theorem Gardiner.Ito.l2_tendsto_add {α : Type u_2} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {X Y : ℕ → α → ℝ} {X' Y' : α → ℝ} (hX : ∀ (n : ℕ), MeasureTheory.MemLp (X n) 2 μ) (hY : ∀ (n : ℕ), MeasureTheory.MemLp (Y n) 2 μ) (hX' : MeasureTheory.MemLp X' 2 μ) (hY' : MeasureTheory.MemLp Y' 2 μ) (h1 : Filter.Tendsto (fun (n : ℕ) => ∫ (a : α), (X n a - X' a) ^ 2 ∂μ) Filter.atTop (nhds 0)) (h2 : Filter.Tendsto (fun (n : ℕ) => ∫ (a : α), (Y n a - Y' a) ^ 2 ∂μ) Filter.atTop (nhds 0)) :
      Filter.Tendsto (fun (n : ℕ) => ∫ (a : α), (X n a + Y n a - (X' a + Y' a)) ^ 2 ∂μ) Filter.atTop (nhds 0)

      Mean-square convergence of sums.

      theorem Gardiner.Ito.l2_tendsto_const_mul {α : Type u_2} [MeasurableSpace α] {μ : MeasureTheory.Measure α} {X : ℕ → α → ℝ} {X' : α → ℝ} (c : ℝ) (h1 : Filter.Tendsto (fun (n : ℕ) => ∫ (a : α), (X n a - X' a) ^ 2 ∂μ) Filter.atTop (nhds 0)) :
      Filter.Tendsto (fun (n : ℕ) => ∫ (a : α), (c * X n a - c * X' a) ^ 2 ∂μ) Filter.atTop (nhds 0)
      theorem Gardiner.Ito.itoInt_spec {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      MeasureTheory.MemLp (itoInt W 𝓕 P T G) 2 P ∧ MeasureTheory.StronglyMeasurable (itoInt W 𝓕 P T G) ∧ ∀ (ι : Type) [inst : Fintype ι] (H : ElemIntegrand 𝓕 P ι), H.SupportedIn T → ∫ (ω : Ω), (ElemIntegrand.integral W H ω - itoInt W 𝓕 P T G ω) ^ 2 ∂P = ∫ (p : ℝ × Ω), (Function.uncurry H.process p - Function.uncurry G p) ^ 2 ∂tsMeasure P T

      The approximating elementary integrals converge in mean square to itoInt, which is 𝓕 T-measurable, square integrable, and satisfies the isometry against every elementary integrand.

      theorem Gardiner.Ito.memLp_itoInt {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      MeasureTheory.MemLp (itoInt W 𝓕 P T G) 2 P
      theorem Gardiner.Ito.itoInt_stronglyMeasurable {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :

      ∫₀ᵀ G dW is 𝓕 T-measurable.

      theorem Gardiner.Ito.itoInt_sub_sq {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) {ι : Type} [Fintype ι] (H : ElemIntegrand 𝓕 P ι) (hH : H.SupportedIn T) :
      ∫ (ω : Ω), (ElemIntegrand.integral W H ω - itoInt W 𝓕 P T G ω) ^ 2 ∂P = ∫ (p : ℝ × Ω), (Function.uncurry H.process p - Function.uncurry G p) ^ 2 ∂tsMeasure P T

      The isometry against elementary integrands: ⟨(∫ H dW - ∫ G dW)²⟩ = ⟨∫₀ᵀ (H - G)² dt⟩.

      theorem Gardiner.Ito.itoInt_tendsto {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) {ι : ℕ → Type} [(k : ℕ) → Fintype (ι k)] (H : (k : ℕ) → ElemIntegrand 𝓕 P (ι k)) (hH : ∀ (k : ℕ), (H k).SupportedIn T) (hconv : Filter.Tendsto (fun (k : ℕ) => ∫ (p : ℝ × Ω), (Function.uncurry (H k).process p - Function.uncurry G p) ^ 2 ∂tsMeasure P T) Filter.atTop (nhds 0)) :
      Filter.Tendsto (fun (k : ℕ) => ∫ (ω : Ω), (ElemIntegrand.integral W (H k) ω - itoInt W 𝓕 P T G ω) ^ 2 ∂P) Filter.atTop (nhds 0)

      Any sequence of elementary integrands converging to G gives elementary integrals converging to ∫ G dW in mean square.

      theorem Gardiner.Ito.itoInt_elem {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {ι : Type} [Fintype ι] (H : ElemIntegrand 𝓕 P ι) (hH : H.SupportedIn T) :

      The Itô integral of an elementary integrand is its elementary integral.

      theorem Gardiner.Ito.approxSeq_tendsto {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      Filter.Tendsto (fun (k : ℕ) => ∫ (ω : Ω), (ElemIntegrand.integral W (approxSeq 𝓕 P T G k).snd ω - itoInt W 𝓕 P T G ω) ^ 2 ∂P) Filter.atTop (nhds 0)

      The approximating sequence converges to the Itô integral in mean square.

      theorem Gardiner.Ito.approxSeq_tendsto_ts {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      theorem Gardiner.Ito.itoInt_mean {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      ∫ (ω : Ω), itoInt W 𝓕 P T G ω ∂P = 0

      (4.2.40): the mean value of the Itô integral vanishes, ⟨∫₀ᵀ G dW⟩ = 0.

      theorem Gardiner.Ito.itoInt_mul {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G K : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (hK : IsItoIntegrand 𝓕 P T K) :
      ∫ (ω : Ω), itoInt W 𝓕 P T G ω * itoInt W 𝓕 P T K ω ∂P = ∫ (p : ℝ × Ω), Function.uncurry G p * Function.uncurry K p ∂tsMeasure P T

      (4.2.42): the correlation formula, time–space form: ⟨∫₀ᵀ G dW ∫₀ᵀ H dW⟩ = ∫ G H d(tsMeasure P T).

      theorem Gardiner.Ito.itoInt_mul_iter {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G K : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (hK : IsItoIntegrand 𝓕 P T K) :
      ∫ (ω : Ω), itoInt W 𝓕 P T G ω * itoInt W 𝓕 P T K ω ∂P = ∫ (t : ℝ) in Set.Ioc 0 T, ∫ (ω : Ω), G t ω * K t ω ∂P

      (4.2.42): the correlation formula, ⟨∫₀ᵀ G dW ∫₀ᵀ H dW⟩ = ∫₀ᵀ dt ⟨G(t) H(t)⟩.

      theorem Gardiner.Ito.itoInt_sq {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) :
      ∫ (ω : Ω), itoInt W 𝓕 P T G ω ^ 2 ∂P = ∫ (t : ℝ) in Set.Ioc 0 T, ∫ (ω : Ω), G t ω ^ 2 ∂P

      The Itô isometry, ⟨(∫₀ᵀ G dW)²⟩ = ∫₀ᵀ dt ⟨G(t)²⟩.

      Linearity #

      theorem Gardiner.Ito.itoInt_add {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G K : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (hK : IsItoIntegrand 𝓕 P T K) :
      (itoInt W 𝓕 P T fun (t : ℝ) (ω : Ω) => G t ω + K t ω) =ᵐ[P] fun (ω : Ω) => itoInt W 𝓕 P T G ω + itoInt W 𝓕 P T K ω

      Linearity: ∫ (G + K) dW = ∫ G dW + ∫ K dW a.s.

      theorem Gardiner.Ito.itoInt_const_mul {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (c : ℝ) :
      (itoInt W 𝓕 P T fun (t : ℝ) (ω : Ω) => c * G t ω) =ᵐ[P] fun (ω : Ω) => c * itoInt W 𝓕 P T G ω

      Linearity: ∫ c G dW = c ∫ G dW a.s.

      theorem Gardiner.Ito.itoInt_sub {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G K : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (hK : IsItoIntegrand 𝓕 P T K) :
      (itoInt W 𝓕 P T fun (t : ℝ) (ω : Ω) => G t ω - K t ω) =ᵐ[P] fun (ω : Ω) => itoInt W 𝓕 P T G ω - itoInt W 𝓕 P T K ω

      Linearity: ∫ (G - K) dW = ∫ G dW - ∫ K dW a.s.

      theorem Gardiner.Ito.itoInt_congr {Ω : Type u_1} [mΩ : MeasurableSpace Ω] {𝓕 : MeasureTheory.Filtration ℝ mΩ} {P : MeasureTheory.Measure Ω} {W : ℝ → Ω → ℝ} (hW : IsFWiener W 𝓕 P) {T : ℝ} {G K : ℝ → Ω → ℝ} (hG : IsItoIntegrand 𝓕 P T G) (hK : IsItoIntegrand 𝓕 P T K) (h : Function.uncurry G =ᵐ[tsMeasure P T] Function.uncurry K) :
      itoInt W 𝓕 P T G =ᵐ[P] itoInt W 𝓕 P T K

      Integrands that agree almost everywhere on (0, T] × Ω have the same Itô integral.