scientific article; zbMATH DE number 3932247

From MaRDI portal
Publication:3704772





[For the entire collection see Zbl 0572.00015.] Consider a system S, observed in state \(X_ n\in I\) at time n; \(n=0,1,2,..\). The evolution of S is assumed Markovian, but dependent on the value of a control parameter \(Z_ n=\sigma_ n(X_ 0,...,X_ n)\) (where \(Z=\{Z_ n\}\) is called stationary if \(Z_ n=\sigma (X_ n))\). The probability law of the system process involves, furthermore, an unknown parameter \(\alpha \in A\subseteq R^ 1\), the estimation of which corresponds to the identification of S. The paper is concerned with establishing conditions under which the family \(\{P_{\alpha}^{Z,N}:\alpha \in A\); \(N=0,1,2,...\}\) of joint distributions of \((X_ 0,...,X_ N)\) is locally asymptotically normal, for given asymptotically stationary control Z. This is done for both finite and countable I. The results, combined with a theorem of Hajek, provide bounds for the quality of identification of S.











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 Q3704772)