Abstract:We develop a new framework to prove the mixing or relaxation time for the Glauber dynamics on spin systems with unbounded degree. It works for general spin systems including both $2$-spin and multi-spin systems. As applications for this approach:
$\bullet$ We prove the optimal $O(n)$ relaxation time for the Glauber dynamics of random $q$-list-coloring on an $n$-vertices triangle-tree graph with maximum degree $\Delta$ such that $q/\Delta > \alpha^\star$, where $\alpha^\star \approx 1.763$ is the unique positive solution of the equation $\alpha = \exp(1/\alpha)$. This improves the $n^{1+o(1)}$ relaxation time for Glauber dynamics obtained by the previous work of Jain, Pham, and Vuong (2022). Besides, our framework can also give a near-linear time sampling algorithm under the same condition.
$\bullet$ We prove the optimal $O(n)$ relaxation time and near-optimal $\widetilde{O}(n)$ mixing time for the Glauber dynamics on hardcore models with parameter $\lambda$ in $\textit{balanced}$ bipartite graphs such that $\lambda < \lambda_c(\Delta_L)$ for the max degree $\Delta_L$ in left part and the max degree $\Delta_R$ of right part satisfies $\Delta_R = O(\Delta_L)$. This improves the previous result by Chen, Liu, and Yin (2023).
At the heart of our proof is the notion of $\textit{coupling independence}$ which allows us to consider multiple vertices as a huge single vertex with exponentially large domain and do a "coarse-grained" local-to-global argument on spin systems. The technique works for general (multi) spin systems and helps us obtain some new comparison results for Glauber dynamics.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to prove the mixing or relaxation time of Glauber dynamics in spin systems on graphs with unbounded degrees. Specifically, the author develops a new framework to prove the mixing or relaxation time of such systems, and this framework is applicable to general spin systems, including 2 - spin and multi - spin systems. The following are two main applications in the paper:
1. **Random q - list coloring**:
- The author proves that on a triangular tree graph with maximum degree Δ, when q/Δ > α⋆ (where α⋆≈ 1.763 is the unique positive solution of the equation \( \alpha=\exp(1 / \alpha) \)), the optimal relaxation time of Glauber dynamics is O(n). This improves the \( n^{1 + o(1)} \) relaxation time obtained previously by Jain, Pham, and Vuong (2022). In addition, this framework can also give a nearly - linear sampling algorithm under the same conditions.
2. **Hard - core model**:
- The author proves that on a balanced bipartite graph, when the parameter λ < λc(ΔL) (where ΔL is the maximum degree of the left part, ΔR is the maximum degree of the right part and satisfies ΔR = O(ΔL)), the optimal relaxation time of Glauber dynamics is O(n), and the approximately optimal mixing time is \( \tilde{O}(n) \). This improves the results obtained previously by Chen, Liu, and Yin (2023).
### Core method of the paper
The core method of the paper is coupling independence. This concept allows researchers to regard multiple vertices as a huge single vertex with an exponentially large domain and conduct "coarse - grained" arguments from local to global on the spin system. This technique is applicable to general (multi - ) spin systems and helps researchers obtain some new comparison results about Glauber dynamics.
### Formulas and definitions
- **Coupling independence**:
- A distribution \( \mu \) is called C - coupling independent (C - CI), if there exists a Hamming weight function \( \rho: V \to \mathbb{N}_{>0} \), such that for any two (possibly partial) configurations \( \sigma_1, \sigma_2 \) on \( S \subseteq V \), which differ only at one vertex \( v_0 \in S \), there exists a coupling \( (X, Y) \), where \( X \sim \mu_{\sigma_1} \) and \( Y \sim \mu_{\sigma_2} \), such that:
\[
\frac{\mathbb{E}[H_\rho(X, Y)]}{\rho(v_0)} \leq C
\]
- where \( H_\rho(\sigma, \tau) \) is defined as:
\[
H_\rho(\sigma, \tau) := \sum_{v \in \Lambda: \sigma(v) \neq \tau(v)} \rho(v)
\]
- **Relaxation time**:
- The relaxation time of Glauber dynamics is defined as:
\[
T_{\text{rel}}^{\text{GD}}(\mu) := \frac{1}{1 - \lambda_2}
\]
- where \( \lambda_2 \) is the second - largest eigenvalue of the transition matrix \( P \).
- **Mixing time**:
- The mixing time of Glauber dynamics is defined as:
\[
T_{\text{mix}}^{\text{GD}}(\mu, \epsilon) := \max_{X_0 \in \Omega(\mu)} \min \{ t > 0 \mid D_{\text{TV}}(X_t \| \mu) \leq \epsilon \}