On self-similar sets with overlaps and inverse theorems for entropy (Q742913)

Let \(\Phi :=\{\varphi_i : i\in \Lambda\}\) be a finite collection of linear contractions of the form \(\varphi_i (x) := r_i x + a_i\), where \(|r_i| < 1\) and \(a_i\in \mathbb R\). Let \(X\neq\emptyset\) be the attractor of the iterated function system \(\Phi\) and let the self-similar measure associated with \(\Phi\) and a probability vector \((p_i)_{i\in \Lambda}\) be denoted by \(\mu\). The author investigates the Hausdorff dimension \(\dim \mu\) of \(\mu\) in case the images \(\varphi_i X\) have significant overlap. The main result is the following: If \(\dim\mu < \min\{1, s\}\), where \(s\) is the similarity dimension of \(X\), then there exist superexponentially close cylinders at small enough scales. As a consequence of this result, the Furstenberg conjecture on the projections of the one-dimensional Sierpiński gasket is proven and some progress on the Bernoulli convolutions problem is achieved.