1: \begin{abstract}

2: The Swendsen-Wang algorithm is a sophisticated, widely-used Markov chain for

3: sampling from the Gibbs distribution for the ferromagnetic Ising and Potts models.

4: %with inverse temperature $\beta>0$.

5: This chain has proved difficult to analyze,

6: due in part to the global nature of its updates.

7: %often mixes rapidly in the low-temperature region (large $\beta$)

8: %where long-range correlations impair the convergence rate of local chains such as the 

9: %Glauber dynamics.

10: We present optimal bounds on the convergence rate of the Swendsen-Wang algorithm

11: for the complete $d$-ary tree. Our bounds extend to the non-uniqueness region and apply to all boundary conditions. 

12: 

13: We show 

14: that the spatial mixing conditions known as \emph{Variance Mixing} and \emph{Entropy Mixing},

15: introduced 

16: in the study of local Markov chains by Martinelli et al.\ (2003),

17: imply $\Omega(1)$

18: spectral gap and $O(\log{n})$ mixing time, respectively, for the Swendsen-Wang dynamics on the $d$-ary tree.

19: We also show that these bounds are asymptotically optimal.

20: As a consequence, we establish $\Theta(\log{n})$ mixing for the Swendsen-Wang dynamics 

21: for \emph{all} boundary conditions throughout the tree uniqueness region;

22: in fact, our bounds hold beyond the uniqueness threshold for the Ising model, 

23: and for the $q$-state Potts model when $q$ is small with respect to $d$. 

24: Our proofs feature a novel spectral view of the Variance Mixing condition 

25: inspired by several recent

26: rapid mixing results on high-dimensional expanders

27: and utilize recent work on block factorization of entropy under spatial mixing conditions.

28: \end{abstract}

29: