The average least quadratic nonresidue modulo \(m\) and other variations on a theme of Erdős (Q413420)

For each \(m\geq 3\), let \(n_2(m)\) denote the least quadratic nonresidue modulo \(m\). The author shows that \(n_2(m)\) has a mean value and determines it. For each prime \(p\), let \(G(p)\) denote the least positive integer \(n\) such that the subgroup generated by \(\{1,2,\dots,n\}\) is all of \((\mathbb Z/p\mathbb Z)^*\). The author shows that \(G(p)\) has a finite mean, on assuming the Generalized Riemann Hypothesis. Let \(K\) be a quadratic number field. Denote by \(n_K\), \(r_K\), the smallest rational prime which is inert, respectively split in \(K\). The author shows that if we order the quadratic fields by the absolute value of their discriminant, both \(n_K\) and \(r_K\) have the same mean value. The averages that occur are relatively complicated infinite prime sums, that the author nevertheless manages to compute with high decimal precision. In the proofs estimates for the number of integers without large prime divisors, the Brun-Titchmarsh inequality and the large sieve for character sums play an important role.