Bounds on short character sums and \(L\)-functions with characters to a powerful modulus (Q2299475)

The authors study the character sum \(S_\chi(M,N)=\sum_{M n\le M+N}\chi(n)\) when \(\chi\) is a primitive Dirichlet character to some modulus \(q\) with a small `core'. On setting \(q_\sharp=\prod_{p|q}p\), \(\gamma_+=\max_{p|q}v_p(q)\) and \(\gamma_-=\min_{p|q}v_p(q)\) where \(v_p\) denotes the usual \(p\)-adic valuation, the first result of this paper is that there exist constants \(A\), \(\gamma_0\) and \(\xi_0\) such that, when \[ \gamma_-\ge 0.7\,\gamma_+, \quad \gamma_+\ge \gamma_0, \quad N\ge q_\sharp^{\gamma_0}, \] we have \(|S_\chi(M,N)|\le A N^{1-\xi_0/\rho^2}\) where \(\rho=(\log q)/\log N\). This improves on earlier work by \textit{H. Iwaniec} and \textit{E. Kowalski} [Analytic number theory. Providence, RI: American Mathematical Society (AMS) (2004; Zbl 1059.11001)]. Similar bounds for the sum \(\displaystyle\sum_{M n\le M+N}\chi(n)e(G(n))\) for some real polynomial of degree bounded by a constant times \(\rho\) and for the sum \(\displaystyle\sum_{M n\le M+N}\chi(n)n^{it}\) are also provided. As consequences, the authors improve on the asymptotic formula for primes in short arithmetic progression modulo a power of a fixed prime, they extend the classical zero-free region for some classes of moduli (a possible Siegel zero remains) and improve the value of the Linnik constant, when considering moduli that are powers of a fixed prime, to \(L=2.1115\). The proof starts by reducing \(S_\chi(M,N)\) to an average of \(\sum_{n}\chi(n+q_\sharp^syz)\) and then to \(\sum_{n}\chi(n)\sum_{y,z}\chi(1+q_\sharp^s\overline{n}yz)\). The sum of \(y\) and \(z\) is then handled in an `additive' way, starting by Lemma 2 of \textit{H. Iwaniec} [Invent. Math. 23, 97--104 (1974; Zbl 0275.10024)] and then exploiting the bounds provided by the work of \textit{N. M. Korobov} in Lemma 3 of [Math. USSR, Sb. 18, 659--676 (1974); translation from Mat. Sb., N. Ser. 89(131), 654--670 (1972; Zbl 0248.10007)] and by the one of \textit{K. Ford} in Theorem 3 of [Proc. Lond. Math. Soc. (3) 85, No. 3, 565--633 (2002; Zbl 1034.11044)] .