Finite precision representation of the Conley decomposition (Q5935516)

In dynamical systems it is of great interest to understand the relationship between the orbits of a system and the dynamic behavior of random perturbations of such system. This is the situation when one considers problems of roundoff in numerical computations and studies the effects that noise introduces in the dynamics of a system. In the paper under review, the author states the problem of obtaining information of the long-time behavior of an unperturbed system when the same behavior of the corresponding randomly perturbed system is known. To that end he explores the relationship between the communication classes of the Markov chain associated to a random perturbation and basic sets (maximal chain transitive sets) of the Conley decomposition of a dynamical system. Given the dynamical system \((X,f)\) with \(X\) a compact set and \(f\) an homeomorphism, Conley's theorem states that the long-time dynamics occur precisely on the chain recurrent set \({\mathcal R}(f)\). Besides, if \({\mathcal R}(f)\) has hyperbolic structure then it can be represented as the finite union of basic sets and the dynamics between them is of gradient-like type while within a basic set itself it is topologically transitive. In the case of nonhomeomorphisms, in the paper sufficient conditions are given for the existence of a chain transitive set and it is shown that basic sets are isolated from each other through some neighborhoods associated with closed communication classes of the chain. Finally, in the setting of measure theory, when the system supports a Sinai-Bowen-Ruelle measure there exists a unique closed communication class being irreducible for the corresponding Markov chain when it is restricted to those states.