A bifurcation theory for a class of discrete time Markovian stochastic systems

We present a bifurcation theory of smooth stochastic dynamical systems that are governed by everywhere positive transition densities. The local dependence structure of the unique strictly stationary evolution of such a system can be expressed by the ratio of joint and marginal probability densities; this ‘dependence ratio’ is a geometric invariant of the system. By introducing an equivalence relation defined on these dependence ratios, we arrive at a bifurcation theory for which in the compact case, the set of stable, i.e. non-bifurcating, systems is open and dense. The theory is illustrated with some simple examples.