CP-chains and dimension preservation for projections of $(\times m,\times n)$-invariant Gibbs measures

TL;DR

This paper extends the dimension preservation results for projections of $( imes m, imes n)$-invariant measures from Bernoulli to Gibbs measures on transitive SFTs using ergodic averages.

Contribution

It generalizes the dimension conservation property to a broader class of Gibbs measures, beyond Bernoulli measures, via ergodic theoretic methods.

Findings

Dimension preservation holds for Gibbs measures on transitive SFTs.

Reduction to pointwise convergence of double ergodic averages.

Applicable to measures invariant under multiplicatively independent transformations.

Abstract

Dimension conservation for almost every projection has been well-established by the work of Marstrand, Mattila and Hunt and Kaloshin. More recently, Hochman and Shmerkin used CP-chains, a tool first introduced by Furstenberg, to prove all projections preserve dimension of measures on $[0,1]^2$ that are the product of a $\times m$-invariant and a $\times n$-invariant measure (for $m$, $n$ multiplicatively independent). Using these tools, Ferguson, Fraser and Sahlsten extended that conservation result to $(\times m,\times n)$-invariant measures that are the pushforward of a Bernoulli scheme under the $(m,n)$-adic symbolic encoding. Their proof relied on a parametrization of conditional measures which could not be extended beyond the Bernoulli case. In this work, we extend their result from Bernoulli measures to Gibbs measures on any transitive SFT. Rather than attempt a similar parametrization, the proof is achieved by reducing the problem to that of the pointwise convergence of a double ergodic average which is known to hold when the system is exact.