English

Fast sampling via spectral independence beyond bounded-degree graphs

Data Structures and Algorithms 2023-10-16 v4

Abstract

Spectral independence is a recently-developed framework for obtaining sharp bounds on the convergence time of the classical Glauber dynamics. This new framework has yielded optimal O(nlogn)O(n \log n) sampling algorithms on bounded-degree graphs for a large class of problems throughout the so-called uniqueness regime, including, for example, the problems of sampling independent sets, matchings, and Ising-model configurations. Our main contribution is to relax the bounded-degree assumption that has so far been important in establishing and applying spectral independence. Previous methods for avoiding degree bounds rely on using LpL^p-norms to analyse contraction on graphs with bounded connective constant (Sinclair, Srivastava, Yin; FOCS'13). The non-linearity of LpL^p-norms is an obstacle to applying these results to bound spectral independence. Our solution is to capture the LpL^p-analysis recursively by amortising over the subtrees of the recurrence used to analyse contraction. Our method generalises previous analyses that applied only to bounded-degree graphs. As a main application of our techniques, we consider the random graph G(n,d/n)G(n,d/n), where the previously known algorithms run in time nO(logd)n^{O(\log d)} or applied only to large dd. We refine these algorithmic bounds significantly, and develop fast n1+o(1)n^{1+o(1)} algorithms based on Glauber dynamics that apply to all dd, throughout the uniqueness regime.

Keywords

Cite

@article{arxiv.2111.04066,
  title  = {Fast sampling via spectral independence beyond bounded-degree graphs},
  author = {Ivona Bezáková and Andreas Galanis and Leslie Ann Goldberg and Daniel Štefankovič},
  journal= {arXiv preprint arXiv:2111.04066},
  year   = {2023}
}

Comments

TALG, To Appear

R2 v1 2026-06-24T07:29:22.149Z