Explicit expression of stationary response probability density for nonlinear stochastic systems (Q2234260)

A nonlinear Newtonian dynamics with Gaussian white noise input is considered wherein the coefficients are polynomial nonlinearities. A stationary Gibbs distribution is assumed and its exponent is approximated by a truncated series expansion. This is rearranged into a weighted linear combination of functions of parameters and state variables. The weights are analyzed using dimensional analysis and the unknown component is obtained by minimizing the mean square approximation error. Numerical examples are also provided in support of the proposed scheme.