AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation
7 named declarations scanned from AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean.
Declarations
def AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.dualAction Partial Not mapped
- The dual action of a semigroup on a continuous linear functional: `Pₜ* ℓ = ℓ ∘ Pₜ`.
def dualAction
(S : ContinuousLinearSemigroup M) (t : ℝ≥0)
(ell : M →L[ℝ] ℝ) : M →L[ℝ] ℝ :=
ell.comp (S.op t)
@[simp]
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:36published source at 0e31a3cda412
theorem AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.dualAction_apply Partial Not mapped
No declaration docstring.
theorem dualAction_apply
(S : ContinuousLinearSemigroup M) (t : ℝ≥0)
(ell : M →L[ℝ] ℝ) (f : M) :
dualAction S t ell f = ell (S.op t f) :=
rfl
/-- The dual action starts at the identity. -/
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:42published source at 0e31a3cda412
theorem AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.dualAction_zero Partial Not mapped
- The dual action starts at the identity.
theorem dualAction_zero
(S : ContinuousLinearSemigroup M) (ell : M →L[ℝ] ℝ) :
dualAction S 0 ell = ell := by
ext f
rw [dualAction_apply, S.op_zero]
rfl
/-- Dual Chapman--Kolmogorov law. The order is reversed by composition, as
expected for the adjoint action. -/
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:49published source at 0e31a3cda412
theorem AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.dualAction_add Partial Not mapped
- Dual Chapman--Kolmogorov law. The order is reversed by composition, as expected for the adjoint action.
theorem dualAction_add
(S : ContinuousLinearSemigroup M) (ell : M →L[ℝ] ℝ)
(s t : ℝ≥0) :
dualAction S (s + t) ell = dualAction S t (dualAction S s ell) := by
ext f
simp only [dualAction_apply]
rw [S.op_add_apply]
/-- Right difference quotient of the weak pairing
`⟨f, Pₜ* ell⟩ = ell (Pₜ f)`. -/
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:58published source at 0e31a3cda412
def AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.rightDualPairingDifferenceQuotient Partial Not mapped
- Right difference quotient of the weak pairing `⟨f, Pₜ* ell⟩ = ell (Pₜ f)`.
def rightDualPairingDifferenceQuotient
(S : ContinuousLinearSemigroup M) (ell : M →L[ℝ] ℝ)
(t h : ℝ≥0) (f : M) : ℝ :=
((h : ℝ)⁻¹) •
(dualAction S (t + h) ell f - dualAction S t ell f)
/-- The weak dual quotient is obtained by applying the functional to the
ordinary semigroup orbit quotient. -/
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:68published source at 0e31a3cda412
theorem AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.rightDualPairingDifferenceQuotient_eq Partial Not mapped
- The weak dual quotient is obtained by applying the functional to the ordinary semigroup orbit quotient.
theorem rightDualPairingDifferenceQuotient_eq
(S : ContinuousLinearSemigroup M) (ell : M →L[ℝ] ℝ)
(t h : ℝ≥0) (f : M) :
rightDualPairingDifferenceQuotient S ell t h f =
ell (rightOrbitDifferenceQuotient S t h f) := by
simp only [rightDualPairingDifferenceQuotient, dualAction_apply,
rightOrbitDifferenceQuotient]
rw [map_smul, map_sub]
/-- Weak right-hand Kolmogorov forward equation.
For an observable in the canonical generator domain, differentiating the dual
pairing gives the evolved functional applied to the generator:
`d⁺/dt (Pₜ* ell)(f) = (Pₜ* ell)(A f)`.
This is the rigorous operator-level content of the calculation immediately
preceding Chewi Proposition 1.2.6. A measure-valued identity involving `A*`
requires an additional adjoint-domain realization and is not claimed here. -/
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:76published source at 0e31a3cda412
theorem AutoSamplingTheory.TechnicalLemmas.StochasticProcesses.WeakForwardEquation.kolmogorov_forward_weak_right Partial Not mapped
- Weak right-hand Kolmogorov forward equation. For an observable in the canonical generator domain, differentiating the dual pairing gives the evolved functional applied to the generator: `d⁺/dt (Pₜ* ell)(f) = (Pₜ* ell)(A f)`. This is the rigorous operator-level content of the calculation immediately preceding Chewi Proposition 1.2.6. A measure-valued identity involving `A*` requires an additional adjoint-domain realization and is not claimed here.
theorem kolmogorov_forward_weak_right
(S : ContinuousLinearSemigroup M) (ell : M →L[ℝ] ℝ)
(f : generatorDomainSubmodule S) (t : ℝ≥0) :
Tendsto
(fun h : ℝ≥0 =>
rightDualPairingDifferenceQuotient S ell t h (f : M))
(nhdsWithin 0 (Ioi 0))
(𝓝 (dualAction S t ell (rightGenerator S f))) := by
have hell :
Tendsto ell
(𝓝 (S.op t (rightGenerator S f)))
(𝓝 (ell (S.op t (rightGenerator S f)))) :=
ell.continuous.continuousAt
have hmap :
Tendsto
(fun h : ℝ≥0 =>
ell (rightOrbitDifferenceQuotient S t h (f : M)))
(nhdsWithin 0 (Ioi 0))
(𝓝 (ell (S.op t (rightGenerator S f)))) :=
hell.comp (kolmogorov_backward_right_generator S f t)
simpa only [rightDualPairingDifferenceQuotient_eq, dualAction_apply] using hmap
end
end WeakForwardEquation
end StochasticProcesses
end TechnicalLemmas
end AutoSamplingTheory
AutoSamplingTheory/TechnicalLemmas/StochasticProcesses/WeakForwardEquation.lean:95published source at 0e31a3cda412