\(I\)-projection and conditional limit theorems for discrete parameter Markov processes (Q686760)

Let \((X,{\mathcal B})\) be a compact metric space with \({\mathcal B}\) the \(\sigma\)-field of Borel sets. Suppose this is the state space of a discrete parameter Markov process. Let \(C\) be a closed convex set of probability measures on \(X\). Known results on the asymptotic behaviour of the probability that the empirical distributions \(\widehat{P}_ n\) belong to \(C\) and new results on the Markov process distribution of \(\omega_ 0,\ldots,\omega_{n-1}\) under the condition \(\widehat{P}_ n\in C\) are obtained simultaneously through a large deviations estimate. In particular, the Markov process distribution under the condition \(\widehat{P}_ n\in C\) is shown to have an asymptotic quasi-Markov property, generalizing a concept of Csiszár. Large deviations are understood in the spirit of \textit{M. D. Donsker} and \textit{S. R. S. Varadhan} [Commun. Pure Appl. Math. 28, 1-47 (1975; Zbl 0323.60069) and ibid. 29, 389-461 (1976; Zbl 0348.60032)]. Other corner- stones of this paper are the Sanov property and \(I\)-projection [see \textit{I. Csiszár}, Stud. Sci. Math. Hungar. 2, 299-318 (1967; Zbl 0157.258), Ann. Probab. 3, 146-158 (1975; Zbl 0318.60013) and ibid. 12, 768-793 (1984; Zbl 0544.60011); \textit{I.Csiszár, T. M. Cover} and \textit{B.-S. Cho}, IEEE Trans. Inf. Theory IT-33, 788-801 (1987; Zbl 0628.60037)].