The Euler–Maruyama (Cauchy–Euler) scheme #
Gardiner §4.3.1 (4.3.5)–(4.3.9) and §10.1.4 (10.1.13)–(10.1.17). On the uniform grid
t_i = t₀ + i τ, the Euler scheme is
x_{i+1} = x_i + a(x_i, t_i) τ + b(x_i, t_i) (W(t_{i+1}) - W(t_i)), x_0 = ξ (eulerSeq).
For the error analysis we use the continuous-time interpolation (eulerInterp)
X(u) = ξ + ∫_{t₀}^u a(x_{k(s)}, t_{k(s)}) ds + Σ_i b(x_i, t_i) (W(t_{i+1} ∧ u) - W(t_i ∧ u)),
which satisfies X(t_k) = x_k (eulerInterp_grid) and is the Itô process with the step
coefficients (eulerInterp_eq).
The uniform grid t_i = t₀ + i τ.
Equations
- Gardiner.Ito.StochDE.eulerGrid t₀ τ i = t₀ + ↑i * τ
Instances For
The Euler–Maruyama scheme (4.3.6) / (10.1.15):
x_{i+1} = x_i + a(x_i, t_i) τ + b(x_i, t_i) ΔW_i, x_0 = ξ.
Equations
- One or more equations did not get rendered due to their size.
- Gardiner.Ito.StochDE.eulerSeq W a b t₀ τ ξ 0 = ξ
Instances For
The continuous Euler interpolation
X(u) = ξ + ∫_{t₀}^u a(x_{k(s)}, t_{k(s)}) ds + Σ_i b(x_i, t_i) (W(t_{i+1} ∧ u) - W(t_i ∧ u)).
Equations
- Gardiner.Ito.StochDE.eulerInterp W a b t₀ τ ξ N u ω = (ξ ω + ∫ (s : ℝ) in t₀..u, Gardiner.Ito.StochDE.eulerDrift W a b t₀ τ ξ N s ω) + Gardiner.Ito.StochDE.eulerNoiseRun W a b t₀ τ ξ N u ω
Instances For
Grid arithmetic #
Measurability and integrability of the scheme #
The Euler iterates are adapted: x_i is 𝓕 t_i-measurable.
The Euler iterates are square integrable (up to the final time t_N ≤ T).
Cells of the grid #
The interpolation on the grid #
The step diffusion is the process of an elementary integrand, whose running sum is
eulerNoiseRun.
The interpolation is the Itô process with the step coefficients:
X(u) = ξ + ∫_{t₀}^u a(x_{k(s)}, t_{k(s)}) ds + ∫_{t₀}^u b(x_{k(s)}, t_{k(s)}) dW(s).