Abstract:We study a random dynamical system such that one transformation is randomly selected from a family of transformations and then applied on each iteration. For such random dynamical systems, we consider estimates of absolutely continuous invariant measures. Since the random dynamical systems are made by complicated compositions of many deterministic maps and probability density functions, it is difficult to estimate the invariant measures. To get rid of this difficulty, we present fundamental comparison theorems which make easier the estimates of invariant measures of random maps. We also demonstrate how to apply the comparison theorems to random maps with indifferent fixed points and/or with unbounded derivatives.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the difficulty in estimating absolutely continuous invariant measures in stochastic dynamical systems. Specifically, the paper focuses on how to simplify the estimation of absolutely continuous invariant measures for one - dimensional stochastic mappings with different fixed points or unbounded derivatives. Since these stochastic dynamical systems are composed of complex combinations of many deterministic mappings and probability density functions, it is very difficult to directly estimate their invariant measures. For this reason, the author proposes some basic comparison theorems, which can simplify the estimation process of the invariant measures of these stochastic mappings.
### Main contributions of the paper
1. **Proposing comparison theorems**: The author proposes two basic comparison theorems, one for lower - bound estimation and the other for upper - bound estimation. These theorems hold in general settings and can therefore be applied to multiple types of stochastic mappings.
2. **Application of auxiliary stochastic mappings**: In order to use the comparison theorems, it is necessary to select appropriate auxiliary stochastic mappings whose invariant measures are easier to estimate. The paper shows through specific examples how to select these auxiliary mappings and apply the comparison theorems to estimate the invariant measures of stochastic mappings with different fixed points or unbounded derivatives.
3. **Combination of theory and practice**: The paper not only provides theoretical comparison theorems but also verifies the effectiveness of these theorems through specific examples, showing how to apply these theorems in practical problems.
### Solutions to specific problems
- **Lower - bound estimation**: By introducing an auxiliary stochastic mapping \(\bar{\tau}\), the invariant measure \(\bar{\mu}\) of \(\bar{\tau}\) is made easier to estimate. Then, using the comparison theorem, a lower - bound estimate of the invariant measure \(\mu\) of the original stochastic mapping \(T\) is obtained.
- **Upper - bound estimation**: By introducing another auxiliary stochastic mapping \(\hat{\tau}\), the invariant measure \(\hat{\mu}\) of \(\hat{\tau}\) is made easier to estimate. Similarly, using the comparison theorem, an upper - bound estimate of the invariant measure \(\mu\) of the original stochastic mapping \(T\) is obtained.
### Formula summary
- **Lower - bound estimation formula**:
\[
\frac{\gamma_1}{\bar{\gamma}_2} \bar{\mu}([0,\epsilon]) \leq \mu([0,\epsilon]) \quad \text{for any } \epsilon \in (0, c^*).
\]
- **Upper - bound estimation formula**:
\[
\mu([0,\epsilon]) \leq \frac{\gamma_2}{\hat{\gamma}_1} \inf_{x \in [0, c^*)} \hat{p}(x) \hat{\mu}([0,\epsilon]) \quad \text{for any } \epsilon \in (0, c^*).
\]
### Conclusion
Through these comparison theorems, the author provides an effective method to estimate the absolutely continuous invariant measures of one - dimensional stochastic mappings with different fixed points or unbounded derivatives, thus solving a key problem in this field.