The basic L² estimates for Itô processes with state-dependent coefficients #
isItoIntegrand_drive: for a jointly measurable adapted processxwith∫_{t₀}^T E x(s)² ds < ∞,s ↦ 1_{s > t₀} b(x(s), s)is an Itô integrand on[0, T];sup_sq_le: ifY(u) = ζ + ∫_{t₀}^u A ds + ∫_{t₀}^u B dWfor everyu ∈ [t₀, T]andYhas continuous paths, thenE sup_{t₀ ≤ u ≤ t} |Y(u)|² ≤ 3 E ζ² + 3 (t - t₀) ∫_{t₀}^t E A² ds + 12 ∫_{t₀}^t E B² ds. The factor12 = 3 · 4comes from Doob's inequality for the Itô integral.
def
Gardiner.Ito.StochDE.tsSq
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
(P : MeasureTheory.Measure Ω)
(A : ℝ → Ω → ℝ)
(t₀ t : ℝ)
:
∫_{(t₀, t]} E[A(s)²] ds as an element of ℝ≥0∞.
Equations
Instances For
def
Gardiner.Ito.StochDE.supSq
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
(P : MeasureTheory.Measure Ω)
(Y : ℝ → Ω → ℝ)
(t₀ t : ℝ)
:
E[sup_{t₀ ≤ u ≤ t} |Y(u)|²] as an element of ℝ≥0∞.
Instances For
theorem
Gardiner.Ito.StochDE.supSq_mono_right
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{P : MeasureTheory.Measure Ω}
{Y : ℝ → Ω → ℝ}
{t₀ s t : ℝ}
(hst : s ≤ t)
:
Pointwise bounds for the coefficients #
The noise integrand #
theorem
Gardiner.Ito.StochDE.memLp_two_of_lintegral_lt_top
{α : Type u_2}
[MeasurableSpace α]
{μ : MeasureTheory.Measure α}
{f : α → ℝ}
(hf : MeasureTheory.AEStronglyMeasurable f μ)
(h : ∫⁻ (a : α), ‖f a‖ₑ ^ 2 ∂μ < ⊤)
:
MeasureTheory.MemLp f 2 μ
theorem
Gardiner.Ito.StochDE.measurable_drive
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{a b : ℝ → ℝ → ℝ}
{t₀ T K : ℝ}
(hc : SDECoeffs a b t₀ T K)
{x : ℝ → Ω → ℝ}
(hxm : Measurable (Function.uncurry x))
:
Measurable (Function.uncurry (drive t₀ b x))
theorem
Gardiner.Ito.StochDE.drive_adapted
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{𝓕 : MeasureTheory.Filtration ℝ mΩ}
{a b : ℝ → ℝ → ℝ}
{t₀ T K : ℝ}
(hc : SDECoeffs a b t₀ T K)
{x : ℝ → Ω → ℝ}
(hxa : ∀ t ∈ Set.Icc t₀ T, MeasureTheory.StronglyMeasurable (x t))
{t : ℝ}
(htT : t ≤ T)
:
MeasureTheory.StronglyMeasurable (drive t₀ b x t)
theorem
Gardiner.Ito.StochDE.lintegral_tsMeasure_drive_le
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsFiniteMeasure P]
{a b : ℝ → ℝ → ℝ}
{t₀ T K : ℝ}
(hc : SDECoeffs a b t₀ T K)
(ht₀ : 0 ≤ t₀)
{x : ℝ → Ω → ℝ}
(hxm : Measurable (Function.uncurry x))
:
∫_{(0,T]} ∫ |1_{s>t₀} b(x(s), s)|² ≤ K² ∫_{(t₀,T]} ∫ (1 + |x(s)|²).
theorem
Gardiner.Ito.StochDE.lintegral_one_add_lt_top
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsFiniteMeasure P]
{x : ℝ → Ω → ℝ}
(hxm : Measurable (Function.uncurry x))
{t₀ T : ℝ}
(hsq : ∫⁻ (s : ℝ) in Set.Ioc t₀ T, ∫⁻ (ω : Ω), ‖x s ω‖ₑ ^ 2 ∂P < ⊤)
:
theorem
Gardiner.Ito.StochDE.isItoIntegrand_drive
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{𝓕 : MeasureTheory.Filtration ℝ mΩ}
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsFiniteMeasure P]
{a b : ℝ → ℝ → ℝ}
{t₀ T K : ℝ}
(hc : SDECoeffs a b t₀ T K)
(ht₀ : 0 ≤ t₀)
{x : ℝ → Ω → ℝ}
(hxm : Measurable (Function.uncurry x))
(hxa : ∀ t ∈ Set.Icc t₀ T, MeasureTheory.StronglyMeasurable (x t))
(hsq : ∫⁻ (s : ℝ) in Set.Ioc t₀ T, ∫⁻ (ω : Ω), ‖x s ω‖ₑ ^ 2 ∂P < ⊤)
:
IsItoIntegrand 𝓕 P T (drive t₀ b x)
The noise integrand of (4.3.4) is an Itô integrand.
The basic estimate #
theorem
Gardiner.Ito.StochDE.ofReal_integral_sq_le_tsSq
{Ω : Type u_1}
[mΩ : MeasurableSpace Ω]
{P : MeasureTheory.Measure Ω}
{B : ℝ → Ω → ℝ}
{t₀ t : ℝ}
(ht₀ : 0 ≤ t₀)
(hB0 : ∀ (s : ℝ) (ω : Ω), s ≤ t₀ → B s ω = 0)
:
∫_{(0,t]} E B² ≤ ∫_{(t₀,t]} E B² for an integrand vanishing on (0, t₀] (as ℝ≥0∞).
theorem
Gardiner.Ito.StochDE.sup_sq_le
{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)
{Y A B : ℝ → Ω → ℝ}
{ζ : Ω → ℝ}
{t₀ T t : ℝ}
(ht₀ : 0 ≤ t₀)
(ht : t ∈ Set.Icc t₀ T)
(hB : IsItoIntegrand 𝓕 P T B)
(hB0 : ∀ (s : ℝ) (ω : Ω), s ≤ t₀ → B s ω = 0)
(hA : Measurable (Function.uncurry A))
(hζ : AEMeasurable ζ P)
(hYc : ∀ᵐ (ω : Ω) ∂P, ContinuousOn (fun (u : ℝ) => Y u ω) (Set.Icc t₀ T))
(hY : ∀ u ∈ Set.Icc t₀ T, Y u =ᵐ[P] fun (ω : Ω) => (ζ ω + ∫ (s : ℝ) in t₀..u, A s ω) + itoInt W 𝓕 P u B ω)
:
The basic L² estimate. If Y(u) = ζ + ∫_{t₀}^u A ds + ∫_{t₀}^u B dW almost surely for
each u ∈ [t₀, T] and Y has almost surely continuous paths, then for t ∈ [t₀, T]
E sup_{t₀ ≤ u ≤ t} |Y(u)|² ≤ 3 E ζ² + 3 (t - t₀) ∫_{t₀}^t E A² + 12 ∫_{t₀}^t E B².