No label defined (Q4345950)

Even if a number-theoretic conjecture remains open for a long time, an analoguous conjecture in the arithmetic theory of function fields sometimes turns out to be provable, although by no means trivial. The paper under review offers another beautiful example of this phenomenon. Originally, Artin's conjecture says that given an integer \(a\neq\pm 1\) which is not a perfect square, there should be infinitely many primes \(p\) such that \(a\) is a primitive root mod \(p\), and these primes should have positive density in the set of all primes. This was proved 40 years later by \textit{C. Hooley} [J. Reine Angew. Math. 225, 209-220 (1967; Zbl 0221.10048)] under the assumption of some generalized Riemann hypothesis, but an unconditional proof still has not been found. Now replace \(\mathbb Z\) by the polynomial ring \(A=\mathbb F_q[t]\). In many aspects (and especially from the class field theoretic point of view) the best function field analog of the multiplicative group is the Carlitz module, a new \(A\)-module structure on the additive group of \(A\) where \(t\) acts as \(X^q+tX\). For every prime ideal \({\mathfrak P}\) of \(A\) this induces a new \(A\)-module structure \(C(A/{\mathfrak P})\) on \(A/{\mathfrak P}\). The analog of Artin's conjecture now says that given a nonzero \(a\in A\) there should be infinitely many \({\mathfrak P}\) such that \(\bar{a}=a+{\mathfrak P}\) is a generator of \(C(A/{\mathfrak P})\). This problem is translated into a question concerning the splitting of \({\mathfrak P}\) in so-called Kummer-Carlitz extensions of \(\mathbb F_q(t)\) and then solved with methods from analytic number theory. For \(q\neq 2\) the author proves that the number of such \({\mathfrak P}\) with \(\deg({\mathfrak P})=x\) is \(\delta_a\cdot\frac{q^x}{x}+o(\frac{q^x}{x})\) with a constant \(\delta_a>0\). For \(q=2\) there are some conditions on \(a\) corresponding to the conditions in the classical setting.