Weak second order multirevolution composition methods for highly oscillatory stochastic differential equations with additive or multiplicative noise (Q6486755)

Deterministic multirevolution composition methods are extended to apply to systems of highly oscillatory nonlinear Itô stochastic differential equations of the form NEWLINE\[NEWLINEJX(t)- (e^{-1} AX(t)+ f(X(t)))\,dt+ \sum^m_{r=1} g^r(X(t))\,dW_r(t),\;0\leq r\leq I,NEWLINE\]NEWLINE where \(\varepsilon>0\) is fixed, the \(W_r\) are independent standard Wiener processes, and on the oscillatory part of the system \(e^A=I\). Algorithms that implement the methods are described. Second-order error estimates are proved for the methods and algorithms. Numerical results for four examples are presented that provide evidence of second-order weak convergence and enable a comparison with known methods that indicates a significant computational cost advantage of the new methods.