Arguments for the Stark units and periods of Eisenstein series (Q1005881)

Let \(K/k\) be a finite abelian extension, \(S\) a finite set of places of \(k\) (containing all archimedean and ramified ones), \(S_K\) a finite extension of \(S\), \(G\) the Galois group and \(\chi\) a character of \(G\). Under the hypothesis that \(S\) contains at least two, and among them one totally split place the abelian Stark conjecture (cf. section IV.2 of [\textit{J. Tate}, Les conjectures de Stark sur les fonctions \(L\) d'Artin en \(s=0\). Prog. Math. 47. Boston etc.: Birkhäuser (1984; Zbl 0545.12009)]) predicts the existence of an \(S_K\)-unit in \(K^{\times}\) the logarithm of the \textit{absolute value} of which can be identified through the derivative at \(0\) of the partial Artin \(L\)-function associated to \(\chi\). The point of the existence of such being its relevance to Hilbert's twelfth problem. The conjecture is known to be trivially true if \(S\) contains two totally split places (cf. section IV.3 of \textit{Tate}), which leaves the two cases of totally real and what the authors of the present paper call \textit{almost totally real} extensions, i.e. quadratic extensions of a totally real field with exactly one complex place. For the totally real case Stark (see \textit{H. Stark} [Adv. Math. 22, 64--84 (1976; Zbl 0348.12017)]) explicitly describes class fields of \(k\) by virtue of these \(S_K\)-units. The paper under review presents a strengthening of the abelian Stark conjecture in the ATR case. It is formulated as Conjecture 4.1 and identifies the logarithm -- lifted to \(\mathbb{C}/2\pi i\mathbb{Z}\) -- of the \(S_K\)-unit up to correcting factors through the period of an Eisenstein form. To give an idea of how this is accomplished start from the situation of an imaginary quadratic extension \(K|\mathbb{Q}\). Let \(\eta\) be the Dedekind eta function, \(\tau_1,\tau_2\) two \(K\)-rational points in the upper half plane \(\mathcal{H}\) and consider the modular unit \(u(\tau_1,\tau_2)=\eta(\tau_2)\eta(\tau_1)^{-1}\). The essential observation is that one can describe the logarithm of such in terms of the corresponding period of the classical Eisenstein series \(E_2\) for \(SL_2(\mathbb{Z})\), explicitly \[ \log u(\tau_1,\tau_2)=-i\pi\int_{\tau_1}^{\tau_2}E_2(z)\,dz \mod2\pi i\mathbb{Z}. \] While the modular units themselves do not permit a straightforward generalization to fields of higher degree, the periods of Eisenstein series do. Now, in generalizing the latter to the situation of a totally real field \(k\) of degree \(n>1\) over the rationals, \(E_2\) gets replaced by an Eisenstein form associated with the straightforward generalization of it in the sense of Hilbert modular forms for \(\text{SL}_2(\mathcal{O}_k)\); and the integral is taken over any differentiable chain the boundary of which is given by a certain \(n-1\)-cycle on \(\text{SL}_2(\mathcal{O}_k)\backslash\mathcal{H}^n\) depending on the algebraic data the set-up rests on. Working out this idea and rendering precise the relations hinted at here the authors come up with the remarkably precise formula of Conjecture 4.1., a feature they demonstrate by describing an algorithm for \(n=2\) and subsequently making use of it to provide a fair number of numerical examples. At any rate, the reader is advised to take a look at the paper's introduction, which makes for a succinct presentation of the problems in question.