Mean and quasideterministic equivalence for linear stochastic dynamics (Q1823190)

A linear stochastic population model is considered where growth is permitted by linear Gaussian white noise and a discrete set of linear Poisson noise terms. After discussing the Markov dynamics of the model, an exact solution of the governing linear stochastic differential equation of the constant parameter version is given by using Ito's generalized chain rule. A similar transformation procedure is then applied to the case of a model with time-dependent coefficients. These results are extended to a generalized Poisson process with independent random jump amplitudes that are conditioned on the occurence of a random jump. This shows that the quasideterministic approximation defined by an ordinary differential equation is equivalent to the mean of the linear stochastic process defined by the linear Ito stochastic differential equation with autonomous coefficients. Finally an exactly solvable counter example indicates that the equivalence result does not hold for nonlinear stochastic dynamics.