Title:On the connection between a skew product IFS and the ergodic optimization for a finite family of potentials

Abstract:We study the skew product IFS on the cylinder $\textbf{S}^{1}\times \mathbb{R}$ defined by the maps $G_{c}(x,y)= (T(x) , A_{c}(x)+\lambda y), c \in \mathcal{C}$ for a finite family $\mathcal{C}$ of $m \geq 2$ potential functions $A_{c}:\textbf{S}^{1} \to \mathbb{R}$ where $T: \textbf{S}^{1} \to \textbf{S}^{1}$ is the doubling map. We show that there exists a compact invariant set $\Lambda $ which has an attractive behavior. Then, we introduce the random SRB measure $\mu \in Prob(\textbf{S}^{1}\times \mathbb{R}\times \mathcal{C}^{\mathbb{N}})$ which is ergodic with respect to the skew dynamical system $ G (x,y,\bar{c})=(G_{c_{0}}(x,y) , \sigma(\bar{c}))$, having the property that $\mu$ a.e.
$$\frac{1}{N}\sum_{j=0}^{N-1} g( G ^{j}(x,y,\bar{b})) \stackrel{N\to\infty}{\to} \int_{X\times \mathbb{R} \times \mathcal{C}^{\mathbb{N}}} g(x,y,\bar{b}) d\mu(x,y,\bar{b}),$$
for all $g \in L^1(\mu)$. In the second part, we study the IFS ergodic optimization problem for a finite family of potentials. As a tool to understand it, we study, for each $\lambda <1$, the discounted IFS ergodic optimization problem for a finite family of potentials. Moreover, we show that if $\mu=\mu_{x_0,\bar{c},\bar{a}}^{\lambda}$ is an empirical discounted holonomic measure, then $\displaystyle \int A_{c}(\tau_a x) d\mu(x,c,a)= (1-\lambda) S_{x_0}(\bar{c},\bar{a}).$ This shows the connection between this maximization problem and the superior boundary of the set $\Lambda$. Additionally, when the discounted holonomic measures has trace $\nu=\delta_{x_0}$ supported in a point of maximum of $v_{\lambda}^{+}$ we show that $m_{\lambda}(R)\stackrel{\lambda \to 1}{\to} m(R)$ and any cluster point of optimal discounted holonomic measures is a optimal holonomic measure.