Mean-square stability of numerical schemes for stochastic differential systems (Q1411128)

Criteria are derived for establishing mean-square (MS)-stability of the system of stochastic differential equations \[ d{\mathbf X}(t)= {\mathbf D}{\mathbf X}(t)\,dt+{\mathbf B}{\mathbf X}(t)\,dW(t),\quad{\mathbf X}(0)= 1, \] where \[ {\mathbf D}= \left[\begin{matrix}\lambda_1 & 0\\ 0 &\lambda_2\end{matrix}\right],\qquad{\mathbf B}= \left[\begin{matrix} \alpha_1 &\beta_1\\ \beta_2 &\alpha_2\end{matrix}\right], \] and \(W(t)\) is a Wiener process. This leads to criteria under which the Euler-Maruyama method for approximating the solution of the system will be numerically MS-stable, and to the identification of its region of MS-stability. Results of numerical experiments are presented which affirm the accuracy of the criteria.