Abstract
We prove that Ising models on the hypercube with general quadratic interactions satisfy a Poincaré inequality with respect to the natural Dirichlet form corresponding to Glauber dynamics, as soon as the operator norm of the interaction matrix is smaller than 1. The inequality implies a control on the mixing time of the Glauber dynamics. Our techniques rely on a localization procedure which establishes a structural result, stating that Ising measures may be decomposed into a mixture of measures with quadratic potentials of rank one, and provides a framework for proving concentration bounds for high temperature Ising models.
This is a preview of subscription content, access via your institution.
References
Anari, N., Jain, V., Koehler, F., Pham, H.T., Vuong, T.D.: Entropic independence in high-dimensional expanders: modified log-sobolev inequalities for fractionally log-concave polynomials and the Ising model (2021). arXiv preprint arXiv:2106.04105
Anderson, G.W., Guionnet, A., Zeitouni, O.: An Introduction to Random Matrices, vol. 118. Cambridge University Press, Cambridge (2010)
Bauerschmidt, R., Bodineau, T.: A very simple proof of the LSI for high temperature spin systems. J. Funct. Anal. 276(8), 2582–2588 (2019)
Bordenave, C.: A new proof of Friedman’s second eigenvalue theorem and its extension to random lifts (2015). arXiv preprint arXiv:1502.04482
Deshpande, Y., Montanari, A., O’Donnell, R., Schramm, T., Sen, S.: The threshold for SDP-refutation of random regular NAE-3SAT. In: Proceedings of the Thirtieth Annual ACM-SIAM Symposium on Discrete Algorithms, pp. 2305–2321. SIAM (2019)
Diaconis, P., Saloff-Coste, L.: Logarithmic Sobolev inequalities for finite Markov chains. Ann. Appl. Probab. 6(3), 695–750 (1996)
Dobrushin, R.L.: The description of a random field by means of conditional probabilities and conditions of its regularity. Theor. Prob. Appl. 13, 197–224 (1968)
Eldan, R.: Thin shell implies spectral gap up to polylog via a stochastic localization scheme. Geom. Funct. Anal. 23(2), 532–569 (2013)
Galanis, A., Stefankovic, D., Vigoda, E.: Inapproximability of the partition function for the antiferromagnetic Ising and hard-core models. Comb. Probab. Comput. 25, 500–559 (2016)
Gross, L.: Logarithmic Sobolev inequalities. Am. J. Math. 97(4), 1061–1083 (1975)
Hayes, T.P.: A simple condition implying rapid mixing of single-site dynamics on spin systems. In: 2006 47th Annual IEEE Symposium on Foundations of Computer Science (FOCS’06), pp. 39–46. IEEE (2006)
Kannan, R., Lovász, L., Simonovits, M.: Isoperimetric problems for convex bodies and a localization lemma. Discrete Comput. Geom. 13(3–4), 541–559 (1995)
Karatzas, I., Shreve, S.E.: Brownian Motion and Stochastic Calculus. Springer, Berlin (1998)
Klartag, B.: Needle decompositions in Riemannian geometry. Mem. Am. Math. Soc. 249(1180), v+77 (2017)
Ledoux, M.: Concentration of measure and logarithmic Sobolev inequalities. In: Seminaire de Probabilites XXXIII, pp. 120–216. Springer (1999)
Levin, D.A., Peres, Y.: Markov Chains and Mixing Times. American Mathematical Soc., Providence (2017)
Martinelli, F.: Lectures on Glauber dynamics for discrete spin models. In: Lectures on Probability Theory and Statistics (Saint-Flour, 1997), volume 1717 of Lecture Notes in Math., pp. 93–191. Springer, Berlin (1999)
Marton, K.: Logarithmic Sobolev inequalities in discrete product spaces. Comb. Probab. Comput. 28(6), 919–935 (2019)
Mohanty, S., O’Donnell, R., Paredes, P.: Explicit near-Ramanujan graphs of every degree. In: Proceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing, pp. 510–523 (2020)
Sly, A., Sun, N.: The computational hardness of counting in two-spin models on d-regular graphs. In: 2012 IEEE 53rd Annual Symposium on Foundations of Computer Science, pp. 361–369. IEEE (2012)
Talagrand, M.: Mean Field Models for Spin Glasses: Volume I: Basic Examples, vol. 54. Springer, Berlin (2010)
Talagrand, M.: Mean Field Models for Spin Glasses: Volume II: Advanced Replica-Symmetry and Low Temperature. Springer, Berlin (2011)
van Handel, R.: Probability in High Dimension. Technical report, Princeton Univ., NJ (2016). https://web.math.princeton.edu/~rvan/APC550.pdf
Wu, L.: Poincaré and transportation inequalities for Gibbs measures under the Dobrushin uniqueness condition. Ann. Probab. 34(5), 1960–1989 (2006)
Acknowledgements
We would like to thank Fanny Augeri for enlightening discussions. We also thank Roland Bauerschmidt for some useful comments. We thank Ahmed El Alaoui, Heng Guo, Vishesh Jain, and the anonymous reviewers for useful feedback.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
R. Eldan: Supported by a European Research Council Starting Grant (ERC StG) Grant Agreement No. 803084 and by an Israel Science Foundation Grant No. 715/16. F. Koehler: This work was supported in part by NSF CAREER Award CCF-1453261, NSF Large CCF-1565235, Ankur Moitra’s ONR Young Investigator Award, and European Research Council (Grant No. 803084). O. Zeitouni: This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant Agreement No. 692452)
Inequivalence of Dirichlet forms
Inequivalence of Dirichlet forms
The Dirichlet form \({\mathcal {E}}_{\nu }(\varphi ,\varphi )\) can be viewed as the expected norm squared for an appropriate notion of gradient of \(\varphi \). On the other hand, another natural notion of discrete gradient for functions on the hypercube is given by \((\nabla \varphi )_i(x) = \varphi (x_{\sim i}, x_i = 1) - \varphi (x_{\sim i}, x_i = -1)\) which is used in the result of Bauerschmidt and Bodineau [3], and the norm of this discrete gradient squared is the Dirichlet form of a different semigroup with a variable transition rate [17]. In the case of the SK model, the transition rate of the variable-rate chain is sometimes exponentially large in \(\sqrt{n}\); in what follows, we give a simple example of a function \(\varphi \) witnessing that the Dirichlet forms similarly can differ in size by an exponentially large factor in \(\sqrt{n}\).
To compare these two Dirichlet forms, we have the following estimates which follow immediately from (27):
In the context of the SK Model, the parenthesized term is of size \(e^{-\Theta (\beta \sqrt{n})}\) and both estimates are tight up to constants. To see this for the lower bound, define \(a \in \{\pm 1\}^n\) by \(a_1 = -1\) and \(a_j = \text {sgn}(J_{1j})\) otherwise; the significance of this choice is that in the SK model, it’s exponentially unlikely to see \(X_1 = a_1\) given \(X_{\sim 1} = a_{\sim 1}\). Let \(\lambda \) be an atomic measure supported on a, so
If we define \(\varphi = \sqrt{\frac{d\lambda }{d\nu }}\) then, see (28),
In comparison, for the discrete gradient \(\nabla \varphi \) we have
where the lower bound follows by considering the \(a'\) which equals a but flipped on the first coordinate, and which (from the definition of the SK model) is \(e^{\Theta (\beta \sqrt{n})}\) more likely under \(\nu \).
Rights and permissions
About this article
Cite this article
Eldan, R., Koehler, F. & Zeitouni, O. A spectral condition for spectral gap: fast mixing in high-temperature Ising models. Probab. Theory Relat. Fields 182, 1035–1051 (2022). https://doi.org/10.1007/s00440-021-01085-x
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00440-021-01085-x
Mathematics Subject Classification
- 60J10
- 60J27