Reciprocal relations between cyclotomic fields (Q982535)

For any positive integer \(n\) let \(\zeta_n\) be a primitive \(n\)-th root of unity, \(\phi_n\) the \(n\)-th cyclotomic polynomial, i.e. the irreducible polynomial of \(\zeta_n\) over \(\mathbb{Q}\), and let \(K_n\) be the cyclotomic field \(\mathbb{Q}(\zeta_n)\) with ring of integers \(\mathfrak{o}_n\). Let \(N_n\) denote the norm map in the cyclotomic extension \(K_n / \mathbb{Q}\). The paper under review describes a reciprocity relation between the prime ideal factorization of \(\phi_n(c - \zeta_m)\) in \(\mathfrak{o}_m\) and \(\phi_m(c - \zeta_n)\) in \(\mathfrak{o}_n\), where \(n\) and \(m\) are positive integers and \(c \in \mathbb{Z}\) such that \(N_m(\phi_n(c - \zeta_m))\) is relatively prime to \(mn\); note that the latter condition is dual in \(n\) and \(m\), since one has an equality (Theorem 1) \[ N_m(\phi_n(c - \zeta_m)) = N_n(\phi_m(c - \zeta_n)). \] More precisely, let \(|N_m(\phi_n(c - \zeta_m))| = \prod_{i=1}^r l_i^{h_i}\) with distinct prime numbers \(l_i\) and positive integers \(h_i\). Then for \(1 \leq i \leq r\), the orders of \(l_i\) modulo \(m\) and \(l_i\) modulo \(n\) agree (Corollary 2 to Theorem 2) and divide \(h_i\) (Corollary 2 to Proposition 7); let \(m_i \in \mathbb{Z}\) be the quotient. Furthermore, let \(\mathfrak{l}_i\) (resp. \(\mathfrak{l}'_i\)) be a prime ideal of \(\mathfrak{o}_m\) (resp. \(\mathfrak{o}_n\)) dividing \(l_i\) and \(\phi_n(c - \zeta_m)\) (resp. \(l_i\) and \(\phi_m(c - \zeta_n)\)). Then one has (Theorem 4) \[ \phi_n(c - \zeta_m) \mathfrak{o}_m = \prod_{i=1}^r \mathfrak{l}_{i,1} \cdot \dots \cdot \mathfrak{l}_{i,m_i}, \] \[ \phi_m(c - \zeta_n) \mathfrak{o}_n = \prod_{i=1}^r \mathfrak{l}'_{i,1} \cdot \dots \cdot \mathfrak{l}'_{i,m_i}, \] where \(\mathfrak{l}_{i,1}, \dots, \mathfrak{l}_{i,m_i}\) (resp. \(\mathfrak{l}'_{i,1}, \dots, \mathfrak{l}'_{i,m_i}\)) are \(m_i\), not necessarily distinct, prime ideal conjugates of \(\mathfrak{l}_i\) (resp. \(\mathfrak{l}'_i\)). In particular (Theorem 5), \(\phi_n(c - \zeta_m) \mathfrak{o}_m\) is prime if and only if \(\phi_m(c - \zeta_n) \mathfrak{o}_n\) is prime. Also the following result (Theorem 3) might be useful: Let \(c \in \mathbb{Z}\) such that \(\phi_n(c)\) is relatively prime to \(n\). Then \(\phi_n(c)\) is a rational prime if and only if \((c - \zeta_n) \mathfrak{o}_n\) is prime; in this case, \(\phi_n(c)\) splits completely in \(K_n / \mathbb{Q}\). Note that this in particular applies to Mersenne numbers \(M_p = 2^p-1 = \phi_p(2)\), where \(p\) is a prime, and to Fermat numbers \(F_k = 2^{2^k}+1 = \phi_{2^{k+1}}(2)\), where \(k\) is a positive integer. The paper is well written and requires only basic knowledge on cyclotomic fields.