Two-step Runge-Kutta methods for stochastic differential equations.

We introduce a theory of two-step Runge-Kutta (TSRK) methods for stochastic differential equations, arising from the perturbation of the corresponding TSRK methods for deterministic problems. We present a proof of convergence and study the mean-square stability properties. Numerical experiments confirming the theoretical results are provided. [ABSTRACT FROM AUTHOR]