Multivariate Böttcher equation for polynomials with non-negative coefficients (Q1601709)

``For a given function \(f:\mathbb{R}^d\to\mathbb{R}^d\), a solution to the multivariate Böttcher equation is a function \(\beta:\mathbb{R}^d\to\mathbb{R}^d\) and a constant \(\lambda \in \mathbb{R}\) such that \(\beta(f(x))=\beta(x)^\lambda\) for \(x\in S\subset\mathbb{R}^d\), where \(\beta(x)^\lambda\) is taken componentwise. If there exists a \(\lambda\) such that \((1)\quad F(x):=\lim_{n\to\infty}-\lambda^{-n}\log f^n(x)\) exists, then \(\beta(x)=\exp(-F(x))\) is a solution to the multivariate Böttcher equation for \(f\). Here \(\log\) and \(\exp\) are taken componentwise. Accordingly, we look for conditions under which the limit (1) exists. We will restrict our attention to functions \(f:[0,1]^d\times [0,1]^d\) \dots. The initial motivation for this study was an investigation of large deviations for supercritical multitype branching processes.'' (From the author's introduction.). For the univariate Böttcher equation the author refers to the classical book of \textit{M. Kuczma} [Functional equations in a single variable. PWN Warszawa (1968; Zbl 0196.16403)].