Asymptotic behavior of one-dimensional random dynamical systems (Q1072184)

Let (\(\Omega\),\({\mathcal F},P,\sigma)\) be a measure preserving dynamical system on a Lebesgue space and \(\tau\) : \(\Omega\) \(\times I\to I\) be a measurable map such that for every \(\omega\) \(\tau_{\omega}=\tau (\omega,\cdot)\) is a nondegenerate piecewise (monotone and) \(C^ 2\) transformation of the unit interval I (thus nonsingular with respect to the Lebesgue measure m). The sequence of iterations \(\tau_{\sigma^{n- 1}\omega}\circ...\circ \tau_{\omega}\circ \tau_{\omega}\) is called a random dynamical system. Results: Under some conditions on the statistics of the derivatives of \(\tau\) there exist (m\(\times P)\)-absolutely continuous finite measures invariant under the associated skew product \(T: I\times \Omega \hookleftarrow\); \((x,\omega)\mapsto (\tau_{\omega}(x),\sigma \omega)\), called a.c.i. measures. If furthermore \(\sigma\) is ergodic there are finitely many ergodic a.c.i. probability measures; if \(\sigma\) is totally ergodic they are totally ergodic for certain powers of T and will be exact if \(\sigma\) has been exact. The proofs employ the investigation of asymptotic properties of the Frobenius-Perron operator of T.