Prime Factorization of the Kirchhoff Polynomial: Compact Enumeration of Arborescences

TL;DR

This paper introduces a method to efficiently factorize the Kirchhoff polynomial of a directed graph, enabling compact enumeration of all arborescences and facilitating applications like steady state calculations and gcd computations.

Contribution

The authors develop a linear time algorithm for prime factorization of the Kirchhoff polynomial based on digraph decomposition rules, linking algebraic factorization to graph connectivity properties.

Findings

Linear time decomposition algorithm for digraphs

Prime factorization of Kirchhoff polynomial based on connectivity

Application to efficient enumeration of arborescences

Abstract

We study the problem of enumerating all rooted directed spanning trees (arborescences) of a directed graph (digraph) $G=(V,E)$ of $n$ vertices. An arborescence $A$ consisting of edges $e_1,\ldots,e_{n-1}$ can be represented as a monomial $e_1\cdot e_2 \cdots e_{n-1}$ in variables $e \in E$. All arborescences $\mathsf{arb}(G)$ of a digraph then define the Kirchhoff polynomial $\sum_{A \in \mathsf{arb}(G)} \prod_{e\in A} e$. We show how to compute a compact representation of the Kirchhoff polynomial -- its prime factorization, and how it relates to combinatorial properties of digraphs such as strong connectivity and vertex domination. In particular, we provide digraph decomposition rules that correspond to factorization steps of the polynomial, and also give necessary and sufficient primality conditions of the resulting factors expressed by connectivity properties of the corresponding decomposed components. Thereby, we obtain a linear time algorithm for decomposing a digraph into components corresponding to factors of the initial polynomial, and a guarantee that no finer factorization is possible. The decomposition serves as a starting point for a recursive deletion-contraction algorithm, and also as a preprocessing phase for iterative enumeration algorithms. Both approaches produce a compressed output and retain some structural properties in the resulting polynomial. This proves advantageous in practical applications such as calculating steady states on digraphs governed by Laplacian dynamics, or computing the greatest common divisor of Kirchhoff polynomials. Finally, we initiate the study of a class of digraphs which allow for a practical enumeration of arborescences. Using our decomposition rules we observe that various digraphs from real-world applications fall into this class or are structurally similar to it.