A level 1 large-deviation principle for the autocovariances of uniquely ergodic transformations with additive noise (Q1284901): Difference between revisions

A large deviation principle (LDP) at level 1 for random means of the type \[ M_n={1\over n}\sum_{j=0}^{n-1}Z_j Z_{j+1},\quad j=1,2,\ldots, \] is established. The random process \(\{Z_n\}_{n\geq 0}\) is given by \(Z_n=\Phi(X_n)+\xi_n\), \(n=0,1,2,\ldots\), where \(\{X_n\}_{n\geq 0}\) and \(\{\xi_n\}_{n\geq 0}\) are independent random sequences: the former is a stationary process defined by \(X_n=T^n(X_0)\), \(X_0\) is uniformly distributed on the circle \(S^1\), \(T:S^1\to S^1\) is a continuous, uniquely ergodic transformation preserving the Lebesgue measure on \(S^1\), and \(\{\xi_n\}_{n\geq 0}\) is a random sequence of independent and identically distributed random variables on \(S^1\); \(\Phi\) is a continuous real function. The motivation of analyzing such processes is given by \textit{A. Lopes} and \textit{S. Lopes} [Adv. Appl. Probab. 30, No.~3, 757-776 (1998; Zbl 0917.60043); ``Unique ergodicity, large deviations and parametric estimation'' (submitted 1998)]. The LDP at level 1 for the means \(M_n\) is obtained by using the level 2 LDP for the Markov process \(\{V_n=(X_n,\xi_n,\xi_{n+1})\}_{n\geq 0}\) and the contraction principle. For establishing this level 2 LDP, one can consider a more general setting: \(T:[0,1)\to[0,1)\) is a uniquely ergodic transformation (not necessarily continuous) preserving the Lebesgue measure, \(\Phi:[0,1)\to\mathbb{R}\) is a real measurable function, and \(\xi_n\) are independent and identically distributed random variables on \(\mathbb{R}\). The analogous result for the case of autocovariance of order \(k\) is also true.