On an Erdős-pomerance conjecture for rank one Drinfeld modules (Q2517120)

Let \(n\in{\mathbb N}\) and consider \(b\in{\mathbb Z}\) such that \(\gcd(b,n)=1\). Let \(l_b(n)\) be the order of \(b\bmod n\). \textit{P. Erdős} and \textit{C. Pomerance} conjectured in [Rocky Mt. J. Math. 15, 343--352 (1985; Zbl 0617.10037)] that if \(|b|>1\), then for any \(\gamma \in\mathbb R\), \[ \lim_{x\to\infty}\frac{1}{x}\big|\big\{n\leq x: \gcd(n,b)=1 \text{\;and\;}\frac{\nu(l_b(n))-\frac{1}{2}(\log\log n)^2}{ \frac{1}{\sqrt{3}}(\log\log n)^{3/2}}\leq \gamma\big\}\big|= \frac{\varphi(b)}{|b|}G(\gamma) \] where for \(m\in{\mathbb N}\), \(\nu(m)\) is the number of distinct prime divisors of \(m\), \(\varphi\) is the Euler's function, and \(G(\gamma)=\frac{1}{\sqrt{2\pi}}\int_{-\infty}^{\gamma} e^{-t^2/2}\,dt\). \textit{W. Kuo} and \textit{Y.-R. Liu} proved an analogous conjecture for the Carlitz module in [Trans. Am. Math. Soc. 361, No. 9, 4519--4539 (2009; Zbl 1181.11050)]. The objective of this paper is to generalize this question to some Drinfeld modules of rank one. For this end, consider \(k\) a global function field over \({\mathbb F}_q\) with \(q\neq 2\) containing a place of degree one. For a fixed place \(\infty\) of degree one, let \(A\) be the ring of elements of \(K\) regular away from \(\infty\). Let \(\psi\) be a sign-normalized \(A\)-Drinfeld module over \({\mathcal O}\), the integral closure of \(A\) in \(H_A\), the Hilbert class field of \(A\). For a fixed non-zero element \(\alpha\) of \({\mathcal O}\) and an ideal \({\mathfrak m}\) of \({\mathcal O}\), let \(f_{\alpha}({\mathfrak m})=\{f\in A: \psi_f(\alpha)\equiv 0\bmod {\mathfrak m}\}\). For an ideal \({\mathfrak n}\) of \(A\) let \(\omega({\mathfrak n})\) be the number of distinct prime ideals factors of \({\mathfrak m}\). The first main result (Theorem 1) is: \[ \sum_{\deg {\mathfrak P}=x} (\omega(f_{\alpha}({\mathfrak P}))-\log x)^2 \ll \pi_{H_A}(x)\log x \] where \({\mathfrak P}\) runs through all prime ideals of degree \(x\), and \(\pi_{H_A}(x)\) is the number of finite places of \(H_A\) of degree \(x\). A corollary of this result, is that for any \(\varepsilon>0\) \[ \big|\{\deg {\mathfrak P}=x: |\omega(f_{\alpha}({\mathfrak P}))-\log\deg{\mathfrak P}| >\varepsilon \log\deg {\mathfrak P}\}\big|=o(\pi_{H_A}(x)). \] The second main result, Theorem 2, is that for a prime ideal \({\mathfrak P}\) of \({\mathcal O}\), then for \(\gamma \in {\mathbb R}\) and \(x\in{\mathbb N}\), we have \[ \lim_{x\to\infty}\frac{1}{\pi_{H_A(x)}}\big|\big\{\deg {\mathfrak P}=x: \frac{\omega(f_{\alpha}({\mathfrak P}))-\log\deg {\mathfrak P}}{\sqrt{\log\deg{\mathfrak P}}} \leq \gamma\big\}\big|=G(\gamma). \] The main tools in the proof of Theorem 2 is the study of Kummer's theory for sign-normalized rank one Drinfeld modules and apply Chebotarev density theorem to estimate error terms. Next, the authors establish (Theorem 3) \[ \sum_{\deg{\mathfrak m}=x}\big(\omega(f_{\alpha}({\mathfrak m}))- \frac{1}{2}(\log x)^2\big)\ll q^x(\log x)^3 \] where the sum runs through all ideals \({\mathfrak m}\). The last main result, Theorem 4, is an analogue of Erdős-Pomerance's conjecture for rank one Drinfeld modules: \[ \lim_{x\to\infty}\frac{1}{N(x)}\big|\big\{\deg{\mathfrak m}=x: \frac{ \omega(f_{\alpha}({\mathfrak m}))-\frac{1}{2}(\log\deg{\mathfrak m})^2}{ \frac{1}{\sqrt{3}}(\log\deg{\mathfrak m})^{3/2}}\leq \gamma\big\}\big|=G(\gamma) \] where \(N(x)\) is the number of ideals of degree \(x\). After some preliminaries in Section 2, the authors prove Theorems 1 and 2 in Section 3 and Theorems 3 and 4 in last section.