Kronecker's solution of the Pell equation on a computer (Q2500794)

The computation of the unit group of an algebraic number field is in general closely connected with the computation of the class number, and in most implementations, the class number \(h\) and the regulator \(R\) of a number field are determined simultaneously. One reason for this is the fact that the class number formula expresses the product \(hR\) as the residue of the Dedekind zeta function at \(s = 1\), and no one knows how to separate these siamese twins. Already Dirichlet and Jacobi observed that these class number formulas can be used to compute nontrivial solutions of the Pell equation. In this article, the author suggests using Kronecker's limit formula for computing units, and shows that this approach works in practice for discriminants of modest size. This limit formula expresses a rational multiple of \(\log \varepsilon_t\) (here \(\varepsilon_t\) is the fundamental unit of \(\mathbb Q(\sqrt{t}\,)\)) as a sum over the ideal classes of a quadratic field \(\mathbb Q(\sqrt{tu}\,)\) for a suitable negative integer \(u\); the summands are essentially logarithms of Dedekind's eta function evaluated at points in the upper half plane determined by the ideal classes. The article begins with a brief introduction to algebraic number theory, then discusses the genus theory of quadratic number fields and Dirichlet \(L\)-series, and finally the limit formulas. It was written for non-specialists and is warmly recommended for everyone who hasn't seen Kronecker's limit formula before.