Spanning trees and their relations in Galois covers
Kosuke Mizuno
TL;DR
The paper generalizes spanning-tree relations in Galois covers from the $(\mathbb{Z}/2\mathbb{Z})^{m}$ case to arbitrary finite groups by proving two Möbius-inversion formulas that express the spanning-tree count of the top and intermediate graphs via intermediate covers. Using the Ihara zeta function and Artin–Ihara $L$-functions, the authors derive graph-theoretic analogues of Kuroda's formula and Brauer–Kuroda relations, establishing a broad framework that links intermediate-graph complexities to the global complexity $\kappa(Y)$. A key insight is the role of Möbius functions on the posets $\mathscr{H}_G$ and $\mathscr{C}_G$ in weighting these products, which recovers the known HKSV result in the $\mathbb{Z}/2\mathbb{Z}$-power case and clarifies when such formulas do not exist, notably for cyclic G. The results deepen the connection between graph theory and algebraic number theory, provide practical means to compute $\kappa(Y)$ from intermediate-data when $G$ is noncyclic, and reveal intrinsic limitations for monomial-type spanning-tree relations in the cyclic setting.
Abstract
This paper studies the relation among the number of spanning trees of intermediate graphs in a Galois cover, building on results for $(\mathbb{Z}/2\mathbb{Z})^m$-covers previously established by Hammer, Mattman, Sands, and Vallières. We generalize their results to arbitrary finite Galois covers. Using the Ihara zeta function and the Artin--Ihara $L$-function, we prove two formulas which are graph-theoretic analogues of Kuroda's formula and the Brauer--Kuroda relations in algebraic number theory. Furthermore, we prove that a spanning tree formula does not exist if the Galois group is cyclic.
