English

Eigenvarieties and invariant norms: Towards p-adic Langlands for U(n)

Number Theory 2016-01-20 v1 Representation Theory

Abstract

We give a proof of the Breuil-Schneider conjecture in a large number of cases, which complement the indecomposable case, which we dealt with earlier in [Sor]. In some sense, only the Steinberg representation lies at the intersection of the two approaches. In this paper, we view the conjecture from a broader global perspective. If U/FU_{/F} is any definite unitary group, which is an inner form of \GL(n)\GL(n) over \K\K, we point out how the eigenvariety \X(Kp)\X(K^p) parametrizes a global pp-adic Langlands correspondence between certain nn-dimensional pp-adic semisimple representations ρ\rho of \Gal(\Qˉ\K)\Gal(\bar{\Q}|\K) (or what amounts to the same, pseudo-representations) and certain Banach-Hecke modules B\mathcal{B} with an admissible unitary action of U(F\Qp)U(F\otimes \Q_p), when pp splits. We express the locally regular-algebraic vectors of B\mathcal{B} in terms of the Breuil-Schneider representation of ρ\rho. Upon completion, this produces a candidate for the pp-adic local Langlands correspondence in this context. As an application, we give a weak form of local-global compatibility in the crystalline case, showing that the Banach space representations Bξ,ζB_{\xi,\zeta} of Schneider-Teitelbaum [ScTe] fit the picture as predicted. There is a compatible global mod pp (semisimple) Langlands correspondence parametrized by \X(Kp)\X(K^p). We introduce a natural notion of refined Serre weights, and link them to the existence of crystalline lifts of prescribed Hodge type and Frobenius eigenvalues. At the end, we give a rough candidate for a local mod pp correspondence, formulate a local-global compatibility conjecture, and explain how it implies the conjectural Ihara lemma in [CHT].

Keywords

Cite

@article{arxiv.1208.0703,
  title  = {Eigenvarieties and invariant norms: Towards p-adic Langlands for U(n)},
  author = {Claus M. Sorensen},
  journal= {arXiv preprint arXiv:1208.0703},
  year   = {2016}
}

Comments

Comments and suggestions are very welcome