Random \beta-transformation on fat Sierpinski gasket

Abstract

We consider the iterated function system (IFS) $$f_{\vec{q}}(\vec{z})=\frac{\vec{z}+\vec{q}}{\beta},\vec{q}\in\{(0,0),(1,0),(0,1)\}.$$ As is well known, for $\beta = 2$ the attractor, $S_\beta$, is a fractal called the Sierpi\'nski gasket(or sieve) and for $\beta>2$ it is also a fractal. Our goal is to study greedy, lazy and random $\beta$-transformations on the attractor for this IFS with $1<\beta<2$. For $1<\beta\leq 3/2$, $S_\beta$ is a triangle and it is shown that the greedy transformation $T_\beta$ and the lazy transformation $L_\beta$ are isomorphic and they both admit an absolutely continuous invariant measure. We show that all $\beta$-expansions of a point $\vec{z}$ in $S_\beta$ can be generated by a random map $K_\beta$ defined on $\{0,1\}^\mathbb{N}\times\{0,1,2\}^\mathbb{N}\times S_\beta$ and $K_\beta$ has a unique invariant measure of maximal entropy when $1<\beta\leq\beta_*$, where $\beta_*\approx 1.4656$ is the root of $x^3-x^2-1=0$. We also show existence of a $K_\beta$-invariant probability measure, absolutely continuous with respect to $m_1\otimes m_2 \otimes \lambda_2$, where $m_1, m_2$ are product measures on $\{0,1\}^\mathbb{N},\{0,1,2\}^\mathbb{N}$, respectively, and $\lambda_2$ is the normalized Lebesgue measure on $S_\beta$. For $3/2<\beta\leq \beta^*$, where $\beta^*\approx 1.5437$ is the root of $x^3-2x^2+2x=2$, there are radial holes in $S_\beta$. In this case, $K_\beta$ is defined on $\{0,1\}^\mathbb{N}\times S_\beta$. We also show that it has a unique invariant measure of maximal entropy.