Kudla-Millson lift of toric cycles and restriction of Hilbert modular forms
Romain Branchereau
TL;DR
The paper develops a Kudla-Millson theta lift framework for cycles attached to anisotropic maximal $\\mathbb{Q}$-tori in an orthogonal group of signature $(p,q)$ with $p\\ge q>0$, showing that the lift of toric cycles is always a cusp form and arises as a diagonal restriction of a parallel-weight Hilbert modular form. Using a seesaw argument and restriction-of-scalars, the authors express these diagonal restrictions as products of diagonally restricted Hilbert modular forms, leading to a dimension formula relating the space of cusp forms to the spans of toric and special cycles. They provide explicit computations of the Hilbert modular forms in the torus setting, prove cusp-ness via positivity conditions, and illustrate the constructions with biquadratic and more general étale-algebra examples. The results connect geometric intersection theory on locally symmetric spaces with arithmetic information encoded in Hilbert modular forms, enabling a precise comparison of the spans of diagonal restrictions and toric cycles. Overall, the work clarifies how toric cycles contribute to the cusp-spectrum via KM lifts and offers concrete mechanisms to compute and compare their contributions across S-arithmetic settings.
Abstract
Let $V$ be quadratic space of even dimension and of signature $(p, q)$ with $p \geq q > 0$. We show that the Kudla-Millson lift of toric cycles - attached to algebraic tori - is a cusp form that is the diagonal restriction of a Hilbert modular form of parallel weight one. We deduce a formula relating the dimension of the span of such diagonal restrictions and the dimension of the span of toric and special cycles.
