Twisted zeta-functions and Abelian Stark conjectures (Q1609531): Difference between revisions

This article proposes a new way of thinking about L-values at \(s=1\). It is a fact that the analytic class number formula appears simpler in its version at \(s=0\) than at \(s=1\), when one looks at the constants involved. (The two versions are equivalent via the functional equation.) Stark's conjectures which connect regulators and L-values in a much more precise way were also formulated at \(s=1\) in the beginning, but as time went by there was a remarkable shift towards the \(s=0\) formulation. All this is lucidly explained in the author's introduction. The pretty idea that materializes in the paper under review is to ``twist'' the L-functions in such a way that the value at 1 connects in a tidier way to arithmetical data than it does without the twisting. It seems that a main point is getting rid of certain Gauss sums. Oversimplifying things one can say that the twist just amounts to replacing the multiplicative character used to make an Abelian Artin L-function by an additive character. Since Gauss sums somehow serve as transfer from additive to multiplicative characters, it is quite plausible that this switch eliminates them. Also, to quote an example, the twisted Dirichlet series with an additive character \(\sum_{n>0} \xi^n n^{-s}\) directly gives, upon letting \(s\to 1\), the logarithm of the cyclotomic number \(1-\xi\), where \(\xi\) is a root of unity, so the link to regulators is as tidy as one can ask for. The idea is implemented by defining twisted zeta-functions in great generality: first those that correspond to partial zeta-functions in the usual sense, and then (by assembly) the general group-ring valued version. All this is explained very carefully. Section 3 enunciates the main results on these new functions: behaviour under change of the sets \(S\) and \(T\) (which play the same roles as for the standard L-functions: control of ramification, resp.~so-called T-modification), and the connection with the standard functions. Proofs are deferred to the final section. Section 4 recapitulates Rubin's higher rank conjectures in the Abelian case. This is done leisurely and with precision, so the paper also offers a good entry to this subject to people not so conversant with it. Section 5 has a look at Rubin's conjectures in the light of the new (twisted) L-functions and also reconsiders their definition in the minus part. This is foundational work which may very well turn out useful in future work, for instance on Rubin's conjectures. There is another motivation which is mentioned briefly in the introduction: in the \(p\)-adic version, there is no functional equation, so it is essential to have a good grasp of the cases \(s=0\) and \(s=1\) independently, and it turns out that the constructions of this paper have reasonable and interesting \(p\)-adic analogs. We refer to a paper of \textit{D. Solomon} in [Ann. Inst. Fourier 52, 379-417 (2002; Zbl 1039.11081)], and to a joint paper of \textit{X.-F. Roblot} and \textit{D. Solomon} [J. Number Theory 107, No. 1, 168--206 (2004; Zbl 1052.11076)] which does numerical verifications in the \(p\)-adic case.