A note on a generalized cohomology equation (Q4519005)

Let \(\varphi\) be an ergodic endomorphism of a probability space \((X,{\mathcal M},\mu)\), \(M(\mu)\) the space of \({\mathcal M}\)-measurable complex valued functions defined on \(X\). If \(f\in M(\mu)\), \(\lambda\in S^1\), then \(f\) is said to be \((\varphi,\lambda)\)-cohomologous to 0 in \(M(\mu)\) if there exists \(g\in M(\mu)\) (called a \((\varphi,\lambda)\)-coboundary of \(f\) in \(M(\mu)\)), such that \[ f= \lambda g-g\circ\varphi, \quad \mu\text{ a.e.} \] The aim of this paper is to generalize the result below (for \(\lambda=1\)) of \textit{J. Kwiatkowski, M. Lemaúczyk} and \textit{D. Rudolph} [Isr. J. Math. 80, 33-64 (1992; Zbl 0774.28008)]. Write \(S_n^\lambda(f)= \sum_{k=0}^{n-1} \lambda^{-k}f\circ \varphi^k\) where \(n\in \mathbb{Z}^+\), then the following is proved: Theorem. \(f\) is \((\varphi,\lambda)\)-cohomologous to 0 in \(M(\mu)\) if and only if there exists \(A\in{\mathcal M}\) with \(\mu(A)> 0\) and \(M\in \mathbb{R}^+\) such that \(|S_n^\lambda(f)(x) |\leq M\) for any \(n\in \mathbb{Z}^+\) and \(\mu\)-a.e., \(x\in A\) such that \(\varphi^n(x)\in A\).