On the Minkowski unit in function fields (Q1262900): Difference between revisions

Let K be a cyclic number field of degree \(\ell\) where \(\ell\) is an odd prime; thus K is totally-real. Let \(U_ K\) be the unit group of K. By assumption, one has \(U_ K=\{\pm 1\}\times V_ K\), where \(V_ K\) is the group of units of norm 1 and is free of rank \(\ell -1\). Let \(\sigma\) be a generator of Gal(K/\({\mathbb{Q}})\) and let \(N=1+\sigma +...+\sigma^{\ell -1}\). Let \(\epsilon \in V_ K\). By definition, \(N(\epsilon)=1\). Let \(R={\mathbb{Z}}[G]/(N)\). By definition \(V_ K\) is an R- module. Suppose now that \(V_ K\) is a cyclic R-module and that \(\epsilon\) is a generator of \(V_ K\) (as R-module). We call \(\epsilon\) a Minkowski unit. In 1969, A. Brumer gave a criterion for a Minkowski unit to exist in terms of an auxiliary cyclotomic number field L. (Basically, L determines the structure of R). In the paper being reviewed, the authors give a version of Brumer's theorem for cyclotomic function fields. The result parallels the classical case, but is remarkable in that the structure of a number field (i.e., the analog of L above) is used to determine the structure of units in a function field.