Multiplicative orders of Gauss periods and the arithmetic of real quadratic fields (Q2031650)

Let \(q\) be a prime number, \(n\) be a positive integer, and \(\mathbb F_{q^n}\) be the finite field with \(q^n\) elements. Suppose that \(p = 2n+1\) is a prime, let \(\zeta \in \mathbb F_{q^{2n}}\) be a primitive \(p^{\mathrm{th}}\) root of unity, and let \(\alpha = \zeta + \zeta^{-1}\). This \(\alpha\) is known as Gauss period of type \((n,2)\) over \(\mathbb F_q\). If \(\alpha\) is a primitive element, then it generates a normal basis for \(\mathbb F_{q^n}\), thus it is useful to know more about the multiplicative order ord\((\alpha)\) of \(\alpha\) in \(\mathbb F_{q^n}^*\). Some lower bounds are already known (see [\textit{O. Ahmadi} et al., Int. J. Number Theory 6, No. 4, 877--882 (2010; Zbl 1201.11110); \textit{J. von zur Gathen} and \textit{I. Shparlinski}, Appl. Algebra Eng. Commun. Comput. 9, No. 1, 15--24 (1998; Zbl 0931.11057); \textit{S. Gao} and \textit{S. A. Vanstone}, Math. Comput. 64, No. 211, 1227--1233 (1995; Zbl 0868.11059); \textit{R. Popovych}, Finite Fields Appl. 18, No. 4, 700--710 (2012; Zbl 1251.11086); \textit{R. Popovych}, Ukr. Math. J. 66, No. 6, 916--927 (2014; Zbl 1319.11091)] for more). In this paper, the authors obtain upper bounds for ord\((\alpha)\) using some divisibility conditions. More precisely, let ind\((\alpha) = \frac{q^n-1}{\mathrm{ord}(\alpha)} \in \mathbb Z_+^*\) be the index of \(\alpha\), which is usually small. It is shown that small prime factors of this index can be detected in the arithmetic of the quadratic field \(K = \mathbb Q(\sqrt{p})\). Let \(\varepsilon_p\), \(h_p\), and \(\mathcal O_K\) be the fundamental unit, the class number and the ring of integers of \(K\), respectively. When \(q\) is inert in \(K/\mathbb Q\), let ind\((\varepsilon_p \mod q)\) be the multiplicative index of \((\varepsilon_p \mod {q \mathcal O_K})\) in \((\mathcal O_K/ q\mathcal O_K)^* \simeq \mathbb F_{q^2}^*\). The main result can be stated as follows. Theorem 1.1. Let \(p \equiv 5 \mod 8\) be a prime. Suppose that \((\mathbb Z/ p\mathbb Z)^* = \langle -1,q \rangle\) and let \(\zeta \in \mathbb F_{q^{p-1}}\) be a primitive \(p^{\mathrm{th}}\) root of unity. Then \[ \gcd(\mathrm{ind}(\zeta + \zeta^{-1}), q^2-1) =\mathrm{ind}(\varepsilon_p^{h_p} \mod q). \] They also remark that ord\((\zeta+1)\) when \(q=2\) determine the periods of Ducci sequences (see [\textit{F. Breuer} et al., Finite Fields Appl. 13, No. 2, 293--304 (2007; Zbl 1152.11005); \textit{F. Breuer}, Bull. Aust. Math. Soc. 100, No. 2, 201--205 (2019; Zbl 1459.11077); \textit{F. Breuer} and \textit{I. E. Shparlinski}, Bull. Aust. Math. Soc. 102, No. 1, 31--38 (2020; Zbl 1450.11018)]). Moreover, if \(q\) is a primitive root modulo \(p\), then \(p \nmid q^n-1\), hence \(p \nmid\mathrm{ord}(\zeta + \zeta^{-1})\). It implies that ord\((\zeta+1) = p \mathrm{ord}(\zeta + \zeta^{-1})\). As a consequence of Theorem 1.1., they get the following result. Corollary 1.2. Suppose that \(p \equiv 5 \pmod 8\) is prime and that \(2\) is a primitive root modulo \(p\). Let \(\zeta \in \mathbb F_{2^{p-1}}\) be a primitive \(p^{\mathrm{th}}\) root of unity. Then the following are equivalent: \begin{itemize} \item[(1)] ind\((\zeta+1)\) is divisible by \(3\). \item[(2)] ind\((\zeta+\zeta^{-1})\) is divisible by \(3\). \item[(3)] The eventual period \(P\) of any Ducci sequence in \(\mathbb Z^p\) formed by iterating the map \[ D: \mathbb Z^p \to \mathbb Z^p; \quad D(x_1,x_2,\dots,x_p) = (|x_1-x_2|, |x_2-x_3|, \dots, |x_n-x_1|) \] satisfies \(P \mid \frac{1}{3} p(2^{(p-1)/2} - 1)\). \item[(4)] \((i)\) \(\varepsilon_p \equiv 1 \mod 2\mathcal O_K\) or \((ii)\) \(3 \mid h_p\). \end{itemize} Some of the implications above were already known, such as \(4(i) \Rightarrow (3)\) (see [Zbl 1459.11077]) and \((1) \Leftrightarrow (3)\) (see [Zbl 1152.11005]). Finally, they also discuss the heuristic of how often a given prime prime divides \(\text{ord}(\alpha)\).