Uniqueness and Rapid Mixing in the Bipartite Hardcore Model

Xiaoyu Chen,Jingcheng Liu,Yitong Yin

2023-04-29

Abstract:We characterize the uniqueness condition in the hardcore model for bipartite graphs with degree bounds only on one side, and provide a nearly linear time sampling algorithm that works up to the uniqueness threshold. We show that the uniqueness threshold for bipartite graph has almost the same form of the tree uniqueness threshold for general graphs, except with degree bounds only on one side of the bipartition. The hardcore model from statistical physics can be seen as a weighted enumeration of independent sets. Its bipartite version (#BIS) is a central open problem in approximate counting. Compared to the same problem in a general graph, surprising tractable regime have been identified that are believed to be hard in general. This is made possible by two lines of algorithmic approach: the high-temperature algorithms starting from Liu and Lu (STOC 2015), and the low-temperature algorithms starting from Helmuth, Perkins, and Regts (STOC 2019). In this work, we study the limit of these algorithms in the high-temperature case. Our characterization of the uniqueness condition is obtained by proving decay of correlations for arguably the best possible regime, which involves locating fixpoints of multivariate iterative rational maps and showing their contraction. We also give a nearly linear time sampling algorithm based on simulating field dynamics only on one side of the bipartite graph that works up to the uniqueness threshold. Our algorithm is very different from the original high-temperature algorithm of Liu and Lu, and it makes use of a connection between correlation decay and spectral independence of Markov chains. Last but not the least, we are able to show that the standard Glauber dynamics on both side of the bipartite graph mixes in polynomial time up to the uniqueness.

Data Structures and Algorithms,Probability

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to determine the uniqueness conditions in the hardcore model in bipartite graphs and provide a nearly - linear sampling algorithm that works within the uniqueness threshold. Specifically, the paper focuses on the uniqueness conditions of the hardcore model in bipartite graphs with degree restrictions on only one side and how to sample effectively under such conditions.

### Main contributions of the paper

1. **Characterization of uniqueness conditions**:
- The paper characterizes the uniqueness conditions of the hardcore model in bipartite graphs with degree restrictions on only one side. By proving the decay of correlations, the authors find the best possible region, which involves the fixed points of multivariate iterative rational maps and their contraction properties.
- The authors discover that a region considered to be "low - temperature" (low - temperature) is actually within the uniqueness (high - temperature) region.

2. **Nearly - linear sampling algorithm**:
- The paper proposes a nearly - linear - time sampling algorithm based on field - dynamics simulation, which works effectively within the uniqueness threshold. This algorithm is different from previous high - temperature algorithms in that it utilizes the connection between the decay of correlations and the spectral independence of Markov chains.

3. **Fast convergence of mixing time**:
- The authors prove that the standard Glauber dynamics mixes in polynomial time on bipartite graphs on both sides up to the uniqueness threshold. Although the total influence and the spectral radius of the adjacency matrix can be unbounded, the authors prove the bounds of the mixing time through the framework of spectral independence.

### Technical details

- **Uniqueness conditions**:
- The authors introduce the concept of δ - uniqueness, that is, for all fixed points \(\hat{x}\) satisfying \( F(\hat{x})=\hat{x}\), it is required that \( F'(\hat{x})\leq1 - \delta\).
- By analyzing the decay of correlations, the authors prove that when \(\lambda < \lambda_c(\Delta_L)\), the decay of correlations holds, where \(\Delta_L\) is the degree restriction on only one side.

- **Saming algorithm**:
- The authors design a sampling algorithm based on field - dynamics simulation, which works effectively within the uniqueness threshold. The time complexity of the algorithm is \( n\cdot\left(\frac{\Delta\log n}{\lambda}\right)^{O(C / \delta)}\cdot\log^2\left(\frac{1}{\epsilon}\right)\), where \( C\) is an absolute constant.

- **Mixing time**:
- The authors prove that the standard Glauber dynamics mixes in polynomial time by comparing the variance decay of field - dynamics between one - sided and two - sided Glauber dynamics. The upper bound of the mixing time is \(\left(\frac{\Delta\log n}{\lambda}\right)^{O(C / \delta)}\cdot n^2\cdot\left(\frac{n\log(1 + \lambda)}{\min\{1,\lambda\}}+\log\frac{1}{\epsilon}\right)\).

### Conclusion

The paper solves the problem of determining the uniqueness conditions in the hardcore model in bipartite graphs with degree restrictions on only one side and provides an effective sampling algorithm. These results are not only theoretically breakthroughs but also of great significance in practical applications.