Universality of Spectral Independence with Applications to Fast Mixing in Spin Glasses
Nima Anari, Vishesh Jain, Frederic Koehler, Huy Tuan Pham, Thuy-Duong Vuong
Abstract
We study Glauber dynamics for sampling from discrete distributions on the hypercube ±1 . Recently, techniques based on spectral independence have successfully yielded optimal ( ) relaxation times for a host of different distributions . We show that spectral independence is universal: a relaxation time of ( ) implies spectral independence.
We then study a notion of tractability for , defined in terms of smoothness of the multilinear extension of its Hamiltonian -log -over [-1, +1] . We show that Glauber dynamics has relaxation time ( ) for such , and using the universality of spectral independence, we conclude that these distributions are also fractionally log-concave and consequently satisfy modified log-Sobolev inequalities. We sharpen our estimates and obtain approximate tensorization of entropy and the optimal ( ) mixing time for random Hamiltonians, i.e. the classically studied mixed -spin model at sufficiently high temperature. These results have significant downstream consequences for concentration of measure, statistical testing, and learning.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext c913d1ab-043c-4e60-90f5-12d3d99fffb6Cited by top-tier papers11
- Rapid Mixing at the Uniqueness ThresholdXiaoyu Chen, Zongchen Chen, Yitong Yin, Xinyuan ZhangSTOC 2025 · 15 citations
- Fast Mixing in Sparse Random Ising ModelsKuikui Liu, Sidhanth Mohanty, Amit Rajaraman, David X. WuFOCS 2024 · 14 citations
- Weak Poincaré Inequalities, Simulated Annealing, and Sampling from Spherical Spin GlassesBrice Huang, Sidhanth Mohanty, Amit Rajaraman, David X. WuSTOC 2025 · 13 citations
- Phase Transitions via Complex Extensions of Markov ChainsJingcheng Liu, Chunyang Wang, Yitong Yin, Yixiao YuSTOC 2025 · 6 citations
- A Unified Approach to Learning Ising Models: Beyond Independence and Bounded WidthJason Gaitonde, Elchanan MosselSTOC 2024 · 5 citations
Builds on14
- Spectral Independence in High-Dimensional Expanders and Applications to the Hardcore ModelNima Anari, Kuikui Liu, Shayan Oveis GharanFOCS 2020 · 97 citations
- Optimal mixing of Glauber dynamics: entropy factorization via high-dimensional expansionZongchen Chen, Kuikui Liu, Eric VigodaSTOC 2021 · 61 citations
- Localization Schemes: A Framework for Proving Mixing Bounds for Markov Chains (extended abstract)Yuansi Chen, Ronen EldanFOCS 2022 · 42 citations
- On Mixing of Markov Chains: Coupling, Spectral Independence, and Entropy FactorizationAntonio Blanca, Pietro Caputo, Zongchen Chen, Daniel Parisi et al.SODA 2022 · 41 citations
- Rapid Mixing of Glauber Dynamics up to Uniqueness via ContractionZongchen Chen, Kuikui Liu, Eric VigodaFOCS 2020 · 38 citations
Related papers
- Entropic independence: optimal mixing of down-up random walksNima Anari, Vishesh Jain, Frederic Koehler, Huy Tuan Pham et al.STOC 2022 · 21 citations
- Sampling and Identity-Testing Without Approximate Tensorization of EntropyWilliam Gay, William He, Nicholas Kocurek, Ryan O'DonnellICML 2026
- Fast Sampling of b-Matchings and b-Edge CoversZongchen Chen, Yuzhou GuSODA 2024 · 5 citations
- Rapid Mixing from Spectral Independence beyond the Boolean DomainWeiming Feng, Heng Guo, Yitong Yin, Chihao ZhangSODA 2021 · 18 citations
- Combinatorial Approach for Factorization of Variance and Entropy in Spin SystemsZongchen ChenSODA 2024 · 1 citation
