A weak form of stochastic Newmark method with applications to engineering dynamical systems (Q1401091)

The author considers a stochastic differential equation (SDE) with \(n\) degrees of freedom, which is of practical interest in engeneering: \[ [M] \ddot{X} + C(X,\dot{X})\dot{X}+ K(X,\dot{X})X = \sum_{r=1}^q G_r(X,\dot{X},t)\dot{W}_r + B(t).\tag{1} \] Here \([M]\) is a constant mass matrix, \(C,K\) are (state-dependent for nonlinear systems) the damping and stiffness matrices, respectively, \(G_r\) is the \(r\)th element of a set of \(n\times 1\) diffusion vectors, \(\{ W_r \}\) is a \(q\)-dimensional vector of standard Wiener processes and \(B(t)\) is the external deterministic force vector. The second-order equation is rewritten into an \(n\)-dimensional system of Itô SDEs. Using Itô-Taylor expansions the author derives a stochastic Newmark method. In the numerical analysis literature Newmark methods would more likely be found under the name \(\Theta\)-methods, the version used here is an appropriate adaption for second order equations. The author then analyses the weak convergence properties of the method, in the spirit of \textit{G. Milstein} [``Numerical integration of stochastic differential equations'' (1994; Zbl 0810.65144)]. Numerical examples conclude the article.