Documentation

SHSLib.Gardiner.Ito.WienerIntegral

The Wiener integral of deterministic functions (Gardiner §4.2.1, §4.2.6) #

For f : ℝ → ℝ and 0 ≤ a ≤ b, Gardiner.Ito.wienerInt W P f a b = ∫ₐᵇ f(s) dW(s) is the mean-square limit of the left-point Riemann sums Σᵢ f(tᵢ)(W(tᵢ₊₁) - W(tᵢ)) on the uniform partitions of [a, b] (Gardiner.Ito.riemannSum, (4.2.8) with τᵢ = tᵢ₋₁). Each Riemann sum is the simple integral of a deterministic simple integrand (Gardiner.Ito.riemannSum_eq). For continuous f:

For step functions the same definition gives the simple integral (SHSLib.Gardiner.Ito.WienerStep). Instances for stdWiener: SHSLib.Gardiner.Ito.WienerInstances.

Cells of a partition #

theorem Gardiner.Ito.exists_cell {p : ℕ → ℝ} {N : ℕ} {s : ℝ} (hs : s ∈ Set.Ioc (p 0) (p N)) :
∃ i < N, s ∈ Set.Ioc (p i) (p (i + 1))
theorem Gardiner.Ito.sum_ind_eq {p : ℕ → ℝ} (hp : Monotone p) {N i₀ : ℕ} (hi₀ : i₀ < N) {s : ℝ} (hs : s ∈ Set.Ioc (p i₀) (p (i₀ + 1))) (c : ℕ → ℝ) :
∑ i ∈ Finset.range N, c i * ind (p i) (p (i + 1)) s = c i₀
theorem Gardiner.Ito.sum_ind_eq_zero {p : ℕ → ℝ} (hp : Monotone p) {N : ℕ} {s : ℝ} (hs : s ∉ Set.Ioc (p 0) (p N)) (c : ℕ → ℝ) :
∑ i ∈ Finset.range N, c i * ind (p i) (p (i + 1)) s = 0

Uniform partitions #

@[reducible, inline]
abbrev Gardiner.Ito.grid (a b : ℝ) (n : ℕ) :
ℕ → ℝ

The uniform partition of [a, b] into n + 1 cells.

Equations
Instances For
    theorem Gardiner.Ito.grid_mono {a b : ℝ} (hab : a ≤ b) (n : ℕ) :
    Monotone (grid a b n)
    theorem Gardiner.Ito.grid_zero (a b : ℝ) (n : ℕ) :
    grid a b n 0 = a
    theorem Gardiner.Ito.grid_last' {a b : ℝ} (n : ℕ) :
    grid a b n (n + 1) = b
    theorem Gardiner.Ito.grid_step (a b : ℝ) (n i : ℕ) :
    grid a b n (i + 1) - grid a b n i = (b - a) / (↑n + 1)
    theorem Gardiner.Ito.grid_mem {a b : ℝ} (hab : a ≤ b) {n i : ℕ} (hi : i ≤ n + 1) :
    grid a b n i ∈ Set.Icc a b
    theorem Gardiner.Ito.cells_osc {g : ℝ → ℝ} {a b : ℝ} (hab : a ≤ b) (hg : ContinuousOn g (Set.Icc a b)) {ε : ℝ} (hε : 0 < ε) :
    ∃ (N : ℕ), ∀ n ≥ N, ∀ i < n + 1, ∀ s ∈ Set.Icc (grid a b n i) (grid a b n (i + 1)), ∀ s' ∈ Set.Icc (grid a b n i) (grid a b n (i + 1)), |g s - g s'| ≤ ε

    Uniform continuity on the cells: for g continuous on [a, b] and n large, g oscillates by at most ε on every cell of the uniform partition.

    Step approximations and Riemann sums #

    def Gardiner.Ito.stepFn (f : ℝ → ℝ) (a b : ℝ) (n : ℕ) (s : ℝ) :

    The left-point step approximation of f on the uniform partition of [a, b], fₙ(s) = Σᵢ f(tᵢ) 1_{(tᵢ, tᵢ₊₁]}(s).

    Equations
    Instances For
      def Gardiner.Ito.riemannSum {Ω : Type u_1} (W : ℝ → Ω → ℝ) (f : ℝ → ℝ) (a b : ℝ) (n : ℕ) (ω : Ω) :

      The left-point Riemann sum Σᵢ f(tᵢ)(W(tᵢ₊₁) - W(tᵢ)) on the uniform partition of [a, b] into n + 1 cells (Gardiner (4.2.8) with τᵢ = tᵢ₋₁).

      Equations
      Instances For
        theorem Gardiner.Ito.ind_mem {a b s : ℝ} (hs : s ∈ Set.Ioc a b) :
        ind a b s = 1
        theorem Gardiner.Ito.ind_notMem {a b s : ℝ} (hs : s ∉ Set.Ioc a b) :
        ind a b s = 0
        theorem Gardiner.Ito.ind_nonneg (a b s : ℝ) :
        0 ≤ ind a b s
        theorem Gardiner.Ito.ind_le_one (a b s : ℝ) :
        ind a b s ≤ 1
        theorem Gardiner.Ito.integral_ind {a b : ℝ} (hab : a ≤ b) :
        ∫ (s : ℝ), ind a b s = b - a
        theorem Gardiner.Ito.stepFn_approx {f : ℝ → ℝ} (hf : Continuous f) {a b : ℝ} (hab : a ≤ b) {ε : ℝ} (hε : 0 < ε) :
        ∃ (N : ℕ), ∀ n ≥ N, ∀ (s : ℝ), |stepFn f a b n s - f s * ind a b s| ≤ ε * ind a b s

        The step approximation converges uniformly: for continuous f and large n, |fₙ(s) - f(s) 1_{(a,b]}(s)| ≤ ε 1_{(a,b]}(s) for every s.

        theorem Gardiner.Ito.measurable_stepFn (f : ℝ → ℝ) (a b : ℝ) (n : ℕ) :
        Measurable (stepFn f a b n)
        theorem Gardiner.Ito.integrable_stepFn_mul (f g : ℝ → ℝ) (a b c d : ℝ) (n m : ℕ) :
        MeasureTheory.Integrable (fun (s : ℝ) => stepFn f a b n s * stepFn g c d m s) MeasureTheory.volume
        theorem Gardiner.Ito.memLp_cut {f : ℝ → ℝ} (hf : Continuous f) (a b : ℝ) :
        MeasureTheory.MemLp (fun (s : ℝ) => f s * ind a b s) 2 MeasureTheory.volume

        f 1_{(a,b]} is square integrable for continuous f.

        theorem Gardiner.Ito.stepFn_ms {f : ℝ → ℝ} (hf : Continuous f) {a b : ℝ} (hab : a ≤ b) :
        Filter.Tendsto (fun (n : ℕ) => ∫ (s : ℝ), (stepFn f a b n s - f s * ind a b s) ^ 2) Filter.atTop (nhds 0)

        fₙ → f 1_{(a,b]} in L²(ds).

        Riemann sums as simple integrals #

        def Gardiner.Ito.detSimple {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) (f : ℝ → ℝ) {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n : ℕ) :
        SimpleIntegrand W ⋯ P (Fin (n + 1))

        The deterministic simple integrand Σᵢ f(tᵢ) 1_{(tᵢ, tᵢ₊₁]} on the uniform partition.

        Equations
        • One or more equations did not get rendered due to their size.
        Instances For
          theorem Gardiner.Ito.riemannSum_eq {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n : ℕ) :
          riemannSum W f a b n = (detSimple hW f ha hab n).integral
          theorem Gardiner.Ito.process_detSimple {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n : ℕ) (s : ℝ) (ω : Ω) :
          (detSimple hW f ha hab n).process s ω = stepFn f a b n s
          theorem Gardiner.Ito.memLp_riemannSum {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n : ℕ) :
          theorem Gardiner.Ito.riemannSum_mul {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f g : ℝ → ℝ} {a b c d : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (hc : 0 ≤ c) (hcd : c ≤ d) (n m : ℕ) :
          ∫ (ω : Ω), riemannSum W f a b n ω * riemannSum W g c d m ω ∂P = ∫ (s : ℝ), stepFn f a b n s * stepFn g c d m s

          The isometry for Riemann sums of deterministic functions: ⟨Rₙ(f; a, b) Rₘ(g; c, d)⟩ = ∫ fₙ gₘ ds.

          theorem Gardiner.Ito.riemannSum_sub_sq {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n m : ℕ) :
          ∫ (ω : Ω), (riemannSum W f a b n ω - riemannSum W f a b m ω) ^ 2 ∂P = ∫ (s : ℝ), (stepFn f a b n s - stepFn f a b m s) ^ 2
          theorem Gardiner.Ito.riemannSum_cauchy {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (ha : 0 ≤ a) (hab : a ≤ b) (ε : ℝ) :
          ε > 0 → ∃ (N : ℕ), ∀ m ≥ N, ∀ n ≥ N, ∫ (ω : Ω), (riemannSum W f a b m ω - riemannSum W f a b n ω) ^ 2 ∂P < ε

          The Riemann sums of a continuous function form a mean-square Cauchy sequence.

          theorem Gardiner.Ito.exists_gauss_sum {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {p : ℕ → ℝ} (hp : Monotone p) (h0 : 0 ≤ p 0) (c : ℕ → ℝ) (N : ℕ) :
          ∃ (v : NNReal), MeasureTheory.Measure.map (fun (ω : Ω) => ∑ i ∈ Finset.range N, c i * dW W p i ω) P = ProbabilityTheory.gaussianReal 0 v

          A linear combination of Wiener increments over a partition is centred Gaussian.

          theorem Gardiner.Ito.exists_gauss_riemannSum {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) (n : ℕ) :

          The Wiener integral #

          def Gardiner.Ito.wienerInt {Ω : Type u_1} [MeasurableSpace Ω] (W : ℝ → Ω → ℝ) (P : MeasureTheory.Measure Ω) (f : ℝ → ℝ) (a b : ℝ) :
          Ω → ℝ

          The Wiener integral ∫ₐᵇ f(s) dW(s) of a deterministic function: the mean-square limit of the left-point Riemann sums on the uniform partitions of [a, b] (Gardiner (4.2.8), (4.2.10)). For continuous f and 0 ≤ a ≤ b the limit exists (Gardiner.Ito.IsWiener.wienerInt_spec).

          Equations
          Instances For
            theorem Gardiner.Ito.cut_mul_cut_eq_indicator (f g : ℝ → ℝ) (a b d s : ℝ) :
            f s * ind a b s * (g s * ind a d s) = (Set.Ioc a (min b d)).indicator (fun (s : ℝ) => f s * g s) s
            theorem Gardiner.Ito.integral_cut_mul_cut {f g : ℝ → ℝ} {a b d : ℝ} (hb : a ≤ b) (hd : a ≤ d) :
            ∫ (s : ℝ), f s * ind a b s * (g s * ind a d s) = ∫ (s : ℝ) in a..min b d, f s * g s
            theorem Gardiner.Ito.IsWiener.wienerInt_spec {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (ha : 0 ≤ a) (hab : a ≤ b) :
            MeasureTheory.MemLp (wienerInt W P f a b) 2 P ∧ Filter.Tendsto (fun (n : ℕ) => ∫ (ω : Ω), (riemannSum W f a b n ω - wienerInt W P f a b ω) ^ 2 ∂P) Filter.atTop (nhds 0)

            The Wiener integral of a continuous function exists: it is square integrable and the Riemann sums converge to it in mean square.

            theorem Gardiner.Ito.IsWiener.wienerInt_mean {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (ha : 0 ≤ a) (hab : a ≤ b) :
            ∫ (ω : Ω), wienerInt W P f a b ω ∂P = 0

            The mean of the Wiener integral vanishes (4.2.40).

            theorem Gardiner.Ito.IsWiener.wienerInt_mul {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f g : ℝ → ℝ} {a b c d : ℝ} (hf : Continuous f) (hg : Continuous g) (ha : 0 ≤ a) (hab : a ≤ b) (hc : 0 ≤ c) (hcd : c ≤ d) :
            ∫ (ω : Ω), wienerInt W P f a b ω * wienerInt W P g c d ω ∂P = ∫ (s : ℝ), f s * ind a b s * (g s * ind c d s)

            The covariance of two Wiener integrals (Gardiner §4.4.1, via (4.2.42)): ⟨∫ₐᵇ f dW ∫_c^d g dW⟩ = ∫ f 1_{(a,b]} g 1_{(c,d]} ds.

            theorem Gardiner.Ito.IsWiener.wienerInt_mul_same {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f g : ℝ → ℝ} {a b d : ℝ} (hf : Continuous f) (hg : Continuous g) (ha : 0 ≤ a) (hab : a ≤ b) (had : a ≤ d) :
            ∫ (ω : Ω), wienerInt W P f a b ω * wienerInt W P g a d ω ∂P = ∫ (s : ℝ) in a..min b d, f s * g s

            The covariance for a common left endpoint: ⟨∫ₐᵇ f dW ∫ₐᵈ g dW⟩ = ∫ₐ^{min(b,d)} f g ds.

            theorem Gardiner.Ito.IsWiener.wienerInt_sq {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (ha : 0 ≤ a) (hab : a ≤ b) :
            ∫ (ω : Ω), wienerInt W P f a b ω ^ 2 ∂P = ∫ (s : ℝ) in a..b, f s ^ 2

            The Itô isometry for the Wiener integral: ⟨(∫ₐᵇ f dW)²⟩ = ∫ₐᵇ f² ds.

            theorem Gardiner.Ito.IsWiener.measurable_riemannSum {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) (f : ℝ → ℝ) (a b : ℝ) (n : ℕ) :
            theorem Gardiner.Ito.IsWiener.wienerInt_map {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (ha : 0 ≤ a) (hab : a ≤ b) :

            The Wiener integral is Gaussian: ∫ₐᵇ f dW ~ N(0, ∫ₐᵇ f² ds).

            theorem Gardiner.Ito.IsWiener.wienerInt_lin_sq {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {ι : Type u_2} [Fintype ι] (c : ι → ℝ) (F : ι → ℝ → ℝ) (A B : ι → ℝ) (hF : ∀ (i : ι), Continuous (F i)) (hA : ∀ (i : ι), 0 ≤ A i) (hAB : ∀ (i : ι), A i ≤ B i) :
            ∫ (ω : Ω), (∑ i : ι, c i * wienerInt W P (F i) (A i) (B i) ω) ^ 2 ∂P = ∫ (s : ℝ), (∑ i : ι, c i * (F i s * ind (A i) (B i) s)) ^ 2

            The mean square of a linear combination of Wiener integrals is the L²(ds) norm of the corresponding combination of integrands.

            theorem Gardiner.Ito.IsWiener.wienerInt_lin_ae {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {ι : Type u_2} [Fintype ι] (c : ι → ℝ) (F : ι → ℝ → ℝ) (A B : ι → ℝ) (hF : ∀ (i : ι), Continuous (F i)) (hA : ∀ (i : ι), 0 ≤ A i) (hAB : ∀ (i : ι), A i ≤ B i) (h : ∀ (s : ℝ), ∑ i : ι, c i * (F i s * ind (A i) (B i) s) = 0) :
            (fun (ω : Ω) => ∑ i : ι, c i * wienerInt W P (F i) (A i) (B i) ω) =ᵐ[P] 0

            Linear relations transfer from integrands to Wiener integrals: if Σᵢ cᵢ Fᵢ 1_{(Aᵢ, Bᵢ]} = 0 pointwise, then Σᵢ cᵢ ∫_{Aᵢ}^{Bᵢ} Fᵢ dW = 0 almost surely.

            theorem Gardiner.Ito.IsWiener.wienerInt_add_smul {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f g : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (hg : Continuous g) (ha : 0 ≤ a) (hab : a ≤ b) (α β : ℝ) :
            wienerInt W P (fun (s : ℝ) => α * f s + β * g s) a b =ᵐ[P] fun (ω : Ω) => α * wienerInt W P f a b ω + β * wienerInt W P g a b ω

            Linearity of the Wiener integral: ∫ (αf + βg) dW = α ∫ f dW + β ∫ g dW a.s.

            theorem Gardiner.Ito.IsWiener.wienerInt_one {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {a b : ℝ} (ha : 0 ≤ a) (hab : a ≤ b) :
            wienerInt W P (fun (x : ℝ) => 1) a b =ᵐ[P] fun (ω : Ω) => W b ω - W a ω

            The Wiener integral of the constant 1 is the increment: ∫ₐᵇ dW = W(b) - W(a).

            theorem Gardiner.Ito.hasGaussianLaw_pair {Ω : Type u_1} [MeasurableSpace Ω] {P : MeasureTheory.Measure Ω} {X Y : Ω → ℝ} (hX : AEMeasurable X P) (hY : AEMeasurable Y P) (h : ∀ (α β : ℝ), ∃ (m : ℝ) (v : NNReal), MeasureTheory.Measure.map (fun (ω : Ω) => α * X ω + β * Y ω) P = ProbabilityTheory.gaussianReal m v) :
            ProbabilityTheory.HasGaussianLaw (fun (ω : Ω) => (X ω, Y ω)) P

            A pair of random variables is jointly Gaussian as soon as every linear combination is Gaussian.

            theorem Gardiner.Ito.IsWiener.wienerInt_hasGaussianLaw_pair {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f g : ℝ → ℝ} {a b : ℝ} (hf : Continuous f) (hg : Continuous g) (ha : 0 ≤ a) (hab : a ≤ b) :
            ProbabilityTheory.HasGaussianLaw (fun (ω : Ω) => (wienerInt W P f a b ω, wienerInt W P g a b ω)) P

            Joint law: Wiener integrals over a common interval are jointly Gaussian.

            Integration by parts #

            theorem Gardiner.Ito.sum_by_parts (f w : ℕ → ℝ) (N : ℕ) :
            ∑ i ∈ Finset.range N, f i * (w (i + 1) - w i) = f N * w N - f 0 * w 0 - ∑ i ∈ Finset.range N, (f (i + 1) - f i) * w (i + 1)

            Summation by parts: Σᵢ fᵢ(wᵢ₊₁ - wᵢ) = fₙwₙ - f₀w₀ - Σᵢ (fᵢ₊₁ - fᵢ)wᵢ₊₁.

            theorem Gardiner.Ito.riemann_ibp_tendsto {f f' w : ℝ → ℝ} (hf : ∀ (s : ℝ), HasDerivAt f (f' s) s) (hf' : Continuous f') (hw : Continuous w) {a b : ℝ} (hab : a ≤ b) :
            Filter.Tendsto (fun (n : ℕ) => ∑ i ∈ Finset.range (n + 1), f (grid a b n i) * (w (grid a b n (i + 1)) - w (grid a b n i))) Filter.atTop (nhds (f b * w b - f a * w a - ∫ (s : ℝ) in a..b, f' s * w s))

            Pathwise integration by parts for Riemann sums: for a C¹ function f and a continuous path w, Σᵢ f(tᵢ)(w(tᵢ₊₁) - w(tᵢ)) → f(b)w(b) - f(a)w(a) - ∫ₐᵇ f'(s) w(s) ds.

            theorem Gardiner.Ito.IsWiener.wienerInt_ibp {Ω : Type u_1} [MeasurableSpace Ω] {W : ℝ → Ω → ℝ} {P : MeasureTheory.Measure Ω} (hW : IsWiener W P) {f f' : ℝ → ℝ} {a b : ℝ} (hf : ∀ (s : ℝ), HasDerivAt f (f' s) s) (hf' : Continuous f') (ha : 0 ≤ a) (hab : a ≤ b) :
            wienerInt W P f a b =ᵐ[P] fun (ω : Ω) => f b * W b ω - f a * W a ω - ∫ (s : ℝ) in a..b, f' s * W s ω

            Integration by parts for the Wiener integral of a C¹ function: ∫ₐᵇ f dW = f(b)W(b) - f(a)W(a) - ∫ₐᵇ f'(s) W(s) ds almost surely. This identifies the mean-square integral with an expression in the paths of W.