Stability in distribution of randomly perturbed quadratic maps as Markov processes (Q1769412): Difference between revisions

Stability in distribution means the convergence in distribution of \(\frac 1n \sum_{m=0}^{n-1}p^{(m)}(x,dy)\) to the same limit \(\pi(dy)\) for every initial state \(x\in S\), with \(p^{(n)}\) denoting the \(n\)-step transition probability of a Markov process \((X_n,n\geq 1)\) with the state space \(S=(0,1)\). Then \(\pi\) is the unique invariant probability of this Markov process. Randomly perturbed dynamical systems of random compositions of the form \(X_n=F_{\varepsilon_{n}}F_{\varepsilon_{n-1}}\cdots F_{\varepsilon_{1}} X_0\) are considered, where \(\varepsilon_n\), \(n\geq 1\), is an i.i.d. sequence with values in the parameter space \([0,4]\) of the quadratic map \(F_{\theta}=\theta x(1-x)\), and \(X_0\) is independent of \((\varepsilon_n, n\geq 1)\). This model may be applied to problems of optimization under uncertainty arising in economics. For related papers see: \textit{V. Anantharam} and \textit{T. Konstantopoulos} [Stochastic Processes Appl. 68, 181--194 (1997; Zbl 0911.60076); Correction ibid. 80, 271--278 (1999; Zbl 0962.60086)]; \textit{K. B. Athreya} and \textit{J. J. Dai} [Ann. Probab. 30, 437--442 (2002; Zbl 1013.60054)] and \textit{K. Mitra} [Econ. Theory 11, 457--464 (1998; Zbl 0899.90032)].