Ihara's Lemma for $\mathrm{GL}_d$: the limit case
Pascal Boyer
TL;DR
The paper advances the generalization of Ihara's lemma to $\mathrm{GL}_d$ in the limit case, a key step for non-minimal $R=\mathbb{T}$ theorems in higher rank. It develops a geometric framework using KHT-Shimura varieties, nearby cycles, and Newton filtrations, and leverages level-raising results (BLGGT and Gee–Annalen) to control non-generic contributions. The main result confirms Clozel–Harris–Taylor conjectures in the limit setting, establishing genericity for all irreducible $\mathrm{GL}_d(F_v)$-submodules in the localized middle cohomology and building a robust lattice-theoretic viewpoint. This work deepens the connection between automorphic forms, Galois representations, and geometric methods, with potential implications for $R=\mathbb{T}$-type results and arithmetic applications such as special values of adjoint L-functions and local-global compatibility phenomena.
Abstract
Clozel, Harris and Taylor proposed conjectural generalizations of the classical Ihara's lemma for $\mathrm{GL}_2$, to higher dimensional similitude groups. We prove these conjectures in the so called limit case, which after base change is the essential one, under any hypothesis allowing level raising.
