A field theoretic proof of Hermite's theorem for function fields (Q500858)

For a number field \(K\) we have from Minkowski's theorem that \(|d_{ K/{\mathbb Q}}|\geq \frac{n^n}{n!}\big(\frac{\pi}{4}\big)^{n/2}\), where \(n=[K:{\mathbb Q}]\) and \(d_{K/{\mathbb Q}}\) is the discriminant. Hermite's classical theorem establishes that there exist only finitely many number fields with bounded discriminant. Thus, to prove Hermite's theorem, it suffices to show that there are only finitely many number fields of bounded discriminant and degree. There is a proof of Taguchi of the analogue of Hermite's theorem for global function fields. In this paper the author gives a more elementary proof using field theoretic techniques, inspired by a lemma of Faltings that compares semisimple \(\ell\)-adic Galois representations. Since constant extensions are unramified and inseparable extensions might have all the primes ramified, Hermite's theorem is given for geometric separable extensions. The main result is: For any \(B>0\), there exist only finitely many finite, separable, geometric extensions \(K/k\) with \(\deg_k(d_{K/k})<B\), where \(k={\mathbb F}(T)\), \({\mathbb F}\) is a finite field, and \(d_{K/k}\) is the discriminant of \(K/k\). Using the Riemann-Hurwitz genus formula, the proof of the main result is reduced to prove the weak form of Hermite's theorem: For any integers \(N,C>0\), there exist at most finitely many Galois extensions of global function fields \(K/k\) with \( [K:k]\leq N\) and \(\deg_k d_{K/k}<C\). To prove the weak form of Hermite's theorem, the author shows that if it were false, then it can be produced infinitely many geometric extensions with the same Galois group and bounded discriminant, and a uniform bound \(B^{\prime}\) such that every conjugacy class of the Galois group of norm \(> B^{\prime}\) is realizable. Then, by the Chebotarev density theorem, geometric extensions with the same Galois group and bounded discriminant, are determined by finitely many Frobenius conjugacy classes leading to a contradiction.