On the \(k\)-analogue of a result in the theory of the Riemann zeta function (Q1839898): Difference between revisions

Let \(\chi(n)\) be a real primitive character \(\pmod k\). The author proves that \[ L_{\chi}(1)=\sum_1^\infty \frac{\chi(n)}{n}=\Omega_R(\log\log k), \] and that \[ \sum_1^{k/4} \chi(n)=\Omega_R(\sqrt k\log\log k). \] Weaker results of this type have been proved by \textit{J. E. Littlewood} [Proc. Lond. Math. Soc. 27, 358--372 (1928; JFM 54.0206.02)] and \textit{R. E. A. C. Paley} [J. Lond. Math. Soc. 7, 28--32 (1932; Zbl 0003.34101; JFM 58.0192.01)], respectively. To obtain both these results, it suffices to prove that \[ r_3(m)=\Omega(\sqrt k\log\log m)\tag{1} \] for quadratfrei integers \(m\equiv1\pmod 4\). (\(r_k(m)\) denotes the number of representations of \(m\) as the sum of \(k\) squares.) The author shows that if \(N\) is the product of the first \(r\) primes, then \(r_4(N)=\Omega(N\log\log N)\), and so \[ \sum_{x \text{odd}} r_3(N-x^2)=\Omega(N\log\log N). \] He now proves that in this sum there are only \(O\left(\frac{\sqrt N}{\log N}\right)\) terms for which \(N-x^2\) is not quadratfrei, and that each such term is \(O(\sqrt N\log N)\). From these facts (1) follows.