Noise-induced phase transitions in field-dependent relaxational dynamics: the Gaussian ansatz.

We present an analytic mean-field theory for relaxational dynamics in spatially extended systems that undergo purely noise-induced phase transitions to ordered states. The theory augments the usual mean-field approach with a Gaussian ansatz that yields quantitatively accurate results for strong coupling. We obtain analytic results not only for steady-state mean fields and distribution widths, but also for the dynamical approach to a steady state or to collective oscillatory behaviors in multifield systems. Because the theory yields dynamical information, it can also predict the initial-condition-dependent final state (disordered state, steady or oscillatory ordered state) in multistable arrays.