Overconvergent modular forms are highest weight vectors in the Hodge-Tate weight zero part of completed cohomology

TL;DR

This paper constructs a pairing between overconvergent modular forms and local cohomology, revealing their structure as highest weight vectors in completed cohomology and confirming a conjecture on Galois representations.

Contribution

It introduces a new cup product pairing linking overconvergent modular forms with local cohomology, providing geometric insights and confirming a Gouvea conjecture with a simple, elegant proof.

Findings

Pairing is non-trivial for overconvergent forms of infinitesimal weight not equal to 1.

Describes half of the locally algebraic vectors in completed cohomology.

Confirms Gouvea's conjecture on Hodge-Tate-Sen weights of Galois representations.

Abstract

We construct a $(\mathfrak{gl}_2, B(\mathbb{Q}_p))$ and Hecke-equivariant cup product pairing between overconvergent modular forms and the local cohomology at $0$ of a sheaf on $\mathbb{P}^1$, landing in the compactly supported completed $\mathbb{C}_p$-cohomology of the modular curve. The local cohomology group is a highest-weight Verma module, and the cup product is non-trivial on a highest weight vector for any overconvergent modular form of infinitesimal weight not equal to $1$. For classical weight $k\geq 2$, the Verma has an algebraic quotient $H^1(\mathbb{P}^1, \mathcal{O}(-k))$, and on classical forms the pairing factors through this quotient, giving a geometric description of "half" of the locally algebraic vectors in completed cohomology; the other half is described by a pairing with the roles of $H^1$ and $H^0$ reversed between the modular curve and $\mathbb{P}^1$. Under minor assumptions, we deduce a conjecture of Gouvea on the Hodge-Tate-Sen weights of Galois representations attached to overconvergent modular forms. Our main results are essentially a strict subset of those obtained independently by Lue Pan in arXiv:2008.07099, but the perspective here is different and the proofs are short and use simple tools: a Mayer-Vietoris cover, a cup product, and a boundary map in group cohomology.