Stationary distribution of the stochastic theta method for nonlinear stochastic differential equations (Q2173342)

This paper studies numerical solution of the \(d\)-dimensional stochastic differential equation \[ dx(t) = f (x(t))dt + g(x(t))dB(t) \] where \(B(t)\) is a scalar Brownian motion, when using the stochastic theta numerical method \[ X_{k+1} = X_k +\theta f (X_{k+1})h+(1-\theta)f(X_k)h+g(X_k)\Delta B_k,\qquad X_0 = x(0) = x_0, \] where \(\Delta B_k = B(t_{k+1}) - B(t_k)\). Adapting an existence-uniqueness theorem due to Yuan and Mao, it is proved that the numerical stationary distributions converge to the stationary distribution of the exact solution of the SDE when \(\theta\in [1/2,1]\), and (with additional Lipschitz and growth conditions on \(f\)) when \(\theta\in [0, 1/2)\). Plots of numerical results for three examples are given to demonstrate the convergence.