The smallest inert prime in a cyclic number field of prime degree (Q2434623)

For an odd prime number \(p\), let \(n_2(p)\) denote the least quadratic nonresidue modulo \(p\). A classical result of \textit{P. Erdős} [Mat. Lapok 12, 10--17, 161--169 (1961; Zbl 0154.29403)] shows that \(n_2(p)\) has a finite mean value: if \(p_k\) denotes the \(k\)-th prime number, then Erdős showed that \(\frac1{\pi(x)} \sum_{2 < p \leq x} n_2(p)\) tends to the quickly converging series \(\sum_{k \geq 1} 2^{-k}p_k\) as \(x \to \infty\). In this article, the above result is generalized as follows. For a cyclic number field \(K\) of prime degree \(\ell \geq 3\), let \(n_K\) denote the smallest inert prime and \(f_K\) its conductor. Let \(N(x)\) denote the number of cyclic extensions of the rationals with prime degree \(\ell\) and conductor \(f_K \leq x\); then \(\frac1{N(x)} \sum_{f_K \leq x} n_K\) has a limit as \(x \to \infty\), which can be represented explicitly as a somewhat technical infinite expression. A similar result holds (assuming the generalized Riemann Hypothesis) for the smallest totally split prime \(r_K\).