Documentation

SHSLib.Gardiner.Approx.SystemSize

Van Kampen's system size expansion #

Gardiner, Handbook of Stochastic Methods, §7.2.3, equations (7.2.28)–(7.2.40) and (7.2.59)–(7.2.61).

The master equation has finitely many jumps a → a + r_j with extensive rates W(a; r_j) = Ω ψ_j(a/Ω) (7.2.29), so that the jump moments are αₙ(a) = Ω α̃ₙ(a/Ω) (7.2.31). In the fluctuation variable z of (7.2.30), a = Ωφ(t) + Ω^{1/2} z, the backward generator of the master equation becomes Gardiner.vkGen (Gardiner.finJumpGen_fluct, Gardiner.hasDerivAt_fluct_frame).

Main results:

def Gardiner.extRate {ι : Type u_1} (ψ : ι → ℝ → ℝ) (Ω : ℝ) :
ι → ℝ → ℝ

(7.2.29): the extensive transition rates W(a; r_j) = Ω ψ_j(a/Ω).

Equations
Instances For
    @[reducible, inline]
    abbrev Gardiner.intMoment {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) (n : ℕ) (x : ℝ) :

    (7.2.31): the intensive jump moments α̃ₙ(x) = Σ_j ψ_j(x) r_jⁿ.

    Equations
    Instances For
      theorem Gardiner.jumpMoment_extensive {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) (Ω a : ℝ) (n : ℕ) :
      finJumpMoment s r (extRate ψ Ω) a n = Ω * intMoment s r ψ n (a / Ω)

      (7.2.31): under the scaling (7.2.29), αₙ(a) = Ω α̃ₙ(a/Ω).

      theorem Gardiner.finJumpGen_fluct {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) {Ω : ℝ} (hΩ : 0 < Ω) (phi z : ℝ) (f : ℝ → ℝ) :
      finJumpGen s r (extRate ψ Ω) (fun (a : ℝ) => f ((a - Ω * phi) / √Ω)) (Ω * phi + √Ω * z) = ∑ j ∈ s, Ω * ψ j (phi + (√Ω)⁻¹ * z) * (f (z + (√Ω)⁻¹ * r j) - f z)

      (7.2.30): the master-equation generator acting on a function of the fluctuation variable, g(a) = f((a - Ωφ)/Ω^{1/2}), evaluated at a = Ωφ + Ω^{1/2} z.

      theorem Gardiner.hasDerivAt_fluct_frame {Ω : ℝ} (hΩ : 0 < Ω) (f : ℝ → ℝ) (f' : ℝ) (phi : ℝ → ℝ) (dphi a t : ℝ) (hphi : HasDerivAt phi dphi t) (hf : HasDerivAt f f' ((a - Ω * phi t) / √Ω)) :
      HasDerivAt (fun (t : ℝ) => f ((a - Ω * phi t) / √Ω)) (-(√Ω * dphi) * f') t

      (7.2.32), the transport term: along the moving frame a = Ωφ(t) + Ω^{1/2} z, a function of z changes at rate -Ω^{1/2} φ'(t) ∂_z f.

      def Gardiner.vkGen {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) (Ω phi dphi : ℝ) (D : ℕ → ℝ → ℝ) (z : ℝ) :

      The backward generator of the fluctuation variable z (7.2.30)–(7.2.32): the master-equation generator in the moving frame plus the transport term, Σ_j Ω ψ_j(φ + Ω^{-1/2} z) [f(z + Ω^{-1/2} r_j) - f(z)] - Ω^{1/2} φ' f'(z). The test function is given by its derivative tower D.

      Equations
      Instances For
        def Gardiner.kmGen' {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) (Ω phi : ℝ) (D : ℕ → ℝ → ℝ) (z : ℝ) :

        The Kramers–Moyal route (7.2.36)–(7.2.39): the generator of the Kramers–Moyal Fokker–Planck equation (7.2.37) in the fluctuation variable. It is the small-noise generator snGen of (6.3.8) with A = α̃₁, B = α̃₂, ε = Ω^{-1/2} (7.2.38).

        Equations
        Instances For
          theorem Gardiner.vkGen_sub_kmGen_le {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) {Ω : ℝ} (hΩ : 0 < Ω) (phi z M : ℝ) (D : ℕ → ℝ → ℝ) (hD : ∀ k ≤ 2, ∀ (y : ℝ), HasDerivAt (D k) (D (k + 1) y) y) (hM : ∀ (y : ℝ), |D 3 y| ≤ M) :
          |vkGen s r ψ Ω phi (intMoment s r ψ 1 phi) D z - kmGen' s r ψ Ω phi D z| ≤ (√Ω)⁻¹ * (M / 6 * finAbsJumpMoment s r ψ (phi + (√Ω)⁻¹ * z) 3)

          The master equation and its Kramers–Moyal truncation have the same system-size limit: in the fluctuation variable, with φ' = α̃₁(φ) (7.2.33), the two generators differ by at most Ω^{-1/2} ‖f'''‖ β̃₃(φ + Ω^{-1/2}z)/6.

          theorem Gardiner.kmGen_sub_lnaGen_le {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) {Ω : ℝ} (hΩ : 0 < Ω) (phi z K L : ℝ) (D E : ℕ → ℝ → ℝ) (hE0 : E 0 = intMoment s r ψ 1) (hE : ∀ k ≤ 1, ∀ (u : ℝ), HasDerivAt (E k) (E (k + 1) u) u) (hK : ∀ t ∈ Set.uIcc 0 ((√Ω)⁻¹ * z), |E 2 (phi + t)| ≤ K) (hL : |intMoment s r ψ 2 (phi + (√Ω)⁻¹ * z) - intMoment s r ψ 2 phi| ≤ L * |(√Ω)⁻¹ * z|) :
          |kmGen' s r ψ Ω phi D z - ouGen (E 1 phi) (intMoment s r ψ 2 phi) z (D 1 z) (D 2 z)| ≤ (√Ω)⁻¹ * (K * z ^ 2 * |D 1 z| / 2 + L * |z| * |D 2 z| / 2)

          The Kramers–Moyal route to the linear noise approximation (7.2.37)–(7.2.39): the generator of the Kramers–Moyal equation in the fluctuation variable is within Ω^{-1/2} (K z² |f'| + L |z| |f''|)/2 of the linear-noise operator α̃₁'(φ) z f' + ½ α̃₂(φ) f''. Here E is a derivative tower of α̃₁ with |α̃₁''| ≤ K between φ and φ + Ω^{-1/2} z, and L controls the variation of α̃₂.

          theorem Gardiner.vkGen_sub_lnaGen_le {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) {Ω : ℝ} (hΩ : 0 < Ω) (phi z K L M : ℝ) (D : ℕ → ℝ → ℝ) (hD : ∀ k ≤ 2, ∀ (y : ℝ), HasDerivAt (D k) (D (k + 1) y) y) (hM : ∀ (y : ℝ), |D 3 y| ≤ M) (E : ℕ → ℝ → ℝ) (hE0 : E 0 = intMoment s r ψ 1) (hE : ∀ k ≤ 1, ∀ (u : ℝ), HasDerivAt (E k) (E (k + 1) u) u) (hK : ∀ t ∈ Set.uIcc 0 ((√Ω)⁻¹ * z), |E 2 (phi + t)| ≤ K) (hL : |intMoment s r ψ 2 (phi + (√Ω)⁻¹ * z) - intMoment s r ψ 2 phi| ≤ L * |(√Ω)⁻¹ * z|) :
          |vkGen s r ψ Ω phi (intMoment s r ψ 1 phi) D z - ouGen (E 1 phi) (intMoment s r ψ 2 phi) z (D 1 z) (D 2 z)| ≤ (√Ω)⁻¹ * (K * z ^ 2 * |D 1 z| / 2 + L * |z| * |D 2 z| / 2 + M / 6 * finAbsJumpMoment s r ψ (phi + (√Ω)⁻¹ * z) 3)

          Van Kampen's expansion with an explicit error term. On a test function f with |f'''| ≤ M, the generator of the fluctuation variable (7.2.30)–(7.2.32), with φ solving the deterministic equation (7.2.33), differs from the linear-noise operator α̃₁'(φ) z f'(z) + ½ α̃₂(φ) f''(z) (the corrected (7.2.35)) by at most Ω^{-1/2} [K z² |f'(z)|/2 + L |z| |f''(z)|/2 + M β̃₃(φ + Ω^{-1/2} z)/6].

          theorem Gardiner.vkGen_tendsto_lnaGen {ι : Type u_1} (s : Finset ι) (r : ι → ℝ) (ψ : ι → ℝ → ℝ) (phi z K L M : ℝ) (D : ℕ → ℝ → ℝ) (hD : ∀ k ≤ 2, ∀ (y : ℝ), HasDerivAt (D k) (D (k + 1) y) y) (hM : ∀ (y : ℝ), |D 3 y| ≤ M) (E : ℕ → ℝ → ℝ) (hE0 : E 0 = intMoment s r ψ 1) (hE : ∀ k ≤ 1, ∀ (u : ℝ), HasDerivAt (E k) (E (k + 1) u) u) (hK : ∀ (u : ℝ), |E 2 u| ≤ K) (hL : ∀ (x : ℝ), |intMoment s r ψ 2 x - intMoment s r ψ 2 phi| ≤ L * |x - phi|) (hψ : ∀ j ∈ s, ContinuousAt (ψ j) phi) :
          Filter.Tendsto (fun (Ω : ℝ) => vkGen s r ψ Ω phi (intMoment s r ψ 1 phi) D z) Filter.atTop (nhds (ouGen (E 1 phi) (intMoment s r ψ 2 phi) z (D 1 z) (D 2 z)))

          The system-size limit (7.2.35), corrected: as Ω → ∞ the generator of the fluctuation variable converges to the linear-noise operator on every smooth test function with bounded third derivative, provided α̃₁'' is bounded, α̃₂ is Lipschitz at φ, and the rates are continuous at φ.

          (7.2.33)–(7.2.34): the coefficients of the expansion #

          def Gardiner.vkCoeff (Dα : ℕ → ℕ → ℝ) (Df : ℕ → ℝ) (z : ℝ) (m : ℕ) :

          The coefficient of Ω^{-(m-2)/2} in (7.2.34), in backward (generator) form: Σ_{n=1}^{m} α̃ₙ^{(m-n)}(φ) z^{m-n} f⁽ⁿ⁾(z) / (n! (m-n)!), where Dα n k = α̃ₙ^{(k)}(φ) and Df n = f⁽ⁿ⁾(z). (The forward form has (-∂_z)ⁿ z^{m-n} in place of z^{m-n} f⁽ⁿ⁾.)

          Equations
          Instances For
            theorem Gardiner.vkCoeff_one (Dα : ℕ → ℕ → ℝ) (Df : ℕ → ℝ) (z dphi : ℝ) :
            vkCoeff Dα Df z 1 - dphi * Df 1 = 0 ↔ dphi * Df 1 = Dα 1 0 * Df 1

            (7.2.33): the m = 1 coefficient α̃₁(φ) f' is cancelled by the transport term -Ω^{1/2} φ' f' exactly when φ' = α̃₁(φ).

            theorem Gardiner.vkCoeff_two (Dα : ℕ → ℕ → ℝ) (Df : ℕ → ℝ) (z : ℝ) :
            vkCoeff Dα Df z 2 = ouGen (Dα 1 1) (Dα 2 0) z (Df 1) (Df 2)

            (7.2.34)–(7.2.35): the m = 2 coefficient is the linear-noise operator α̃₁'(φ) z f' + ½ α̃₂(φ) f''.

            The corrected linear-noise Fokker–Planck equation (7.2.35) #

            def Gardiner.lnaFP (a₁ a₂ : ℝ) (P : ℝ → ℝ) (z : ℝ) :

            (7.2.35), corrected: the right-hand side -α̃₁'(φ) ∂_z(z P) + ½ α̃₂(φ) ∂_z² P of the linear-noise Fokker–Planck equation, with a₁ = α̃₁'(φ(t)) and a₂ = α̃₂(φ(t)). (The text prints α̃₁[φ(t)] for a₁.) The corresponding SDE (7.2.40) is dz = a₁ z dt + √a₂ dW(t).

            Equations
            Instances For
              theorem Gardiner.lnaFP_eq_hierRHS_zero (At Bt : ℕ → ℝ) (ph : ℕ → ℝ → ℝ) (z : ℝ) :
              lnaFP (At 1) (Bt 0) (ph 0) z = hierRHS At Bt ph 0 z

              (7.2.38)–(7.2.39): the corrected (7.2.35) is the lowest-order equation (6.3.12) of the small-noise hierarchy for the Kramers–Moyal equation (7.2.37) with ε² = 1/Ω, Ã₁ = α̃₁'(φ), B̃₀ = α̃₂(φ).

              theorem Gardiner.integral_mul_lnaFP (a₁ a₂ : ℝ) (D : ℕ → ℝ → ℝ) (P P' P'' : ℝ → ℝ) (hD : ∀ k ≤ 1, ∀ (y : ℝ), HasDerivAt (D k) (D (k + 1) y) y) (hP : ∀ (z : ℝ), HasDerivAt P (P' z) z) (hP' : ∀ (z : ℝ), HasDerivAt P' (P'' z) z) (i1 : MeasureTheory.Integrable (fun (z : ℝ) => D 0 z * (P z + z * P' z)) MeasureTheory.volume) (i2 : MeasureTheory.Integrable (fun (z : ℝ) => D 1 z * (z * P z)) MeasureTheory.volume) (i3 : MeasureTheory.Integrable (fun (z : ℝ) => D 0 z * (z * P z)) MeasureTheory.volume) (i4 : MeasureTheory.Integrable (fun (z : ℝ) => D 0 z * P'' z) MeasureTheory.volume) (i5 : MeasureTheory.Integrable (fun (z : ℝ) => D 1 z * P' z) MeasureTheory.volume) (i6 : MeasureTheory.Integrable (fun (z : ℝ) => D 0 z * P' z) MeasureTheory.volume) (i7 : MeasureTheory.Integrable (fun (z : ℝ) => D 2 z * P z) MeasureTheory.volume) (i8 : MeasureTheory.Integrable (fun (z : ℝ) => D 1 z * P z) MeasureTheory.volume) :
              ∫ (z : ℝ), D 0 z * lnaFP a₁ a₂ P z = ∫ (z : ℝ), ouGen a₁ a₂ z (D 1 z) (D 2 z) * P z

              The linear-noise operator is the generator of the corrected (7.2.35): for a test function f and a density P, ∫ f · lnaFP P = ∫ (a₁ z f' + ½ a₂ f'') P (integration by parts; surface terms vanish by integrability).

              theorem Gardiner.integral_moments_lnaFP (a₁ a₂ : ℝ) (P P' P'' : ℝ → ℝ) (hP : ∀ (z : ℝ), HasDerivAt P (P' z) z) (hP' : ∀ (z : ℝ), HasDerivAt P' (P'' z) z) (hint : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P z) MeasureTheory.volume) (hint' : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P' z) MeasureTheory.volume) (hint'' : ∀ i ≤ 2, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P'' z) MeasureTheory.volume) :
              ∫ (z : ℝ), z * lnaFP a₁ a₂ P z = a₁ * ∫ (z : ℝ), z * P z ∧ ∫ (z : ℝ), z ^ 2 * lnaFP a₁ a₂ P z = (2 * a₁ * ∫ (z : ℝ), z ^ 2 * P z) + a₂ * ∫ (z : ℝ), P z

              The moments of the corrected linear-noise equation: ∫ z · lnaFP P = a₁ ∫ z P and ∫ z² · lnaFP P = 2 a₁ ∫ z² P + a₂ ∫ P (the m = 2 truncation of (7.2.60)).

              theorem Gardiner.lna_mean_ode (a₁ a₂ : ℝ) (P : ℝ → ℝ → ℝ) (P' P'' : ℝ → ℝ) (t : ℝ) (hP : ∀ (z : ℝ), HasDerivAt (P t) (P' z) z) (hP' : ∀ (z : ℝ), HasDerivAt P' (P'' z) z) (hint : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P t z) MeasureTheory.volume) (hint' : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P' z) MeasureTheory.volume) (hint'' : ∀ i ≤ 2, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P'' z) MeasureTheory.volume) (hswap : HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z * P t z) (∫ (z : ℝ), z * lnaFP a₁ a₂ (P t) z) t) :
              HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z * P t z) (a₁ * ∫ (z : ℝ), z * P t z) t

              The mean of the linear-noise approximation (7.2.48), (4.4.84): if P(t, ·) solves the corrected (7.2.35) with coefficient a₁(t) and d/dt commutes with ∫ dz (hswap), then d⟨z⟩/dt = a₁(t) ⟨z⟩.

              theorem Gardiner.lna_second_moment_ode (a₁ a₂ : ℝ) (P : ℝ → ℝ → ℝ) (P' P'' : ℝ → ℝ) (t : ℝ) (hP : ∀ (z : ℝ), HasDerivAt (P t) (P' z) z) (hP' : ∀ (z : ℝ), HasDerivAt P' (P'' z) z) (hint : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P t z) MeasureTheory.volume) (hint' : ∀ i ≤ 3, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P' z) MeasureTheory.volume) (hint'' : ∀ i ≤ 2, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * P'' z) MeasureTheory.volume) (hswap : HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z ^ 2 * P t z) (∫ (z : ℝ), z ^ 2 * lnaFP a₁ a₂ (P t) z) t) :
              HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z ^ 2 * P t z) ((2 * a₁ * ∫ (z : ℝ), z ^ 2 * P t z) + a₂ * ∫ (z : ℝ), P t z) t

              The second moment of the linear-noise approximation (7.2.49), (4.4.85): under the same hypotheses, d⟨z²⟩/dt = 2 a₁(t) ⟨z²⟩ + a₂(t) ∫ P.

              (7.2.36)–(7.2.37): the Kramers–Moyal equation in intensive variables #

              def Gardiner.kmFP (α₁ α₂ q : ℝ → ℝ) (a : ℝ) :

              The right-hand side -∂ₐ[α₁ q] + ½ ∂ₐ²[α₂ q] of the Kramers–Moyal Fokker–Planck equation (7.2.36).

              Equations
              Instances For
                theorem Gardiner.km_change_of_variables {Ω : ℝ} (hΩ : 0 < Ω) (α₁t α₂t P : ℝ → ℝ) (x : ℝ) :
                Ω * kmFP (fun (a : ℝ) => Ω * α₁t (a / Ω)) (fun (a : ℝ) => Ω * α₂t (a / Ω)) P (Ω * x) = snFP α₁t α₂t (√Ω)⁻¹ (fun (x : ℝ) => Ω * P (Ω * x)) x

                (7.2.36) ⟹ (7.2.37): if αₙ(a) = Ω α̃ₙ(a/Ω) (7.2.31), then the density P_x(x) = Ω P_a(Ωx) of x = a/Ω satisfies ∂ₜ P_x = -∂ₓ[α̃₁ P_x] + (1/2Ω) ∂ₓ²[α̃₂ P_x], i.e. the small-noise equation (6.3.1) with ε² = 1/Ω (7.2.38). Pointwise: Ω · RHS(7.2.36)(Ωx) = RHS(6.3.1)(x).

                (7.2.59)–(7.2.61): the moment hierarchy #

                theorem Gardiner.vk_moment_hierarchy (T : Finset (ℕ × ℕ)) (c : ℕ × ℕ → ℝ) (P : ℝ → ℝ → ℝ) (G : ℕ × ℕ → ℕ → ℝ → ℝ) (dP : ℝ → ℝ) (t : ℝ) (k : ℕ) (hG0 : ∀ p ∈ T, ∀ (z : ℝ), G p 0 z = z ^ (p.1 - p.2) * P t z) (hG : ∀ p ∈ T, ∀ j < p.2, ∀ (z : ℝ), HasDerivAt (G p j) (G p (j + 1) z) z) (hint : ∀ p ∈ T, ∀ j ≤ p.2, ∀ i ≤ k, MeasureTheory.Integrable (fun (z : ℝ) => z ^ i * G p j z) MeasureTheory.volume) (hpde : ∀ (z : ℝ), dP z = ∑ p ∈ T, c p * ((-1) ^ p.2 * G p p.2 z)) (hswap : HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z ^ k * P t z) (∫ (z : ℝ), z ^ k * dP z) t) :
                HasDerivAt (fun (t : ℝ) => ∫ (z : ℝ), z ^ k * P t z) (∑ p ∈ T, c p * ↑(k.descFactorial p.2) * ∫ (z : ℝ), z ^ (k - p.2 + (p.1 - p.2)) * P t z) t

                (7.2.60): suppose ∂ₜ P = Σ_{(m,n) ∈ T} c_{m,n} (-∂_z)ⁿ [z^{m-n} P] (the right-hand side of (7.2.34), truncated to a finite set T of indices; c_{m,n} is Ω^{-(m-2)/2} α̃ₙ^{(m-n)}(φ)/(n!(m-n)!)), where G (m,n) is a derivative tower of z^{m-n} P. If d/dt commutes with ∫ dz (hswap), then d⟨zᵏ⟩/dt = Σ_{(m,n) ∈ T} c_{m,n} k!/(k-n)! ⟨z^{(k-n)+(m-n)}⟩, which is (7.2.60): the falling factorial k!/(k-n)! vanishes for n > k, so the inner sum runs to min(m, k).