Documentation

SHSLib.Gardiner.Escape.Populations

Equilibration of the populations in each well: §9.2, (9.2.1)–(9.2.33), (9.2.46)–(9.2.49) #

Kramers' method reduces the double-well Smoluchowski equation to the three relaxation equations (9.2.8) for the well populations N_a, N_c and the barrier population N₀. We formalise:

Errata recorded. (9.2.4) n_a = ∫_{-∞}^b p_s (not ∫^c). (9.2.13) N_a, N_c must be divided by n_a, n_c. (9.2.14) the coefficient of N₀ is D n₀⁻¹(κ⁻¹ + μ⁻¹) (and N₀ enters as N₀/n₀). (9.2.16) Ṅ_a = -r_a N_a + r_c N_c. (9.3.32) is (9.2.32). (9.2.33) for (9.2.26) the effective potential is U/γ and D = kT/γ, so H carries a factor γ (not γ⁻¹) and the factor (β_aβ_c)^{1/4}; with τ_r(0) = πH(0) the second line needs 2τ_r(0) in place of τ_r(0). (9.2.49) ψ(z) = n_a⁻¹ Σ_{y=z}^{b-1} P_s(y) for z ≤ b and n_c⁻¹ Σ_{y=b}^{z-1} P_s(y) for z > b (the text has upper limits b and z + 1).

(9.2.5)–(9.2.6) #

theorem Gardiner.Escape.flux_identity {U U' p p' : ℝ → ℝ} {D 𝒩 : ℝ} (hD : 0 < D) (h𝒩 : 0 < 𝒩) {x : ℝ} (hU : HasDerivAt U (U' x) x) (hp : HasDerivAt p (p' x) x) :
HasDerivAt (fun (y : ℝ) => p y / (𝒩 * Real.exp (-U y / D))) ((U' x * p x + D * p' x) / (D * (𝒩 * Real.exp (-U x / D)))) x

(9.2.5): with p_s = 𝒩 e^{-U/D}, the probability current of (9.1.1) is U'p + Dp' = D p_s ∂ₓ(p/p_s).

theorem Gardiner.Escape.kramers_integrated {U U' p p' : ℝ → ℝ} {D 𝒩 : ℝ} (hD : 0 < D) (h𝒩 : 0 < 𝒩) (hU : ∀ (x : ℝ), HasDerivAt U (U' x) x) (hp : ∀ (x : ℝ), HasDerivAt p (p' x) x) (hcont : Continuous fun (x : ℝ) => (U' x * p x + D * p' x) / (𝒩 * Real.exp (-U x / D))) (a x₀ : ℝ) :
∫ (x : ℝ) in a..x₀, (U' x * p x + D * p' x) / (𝒩 * Real.exp (-U x / D)) = D * (p x₀ / (𝒩 * Real.exp (-U x₀ / D)) - p a / (𝒩 * Real.exp (-U a / D)))

(9.2.6), exact: ∫_a^{x₀} (U'p + Dp')/p_s = D[p(x₀)/p_s(x₀) - p(a)/p_s(a)].

(9.2.7)–(9.2.18): the three relaxation equations #

theorem Gardiner.Escape.kramers_ansatz {Iint NaDot κ D N₀ n₀ Na na pa psa px psx : ℝ} (h96 : Iint = D * (px / psx - pa / psa)) (hans : Iint = κ * NaDot) (ha : pa / psa = Na / na) (hx : px / psx = N₀ / n₀) :
κ * NaDot = D * (N₀ / n₀ - Na / na)

(9.2.7) ⇒ (9.2.8). Under Kramers' ansatz, expressed as the named hypothesis that ∫_a^{x₀} M/p_s = N_a κ(x₀) with p(a)/p_s(a) = N_a/n_a and p(x₀)/p_s(x₀) = N₀/n₀, the exact identity (9.2.6) (whose time derivative gives d_t ∫ M/p_s) becomes κ Ṅ_a = D[N₀/n₀ - N_a/n_a].

theorem Gardiner.Escape.barrier_population {κ μ D N₀ n₀ Na Nc na nc NaDot : ℝ} (hκ : 0 < κ) (hμ : 0 < μ) (hD : 0 < D) (hn₀ : 0 < n₀) (ha : κ * NaDot = D * (N₀ / n₀ - Na / na)) (hc : μ * -NaDot = D * (N₀ / n₀ - Nc / nc)) :
N₀ = n₀ * (μ * (Na / na) + κ * (Nc / nc)) / (κ + μ)

(9.2.13), corrected. If κṄ_a = D[N₀/n₀ - N_a/n_a], μṄ_c = D[N₀/n₀ - N_c/n_c] and Ṅ_a = -Ṅ_c, then N₀ = n₀[μN_a/n_a + κN_c/n_c]/(κ + μ).

theorem Gardiner.Escape.barrier_population_printed_wrong :
1 * (1 * (1 / 2 / (1 / 4)) + 1 * (1 / 2 / (3 / 4))) / (1 + 1) = 4 / 3 ∧ 1 * (1 * (1 / 2) + 1 * (1 / 2)) / (1 + 1) = 1 / 2

Erratum (9.2.13). The printed N₀ = n₀[μN_a + κN_c]/(κ + μ) is not a consequence of (9.2.8): with κ = μ = D = n₀ = 1, n_a = 1/4, n_c = 3/4, N_a = N_c = 1/2 the equations (9.2.8) force N₀ = 4/3, not 1/2.

theorem Gardiner.Escape.threeState_stationary {κ μ D N₀ n₀ Na Nc na nc : ℝ} (hκ : 0 < κ) (hμ : 0 < μ) (hD : 0 < D) (hn₀ : 0 < n₀) (hna : na ≠ 0) (hnc : nc ≠ 0) :
D * (Na / (na * κ) + Nc / (nc * μ) - N₀ / n₀ * (κ⁻¹ + μ⁻¹)) = 0 ↔ N₀ = n₀ * (μ * (Na / na) + κ * (Nc / nc)) / (κ + μ)

(9.2.14), corrected. The intermediate-state equation Ṅ₀ = D{N_a/(n_aκ) + N_c/(n_cμ) - (N₀/n₀)(κ⁻¹ + μ⁻¹)} has, as its adiabatic (Ṅ₀ = 0) solution, exactly (9.2.13).

theorem Gardiner.Escape.relaxation_rates {κ μ D N₀ n₀ Na Nc na nc NaDot : ℝ} (hκ : 0 < κ) (hμ : 0 < μ) (hD : 0 < D) (hn₀ : 0 < n₀) (hna : 0 < na) (hnc : 0 < nc) (ha : κ * NaDot = D * (N₀ / n₀ - Na / na)) (hc : μ * -NaDot = D * (N₀ / n₀ - Nc / nc)) :
NaDot = -(D / (na * (κ + μ))) * Na + D / (nc * (κ + μ)) * Nc ∧ (na + nc = 1 → D / (na * (κ + μ)) + D / (nc * (κ + μ)) = D / (na * nc * (κ + μ)))

(9.2.16)–(9.2.18), corrected sign. Eliminating N₀ from (9.2.8): Ṅ_a = -r_a N_a + r_c N_c with r_a = D/(n_a I), r_c = D/(n_c I), I = κ + μ = ∫_a^c p_s⁻¹, and (for n_a + n_c = 1) r_a + r_c = D/(n_a n_c I).

theorem Gardiner.Escape.escape_time {D na κ : ℝ} (hD : 0 < D) (hna : 0 < na) (hκ : 0 < κ) :
∫ (t : ℝ) in Set.Ioi 0, Real.exp (-(D / (na * κ)) * t) = na * κ / D

(9.2.19)–(9.2.20). The decay law N_a(t) = N_a(0) e^{-Dt/(n_aκ)} has mean lifetime ∫₀^∞ e^{-Dt/(n_aκ)} dt = n_a κ/D.

§9.2.2: reversible denaturation of chymotrypsinogen #

theorem Gardiner.Escape.chymo_extremum {U₀a a k δp : ℝ} (hk : k ≠ 0) :
have U := fun (x : ℝ) => U₀a + k / 2 * (x - a) ^ 2 + x * δp; have β := -1 / k; HasDerivAt U 0 (a + β * δp) ∧ U (a + β * δp) = U₀a + a * δp + 1 / 2 * β * δp ^ 2

(9.2.30)–(9.2.31), exact for a locally quadratic U₀. If U₀(x) = U₀(a) + k/2 (x-a)² near a (k = U₀''(a) ≠ 0), then U = U₀ + xδp is stationary at a + βδp, β = -1/k, with value U₀(a) + aδp + ½βδp². (The same computation applies at b, with k = U₀''(b) < 0 and β_b = -1/U₀''(b) > 0, and at c.)

theorem Gardiner.Escape.chymo_population_ratio {ka kc U₀a U₀c a c δp kT : ℝ} (hka : 0 < ka) (hkc : 0 < kc) (hkT : 0 < kT) (hbal : √ka * Real.exp (U₀a / kT) = √kc * Real.exp (U₀c / kT)) :
√(kc / ka) * Real.exp ((U₀c + c * δp + 1 / 2 * (-1 / kc) * δp ^ 2 - (U₀a + a * δp + 1 / 2 * (-1 / ka) * δp ^ 2)) / kT) = Real.exp (-(a - c) / kT * δp - (-1 / ka - -1 / kc) / (2 * kT) * δp ^ 2)

(9.2.32) (printed with the label (9.3.32)). With U(x) = U₀(x) + xδp, quadratic extrema (9.2.31) and the balance condition (9.2.29) √k_a e^{U₀(a)/kT} = √k_c e^{U₀(c)/kT}, the ratio (9.2.23) √(k_c/k_a) e^{[U(c) - U(a)]/kT} equals exp[-(a - c)δp/kT - (β_a - β_c)δp²/(2kT)].

def Gardiner.Escape.chymoH (γ ka kb kc Ua Ub Uc kT : ℝ) :

The factor H of (9.2.25) for the chymotrypsinogen model (9.2.26), whose effective potential is U/γ and effective diffusion kT/γ: H = γ k_b^{-1/2}(k_a k_c)^{-1/4} e^{[2U(b) - U(a) - U(c)]/(2kT)} (curvatures k = |U''|).

Equations
Instances For
    theorem Gardiner.Escape.chymoH_eq {γ ka kb kc Ua Ub Uc kT : ℝ} (hγ : 0 < γ) (hka : 0 < ka) (hkb : 0 < kb) (hkc : 0 < kc) (hkT : 0 < kT) :
    chymoH γ ka kb kc Ua Ub Uc kT = γ * (√kb)⁻¹ * (√√(ka * kc))⁻¹ * Real.exp ((2 * Ub - Ua - Uc) / (2 * kT))
    theorem Gardiner.Escape.chymo_H_ratio {γ ka kb kc U₀a U₀b U₀c a b c δp kT : ℝ} (hγ : 0 < γ) (hka : 0 < ka) (hkb : 0 < kb) (hkc : 0 < kc) (hkT : 0 < kT) :
    chymoH γ ka kb kc (U₀a + a * δp + 1 / 2 * (-1 / ka) * δp ^ 2) (U₀b + b * δp + 1 / 2 * (1 / kb) * δp ^ 2) (U₀c + c * δp + 1 / 2 * (-1 / kc) * δp ^ 2) kT = chymoH γ ka kb kc U₀a U₀b U₀c kT * Real.exp (-(a + c - 2 * b) / (2 * kT) * δp + (1 / kb - 1 / 2 * (-1 / ka) - 1 / 2 * (-1 / kc)) / (2 * kT) * δp ^ 2)

    (9.2.33), corrected. With the extrema shifted by (9.2.31) and curvatures unchanged (locally quadratic U₀), H(δp) = H(0) exp[-(a + c - 2b)δp/(2kT) + (β_b - ½β_a - ½β_c)δp²/(2kT)], and H(0) = γ β_b^{1/2}(β_aβ_c)^{1/4} e^{[2U₀(b) - U₀(a) - U₀(c)]/(2kT)} (β_a = -1/k_a, β_c = -1/k_c, β_b = 1/k_b).

    theorem Gardiner.Escape.chymo_relaxTime {γ ka kb kc Ua Ub Uc kT : ℝ} (hγ : 0 < γ) (hka : 0 < ka) (hkb : 0 < kb) (hkc : 0 < kc) (hkT : 0 < kT) :
    2 * Real.pi * chymoH γ ka kb kc Ua Ub Uc kT * √(1 / 2 * (1 / 2)) = Real.pi * γ * √(1 / kb) * √√(1 / (ka * kc)) * Real.exp ((2 * Ub - Ua - Uc) / (2 * kT))

    (9.2.33), first line, corrected. At δp = 0 (n_a = n_c = ½) the relaxation time (9.2.24) is τ_r(0) = 2πH(0)·½ = πH(0), H(0) = γ β_b^{1/2}(β_aβ_c)^{1/4}e^{[2U(b)-U(a)-U(c)]/2kT} in terms of β_b = 1/k_b, β_aβ_c = 1/(k_ak_c).

    (9.2.46)–(9.2.49): the discrete ansatz #

    theorem Gardiner.Escape.discrete_ansatz_M {P Ps : ℕ → ℝ} {Na na : ℝ} {b : ℕ} (hna : na = ∑ y ∈ Finset.range b, Ps y) (hna0 : na ≠ 0) (hP : ∀ y < b, P y = Ps y * Na / na) {z : ℕ} (hz : z ≤ b) :
    ∑ y ∈ Finset.range z, P y = Na * (1 - (∑ y ∈ Finset.Ico z b, Ps y) / na)

    (9.2.46) ⇒ the corrected (9.2.49). If P(y) = P_s(y) N_a/n_a for y < b with n_a = Σ_{y<b} P_s(y), then for z ≤ b, M(z) = Σ_{y<z} P(y) = N_a[1 - ψ(z)] with ψ(z) = n_a⁻¹ Σ_{y=z}^{b-1} P_s(y) (upper limit b - 1, not b).

    theorem Gardiner.Escape.populations_instance :
    ∫ (t : ℝ) in Set.Ioi 0, Real.exp (-(1 / (1 / 2 * 3)) * t) = 1 / 2 * 3 / 1 ∧ 14 / 9 = 1 * (2 * (1 / 2 / (1 / 4)) + 1 * (1 / 2 / (3 / 4))) / (1 + 2) ∧ -4 / 9 = -(1 / (1 / 4 * (1 + 2))) * (1 / 2) + 1 / (3 / 4 * (1 + 2)) * (1 / 2)

    Instances: the mean escape time (9.2.20) for D = 1, n_a = ½, κ = 3 is 3/2; for κ = 1, μ = 2, D = n₀ = 1, n_a = ¼, n_c = ¾, N_a = N_c = ½, the relaxation equations (9.2.8) force N₀ = 14/9 (9.2.13 corrected) and Ṅ_a = -4/9 = -r_aN_a + r_cN_c (9.2.16).