Documentation

LeanPool.AnomalousDiffusion.AVenhance.Infra.Section4.Amnr.CurveJacobian

The actual linearized Volterra operator and its primitive jets.

noncomputable def AVenhance.Infra.Section4.amnrCurveJacobianCoefficient {K E : Type} [TopologicalSpace K] [CompactSpace K] [NormedAddCommGroup E] [NormedSpace ℝ E] (f : ℝ × E → E) (hf : ContDiff ℝ (↑⊤) f) (η : C(K, ℝ)) (V : C(K, E) →L[ℝ] C(K, E)) (u : C(K, E)) :

The linearized Volterra coefficient at an actual continuous trajectory.

Equations
  • One or more equations did not get rendered due to their size.
Instances For

    Lifting an operator curve and composing with time integration is linear with norm at most the integration norm.

    theorem AVenhance.Infra.Section4.amnrCurveJacobianCoefficient_contDiff {K E : Type} [TopologicalSpace K] [CompactSpace K] [NormedAddCommGroup E] [NormedSpace ℝ E] {f : ℝ × E → E} (hf : ContDiff ℝ (↑⊤) f) (hbound : ∀ (j : ℕ), ∃ (C : ℝ), 0 ≤ C ∧ ∀ (z : ℝ × E), ‖iteratedFDeriv ℝ j f z‖ ≤ C) (η : C(K, ℝ)) (V : C(K, E) →L[ℝ] C(K, E)) :

    Primitive joint derivative bounds prove smoothness of the linearized coefficient, including when its state is an actual flow curve.

    theorem AVenhance.Infra.Section4.amnrCurveJacobianCoefficient_iteratedFDeriv_norm_le {K E : Type} [TopologicalSpace K] [CompactSpace K] [NormedAddCommGroup E] [NormedSpace ℝ E] {f : ℝ × E → E} (hf : ContDiff ℝ (↑⊤) f) (hbound : ∀ (j : ℕ), ∃ (C : ℝ), 0 ≤ C ∧ ∀ (z : ℝ × E), ‖iteratedFDeriv ℝ j f z‖ ≤ C) (η : C(K, ℝ)) (V : C(K, E) →L[ℝ] C(K, E)) (n : ℕ) (u : C(K, E)) {B : ℝ} (hB : 0 ≤ B) (hstate : ∀ (t : K) (x : E), ‖iteratedFDeriv ℝ n (fun (y : E) => amnrStateDerivative f (η t, y)) x‖ ≤ B) :

    Every coefficient jet has only the time-integration norm as additional cost, beyond the actual primitive state-Jacobian jet.

    theorem AVenhance.Infra.Section4.amnrCurveResidual_solution_jacobian_equation {K E : Type} [TopologicalSpace K] [CompactSpace K] [NormedAddCommGroup E] [NormedSpace ℝ E] {f : ℝ × E → E} (hf : ContDiff ℝ (↑⊤) f) {C : ℝ} (hC : 0 ≤ C) (hbound : ∀ (z : ℝ × E), ‖iteratedFDeriv ℝ 2 f z‖ ≤ C) (η : C(K, ℝ)) (V : C(K, E) →L[ℝ] C(K, E)) {Q : E → C(K, E)} (hQ : ContDiff ℝ (↑⊤) Q) (heq : ∀ (y : E), amnrCurveResidual f ⋯ η V (Q y) = ContinuousMap.const K y) (x : E) :

    Differentiating the actual Volterra equation yields the Jacobian fixed equation. Its highest derivative is derived, never imposed as a premise.