学术文献库

We state the Brumer-Stark conjecture and motivate it from two perspectives. Stark's perspective arose in his attempts to generalize the classical Dirichlet class number formula for the leading term of the Dedekind zeta function at $s=1$ (equivalently, $s=0$). Brumer's perspective arose by generalizing Stickelberger's work regarding the factorization of Gauss sums and the annihilation of class groups of cyclotomic fields. These viewpoints were synthesized by Tate, who stated the Brumer-Stark conjecture in its current form. The conjecture considers a totally real field $F$ and a finite abelian CM extension $H/F$. It states the existence of $p$-units in $H$ whose valuations at places above $p$ are related to the special values of the $L$-functions of the extension $H/F$ at $s=0$. Essentially equivalently, the conjecture states that a Stickelberger element associated to $H/F$ annihilates the (appropriately smoothed) class group of $H$. This conjecture has been refined by many authors in multiple directions. We conclude by stating our results toward these various conjectures and summarizing the proofs. In particular, we prove the Brumer-Stark conjecture, Rubin's higher rank version, and Kurihara's conjecture, all "away from 2." We also prove strong partial results toward Gross's conjecture and the exact $p$-adic analytic formula for Brumer-Stark units. The key technique involved in the proofs is Ribet's method. We demonstrate congruences between Hilbert modular Eisenstein series and cusp forms, and use the associated Galois representations to construct Galois cohomology classes. These cohomology classes are interpreted in terms of Ritter-Weiss modules, from which results on class groups may be deduced.