scientific article; zbMATH DE number 177232

From MaRDI portal
Publication:4037082





[For the entire collection see Zbl 0754.00005.] This paper is devoted to the logarithmic asymptotics of large deviations for empirical distributions of Markov and stationary processes with discrete and continuous time. The proofs are based on Freĭdlin's method (by which the author seems to mean cumulant generating function method). Various types of assumptions are considered for Markov case, for diffusion processes and for stationary processes; in particular mixing- type condition (5.1) of the paper is equivalent to the \(\psi\)-condition (2.3) in the paper by the reviewer [Ann. Probab. 20, No. 2, 1004--1030 (1992; Zbl 0756.60028)], see Proposition 5.1 there]. Approximation by compacts assumption of Section 3 should perhaps be compared with [\textit{A. De Acosta}, J. Theor. Probab. 3, No. 3, 395--431 (1990; Zbl 0711.60023)]. The conclusions of theorems are slightly weaker than the usual ``large deviation principle in the sense of Varadhan.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4037082)