Table of Contents
Fetching ...

Laplacian eigenvalue conditions for edge-disjoint spanning trees and a forest with constraints

Yongbin Gao, Ligong Wang

TL;DR

This work studies the link between Laplacian spectral data and a graph packing property $P(k,\delta)$, which blends the Nash-Williams/Tutte tree packing framework with a spectral perspective. By introducing the graph classes $\mathcal{G}_1$ and $\mathcal{G}_2$ and leveraging quotient-matrix interlacing, the authors derive explicit sufficient conditions on the Laplacian eigenvalues: for $G\in\mathcal{G}_1$ with $\delta\ge 2k+2$, a bound on $\mu_{n-2}(G)$ ensures $P(k,\delta)$, and for $G\in\mathcal{G}_2$ with $\delta\ge 3k+3$, a bound on $\mu_{n-3}(G)$ ensures $P(k,\delta)$. They further extend these results to the generalized matrix families $aD(G)+A(G)$ and $aD(G)+bA(G)$, obtaining analogous spectral criteria and corollaries involving $\lambda_3$, $q_3$, and $\mu_{n-1}$. The approach combines interlacing with carefully constructed partitions and uses the fractional packing number $\nu_f(G)$ as a bridge to $P(k,\delta)$. Overall, the paper provides concrete, computable spectral thresholds that guarantee enhanced spanning-tree packing and forest structure, with implications for network design and spectral graph theory.

Abstract

Let $k$ be a positive integer and let $G$ be a simple graph of order $n$ with minimum degree $δ$. A graph $G$ is said to have property $P(k, d)$ if it contains $k$ edge-disjoint spanning trees and an additional forest $F$ with edge number $|E(F)| > \frac{d-1}{d}(|V(G)| - 1)$, such that if $F$ is not a spanning tree, then $F$ has a component with at least $d$ edges. Let $D(G)$ be the degree diagonal matrix of $G$. We denote $λ_i$ and $μ_i$ as the $i$th largest eigenvalue of the adjacency matrix $A(G)$ of $G$ and the Laplacian matrix $L(G) = D(G) - A(G)$ of $G$ for $i = 1, 2, \ldots, n$, respectively. In this paper, we investigate the relationship between Laplacian eigenvalues and property $P(k, δ)$. Let $t$ be a positive integer, and define $\mathcal{G}_t$ as the set of simple graphs such that each $G \in \mathcal{G}_t$ contains at least $t+1$ non-empty disjoint proper subsets $V_1, V_2, \ldots, V_{t+1}$ satisfying $V(G) \setminus \bigcup_{i=1}^{t+1} V_i \neq \emptyset$ and edge connectivity $κ'(G) = e(V_i, V(G) \setminus V_i)$ for any $i = 1, 2, \ldots, t+1$. For the class of graphs $\mathcal{G}_1$ with minimum degree $δ$, we provide a sufficient condition involving the third smallest Laplacian eigenvalue $μ_{n-2}(G)$ for a graph $G\in \mathcal{G}_1$ to have property $P(k, δ)$. Similarly, for the class of graphs $\mathcal{G}_2$ with minimum degree $δ$, we establish a corresponding sufficient condition involving the fourth smallest Laplacian eigenvalue $μ_{n-3}(G)$ for a graph $G\in \mathcal{G}_2$ to have property $P(k, δ)$. Furthermore, we extend the spectral conditions for all the results about $μ_{n-2}(G)$, $μ_{n-3}(G)$ and $λ_2(G)$ to the general graph matrices $aD(G) + A(G)$ and $aD(G) + bA(G)$.

Laplacian eigenvalue conditions for edge-disjoint spanning trees and a forest with constraints

TL;DR

This work studies the link between Laplacian spectral data and a graph packing property , which blends the Nash-Williams/Tutte tree packing framework with a spectral perspective. By introducing the graph classes and and leveraging quotient-matrix interlacing, the authors derive explicit sufficient conditions on the Laplacian eigenvalues: for with , a bound on ensures , and for with , a bound on ensures . They further extend these results to the generalized matrix families and , obtaining analogous spectral criteria and corollaries involving , , and . The approach combines interlacing with carefully constructed partitions and uses the fractional packing number as a bridge to . Overall, the paper provides concrete, computable spectral thresholds that guarantee enhanced spanning-tree packing and forest structure, with implications for network design and spectral graph theory.

Abstract

Let be a positive integer and let be a simple graph of order with minimum degree . A graph is said to have property if it contains edge-disjoint spanning trees and an additional forest with edge number , such that if is not a spanning tree, then has a component with at least edges. Let be the degree diagonal matrix of . We denote and as the th largest eigenvalue of the adjacency matrix of and the Laplacian matrix of for , respectively. In this paper, we investigate the relationship between Laplacian eigenvalues and property . Let be a positive integer, and define as the set of simple graphs such that each contains at least non-empty disjoint proper subsets satisfying and edge connectivity for any . For the class of graphs with minimum degree , we provide a sufficient condition involving the third smallest Laplacian eigenvalue for a graph to have property . Similarly, for the class of graphs with minimum degree , we establish a corresponding sufficient condition involving the fourth smallest Laplacian eigenvalue for a graph to have property . Furthermore, we extend the spectral conditions for all the results about , and to the general graph matrices and .
Paper Structure (5 sections, 16 theorems, 46 equations)

This paper contains 5 sections, 16 theorems, 46 equations.

Key Result

Theorem 1.1

Let $k$ and $d$ be positive integers. For a nontrivial graph $G$, if then $G$ has property $P(k, d)$.

Theorems & Definitions (22)

  • Theorem 1.1: $RF10RF9$
  • Theorem 1.2
  • Theorem 1.3
  • Theorem 2.1: Cauchy Interlacing Theorem RF11RF12
  • Theorem 2.2: Weyl's Inequalities $RF11$
  • Lemma 2.3: Lemma 2.8 in $RF2$
  • Lemma 2.4: Adapted from Theorem 3.4 in $RF6$
  • Lemma 2.5: Adapted from Theorem 4.4 in $RF6$
  • proof : Proof of Theorem \ref{['thm:1.2']}
  • Corollary 3.1
  • ...and 12 more