Large deviations for probabilistic cellular automata (Q1397384)

The authors consider a time-homogeneous Markov chain \(X_t\), \(t\in\mathbb Z^+\), with a metric state space \(\Gamma\). It is assumed that the chain \(X_t\) satisfies certain conditions; these conditions are satisfied, for example, for the traditional models of probabilistic cellular automata. For \(X_t\), the authors define the empirical pair distributions \(\Psi_T\) by the formula \[ \Psi_T=\frac{1}{T}\sum_{t=0}^{T-1} \delta(X_t,X_{t+1}), \qquad T\in\mathbb Z^+, \] where \(\delta(x,y)\) is the unit measure concentrated at a point \((x,y)\in\Gamma\times\Gamma\). The main results of the paper are lower large deviations bounds for \(\Psi_T\) which depend on the initial distribution of \(X_t\). The authors also consider upper large deviations bounds for \(\Psi_T\). The latter bounds follow from the results of \textit{M. D. Donsker} and \textit{S. R. S. Varadhan} [Commun. Pure Appl. Math. 36, 183-212 (1983; Zbl 0512.60068)], but the authors prefer to give a direct proof. Large deviations bounds for \(\Psi_T\) allow to obtain the corresponding bounds for the occupation measures \(\zeta_T\) defined by the formula \[ \zeta_T=\frac{1}{T}\sum_{t=0}^{T-1}\delta(X_t), \qquad T\in\mathbb Z^+, \] where \(\delta(x)\) is the unit measure concentrated at a point \(x\in\Gamma\).