Möbius disjointness for analytic skew products (Q2406816)

\textit{P. C. Sarnak} [Not. S. Afr. Math. Soc. 43, No. 2, 89--97 (2012)] formulated the following conjecture, giving a dynamical interpretation of an expected randomness in the behaviour of the Möbius function. For a topological dynamical system \(T: X\to X\), where \(X\) is a compact metric space and \(T\) is a homeomorphism, does zero topological entropy for \(T\) imply that for any function \(f\in C(X)\) and point \(x\in X\) we have \((1/N)\sum_{n=1}^{N}f(T^nx)\mu(n)\to0\) as \(N\to\infty\)? That is, does zero topological entropy imply a form of orthogonality on average to the Möbius function? This has been shown for a great many classes of dynamical systems, but the complete conjecture is still open. Here it is shown that the conjecture holds for all maps of the form \(T:(x,y)\mapsto(x+\alpha,y+h(x))\) on the 2-torus \(\mathbb{T}^2\), where the skewing function \(h:\mathbb{T}^1\to\mathbb{T}^1\) is analytic. A specific point to notice is that -- just as the question of Sarnak concerns the behaviour at every point \(x\) -- there is no hypothesis needed on the underlying rotation \(\alpha\).