Test vectors for finite periods and base change

Abstract

Let E/F be a quadratic extension of finite fields. By a result of Gow, an irreducible representation π of G = GLn(E) has at most one non-zero H-invariant vector, up to multiplication by scalars, when H is GLn(F) or U(n,E/F). If π does have an H-invariant vector it is said to be H-distinguished. It is known, from the work of Gow, that H-distinction is characterized by base change from U(n,E/F), due to Kawanaka, when H is GLn(F) (resp. from GLn(F), due to Shintani, when H is U(n,E/F)). Assuming π is generic and H-distinguished, we give an explicit description of the H-invariant vector in terms of the Bessel function of π. Let be a non-degenerate character of NG/NH and let Bπ, be the (normalized) Bessel function of π on the -Whittaker model. For the H-average \[Wπ, = 1|H| Σh∈ H π(h) Bπ,\] of the Bessel function, we prove that \[Wπ,(In) = dim dim π · | GLn(E)|| GLn(F)| | U(n,E/F)|,\] where is the representation of U(n,E/F) (resp. GLn(F)) that base changes to π when H is GLn(F) (resp. U(n,E/F)). As an application we classify the members of a generic L-packet of SLn(E) that admit invariant vectors for SLn(F). Finally we prove a p-adic analogue of our result for square-integrable representations in terms of formal degrees by employing the formal degree conjecture of Hiraga-Ichino-Ikeda hii08.

0

Turn this paper into a lesson

ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.

Discussion (0)

Sign in to join the discussion.

Loading comments…