Additive functions in short intervals, gaps and a conjecture of Erdős (Q2097510)

Inspired by the impressive work of \textit{K. Matomäki} and \textit{M. Radziwiłł} [Ann. of Math. (2), 183, 1015--1056 (2016; Zbl 1339.11084)] relating short averages of the multiplicative function \(f:\mathbb{N}\to[-1,1]\) to its long averages, in the paper under review, the author establishes similar study concerning additive functions, by proving that if \(g:\mathbb{N}\to\mathbb{C}\) is additive and \(h\in[10,X/100]\cap\mathbb{Z}\), then \begin{multline*} \frac{2}{X}\sum_{\frac{X}{2}<n\leq X}\left|\frac{1}{h}\sum_{n-h<m\leq n} g(m)-\frac{2}{X}\sum_{\frac{X}{2}<m\leq X}g(m)\right|\\ \ll\left(\sqrt{\frac{\log\log h}{\log h}}+(\log X)^{-\frac{1}{800}}\right)B_g(X), \end{multline*} where \[ B_g(X)=\left(\sum_{p^k\leq X}|g(p^k)|^2p^{-k}\right)^{\frac{1}{2}}. \] He also obtains an \(\ell^2\) analogue of the above result with error term \(o(B_g(X)^2)\) for a certain family of additive functions, including the well-known number theoretic omega functions \(\omega\) and \(\Omega\). Generality of the above two mentioned results, allows to provide several applications. For additive and real function \(g\) we let \(\mathcal{B}=\{n\in\mathbb{N}: g(n)<g(n-1)\}\) with \(g(0)=0\), and \(\mathcal{B}(X)=\mathcal{B}\cap [1,X]\). Erdős' conjecture for almost everywhere monotone additive functions asserts that if \(|\mathcal{B}(X)|=o(X)\), then then there exists some real \(c\) such that \(g(n)=c\log n\) for all \(n\in\mathbb{N}\). As an application, the author proves a partial result towards Erdős' conjecture, with some more assumptions. By using a result of Elliott, the author proves an approximate version of Erdős' conjecture under weaker conditions than in the above mentioned application. As another application, the author proves equivalency of approximations concerning the \(\ell^1\) averages of \(g(n)-g(n-1)\) and \(g(n)-A_g(X)\), where \[ A_g(X)=\sum_{p^k\leq X}\frac{g(p^k)}{p^k}\left(1-\frac{1}{p}\right). \]