Twisted character formula for toral supercuspidal representations
Masao Oi
TL;DR
This work extends the Adler–Spice and DeBacker–Spice framework to twisted endoscopy, deriving an explicit formula for twisted Harish-Chandra characters of toral supercuspidal representations in the twisted setting. By developing twisted normal $r$-approximations, twisted Heisenberg–Weil representations, and a twisted ADS-type descent, the authors express the twisted character as a sum over appropriate double cosets with head–tail separation and descent to centralizers. The results leverage Kaletha’s toral data, the descent of root data, and Gérardin-type formulas to produce a computable, case-sensitive character formula applicable to quasi-split groups with involutions. The findings facilitate twisted endoscopy applications and contribute to explicit local Langlands correspondences for classical groups in the twisted context.
Abstract
We establish an explicit formula for twisted Harish-Chandra characters of toral supercuspidal representations of p-adic reductive groups under several technical assumptions. Our setup especially includes the case of a quasi-split group equipped with a involution.
