A combinatorial Li-Yau inequality and rational points on curves

TL;DR

This paper introduces a new method linking graph theory and nonarchimedean geometry to establish lower bounds on the gonality of curves, with applications to Drinfeld modular curves and elliptic curves over function fields.

Contribution

It develops a combinatorial Li-Yau inequality for graphs and applies it to bound gonality and modular degrees in the context of nonarchimedean curves.

Findings

Lower bounds for gonality of nonarchimedean curves using graph eigenvalues.

A Li-Yau type inequality for graphs relating eigenvalues and volume.

Applications to uniform boundedness of torsion and modular degree bounds.

Abstract

We present a method to control gonality of nonarchimedean curves based on graph theory. Let k denote a complete nonarchimedean valued field. We first prove a lower bound for the gonality of a curve over the algebraic closure of k in terms of the minimal degree of a class of graph maps, namely: one should minimize over all so-called finite harmonic graph morphisms to trees, that originate from any refinement of the dual graph of the stable model of the curve. Next comes our main result: we prove a lower bound for the degree of such a graph morphism in terms of the first eigenvalue of the Laplacian and some "volume" of the original graph; this can be seen as a substitute for graphs of the Li-Yau inequality from differential geometry, although we also prove that the strict analogue of the original inequality fails for general graphs. Finally, we apply the results to give a lower bound for the gonality of arbitrary Drinfeld modular curves over finite fields and for general congruence subgroups Gamma of Gamma(1) that is linear in the index [Gamma(1):Gamma], with a constant that only depends on the residue field degree and the degree of the chosen "infinite" place. This is a function field analogue of a theorem of Abramovich for classical modular curves. We present applications to uniform boundedness of torsion of rank two Drinfeld modules that improve upon existing results, and to lower bounds on the modular degree of certain elliptic curves over function fields that solve a problem of Papikian.