Abstract:We study the edge overlap and local limit of the random spanning tree in random environment (RSTRE) on the complete graph with $n$ vertices and weights given by $\exp(-\beta \omega_e)$ for $\omega_e$ uniformly distributed on $[0,1]$. We show that for $\beta$ growing with $\beta = o(n/\log n)$, the edge overlap is $(1+o(1)) \beta$, while for $\beta$ much larger than $n \log^2 n$, the edge overlap is $(1-o(1))n$. Furthermore, there is a transition of the local limit around $\beta = n$. When $\beta = o(n/ \log n)$ the RSTRE locally converges to the same limit as the uniform spanning tree, whereas for $\beta$ larger than $n \log^\lambda n$, where $\lambda = \lambda(n) \rightarrow \infty$ arbitrarily slowly, the local limit of the RSTRE is the same as that of the minimum spanning tree.
What problem does this paper attempt to address?
This paper aims to solve the edge - overlap and local - limit problems of the Random Spanning Tree in Random Environment (RSTRE) as the parameter β changes in a random environment. Specifically:
1. **Edge - overlap problem**: The researchers explored the edge - overlap situation of RSTRE under different β values when the edge weights are given by \(\exp(-\beta \omega_e)\) on a complete graph. The paper points out that:
- When \(\beta = o\left(\frac{n}{\log n}\right)\), the edge - overlap \(O(\beta)=(1 + o(1))\beta\).
- When \(\beta\gg n\log^2 n\), the edge - overlap \(O(\beta)=(1 - o(1))n\).
2. **Local - limit problem**: The researchers analyzed the local - limit of RSTRE under different β values:
- When \(\beta = o\left(\frac{n}{\log n}\right)\), RSTRE locally converges to the same limit as the Uniform Spanning Tree (UST).
- When \(\beta\geq n\log^\lambda n\) (where \(\lambda=\lambda(n)\to\infty\) arbitrarily slowly), the local - limit of RSTRE is the same as that of the Minimum Spanning Tree (MST).
### Formula summary
- Edge - overlap formula:
\[
O(\beta)=\mathbb{E}_{\omega}^{n,\beta}\left[\left|T\cap T_1\right|\right]=\sum_{T,T_1\in\mathcal{T}(G)}|T\cap T_1|P_{\omega}^{n,\beta}(T = T)P_{\omega}^{n,\beta}(T = T_1)
\]
where
\[
P_{\omega}^{n,\beta}(T = T)=\frac{1}{Z_{\omega}^{\beta}}\prod_{e\in T}w(e)=\frac{1}{Z_{\omega}^{\beta}}\prod_{e\in T}\exp(-\beta \omega_e)
\]
- Local - limit results:
- For the low - disorder region (\(\beta = o\left(\frac{n}{\log n}\right)\)):
\[
T^{(1)}\xrightarrow{d}\mathcal{P}^{(o)}
\]
where \(T^{(1)}\) is the RSTRE rooted at vertex 1, and \(\mathcal{P}^{(o)}\) is a conditionally surviving Poisson(1) branching process.
- For the high - disorder region (\(\beta\geq n\log^\lambda n\)):
\[
T^{(1)}\xrightarrow{d}M^{(o)}
\]
where \(M^{(o)}\) is the local - limit of MST.
### Main conclusion
The paper shows that the properties of RSTRE change significantly under different β values. In particular, when \(\beta\) is close to \(n\), the local - limit of RSTRE changes from being similar to UST to being similar to MST. These results are of great significance for understanding the spanning - tree structure in complex networks and its behavior as the environment changes.