Concentration inequalities and Transfer operators for supercritical Dyson models
Mirmukhsin Makhmudov · 2025
Read Concentration inequalities and Transfer operators for supercritical Dyson models by Mirmukhsin Makhmudov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-01703.txt.
Introduction
Mirmukhsin Makhmudov extends work on ferromagnetic Dyson models in a 2025 mathematics paper. These models describe spins on a line that interact over long ranges, a classic test bed in probability and physics. He pushes earlier results into near-critical and strongly interacting regimes. As a main result, he proves Gaussian concentration bounds for the model's unique infinite-volume Gibbs measure across the whole supercritical regime, using transfer operator methods.
Catalog record
| CREATOR | Mirmukhsin Makhmudov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-01703
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Concentration inequalities and Transfer operators for supercritical Dyson models Mirmukhsin Makhmudov Abstract . The present paper extends the results on the ferromagnetic Dyson models from [ 19 , 14 ] to the near-critical and strongly interacting regimes. As part of our main result, we further establish Gaussian concentration bounds for the unique infinite-volume Gibbs measure throughout the entire supercritical regime. 1. Introduction The one-dimensional long-range Ising model, a.k.a. the Dyson model, is one of the archetypical models in Statistical Mechanics. This model has served as the testing ground for many conjectures and ideas in statistical mechanics, and it has also been the first example of a one-dimensional model which undergoes phase transitions at low temperatures. A Dyson interaction \Phi=(\Phi_{\Lambda})_{\Lambda\Subset\mathbb{Z}} is defined on the configuration space X:=\{\pm 1\}^{\mathbb{Z}} of \omega\in X by: (1) \Phi_{\Lambda}(\omega):=\begin{cases}-J(|i-j|)\omega_{i}\omega_{j},&\text{ if }\Lambda=\{i,j\}\subset\mathbb{Z},\ i\neq j;\\ 0,&\text{otherwise},\end{cases} where J(k)\geq 0 for every k\in\mathbb{N} and to have well-defined infinite-volume Gibbs measures, we assume that \sum_{k=1}^{\infty}J(k)<\infty . The most well-known example is the standard Dyson interaction for which J(k)=k^{-\alpha} for every k\in\mathbb{N} and for some \alpha>1 . In this paper, instead of studying one fixed interaction \Phi , we study the family \beta\Phi=(\beta\Phi_{\Lambda})_{\Lambda\Subset\mathbb{Z}} of interactions parameterised by \beta\geq 0 , which is interpreted as the inverse temperature in statistical mechanics. A Gibbs measure for the interaction \beta\Phi is a probability measure \mu on X that satisfies the Dobrushin–Lanford–Ruelle (DLR) equations. Equivalently, for finite volumes \Lambda\Subset\mathbb{Z} , the conditional distribution of \mu given the configuration outside \Lambda is prescribed by the Boltzmann-Gibbs ansatz: (2) \mu(\omega_{\Lambda}\mid\omega_{\Lambda^{c}})=\gamma^{\beta\Phi}_{\Lambda}(\omega_{\Lambda}\mid\omega_{\Lambda^{c}}),\;\;\;\mu-\text{a.e. }\;\omega\in X, where (3) \gamma^{\beta\Phi}_{\Lambda}(\omega_{\Lambda}\mid\omega_{\Lambda^{c}})=\frac{\exp\left(-H_{\Lambda}^{\beta\Phi}(\omega)\right)}{\sum\limits_{\bar{\omega}_{\Lambda}\in\{\pm 1\}^{\Lambda}}\exp\left(-H_{\Lambda}^{\beta\Phi}(\bar{\omega}_{\Lambda}\omega_{\Lambda^{c}})\right)},\;\;\;\;H_{\Lambda}^{\beta\Phi}=\sum_{\begin{subarray}{c}V:V\Subset\mathbb{Z}\\ V\cap\Lambda\neq\emptyset\end{subarray}}\beta\Phi_{V}, and \bar{\omega}_{\Lambda}\omega_{\Lambda^{c}}\in X denotes the configuration constructed by concatenating the strings \bar{\omega}_{\Lambda} and \omega_{\Lambda^{c}} . It is a well-known fact [ 8 , 15 ] for the standard Dyson model that for every \alpha\in(1,2] there exists a critical value \beta_{c}(\alpha)\in(0,\infty) such that for every \beta<\beta_{c}(\alpha) , there exists a unique Gibbs measure for \beta\Phi and for every \beta>\beta_{c}(\alpha) , there are multiple Gibbs measures for \beta\Phi . We also consider the interaction \Phi on the half-line \mathbb{Z}_{+} . Studying the interaction \Phi on {\mathbb{Z}_{+}} amounts to studying the thermodynamic properties of the potential: (4) \phi(x):=\sum_{n=1}^{\infty}J(n)x_{0}x_{n},\;\;\;x\in X_{+}:=\{\pm 1\}^{\mathbb{Z}_{+}}. Note that \phi and \Phi are related through: (5) \phi=-\sum_{0\in V\Subset\mathbb{Z}_{+}}\Phi_{V}, which often appears in the variational characterisation of the Gibbs measures. In fact, by the Gibbs variational principle [ 10 ] *Theorem 15.39, the translation-invariant Gibbs measures for \Phi – that is, those preserved by the shift map S – are precisely the equilibrium states for the potential \phi . These are the measures \mu\in\mathcal{M}_{1,S}(X) satisfying (6) h(\mu)+\int_{X}\phi d\mu=\sup\Big\{h(\tau)+\int_{X}\phi d\tau:\tau\in\mathcal{M}_{1,S}(X)\Big\}, where h(\tau) is the measure-theoretic entropy of the measure \tau , the shift map S:X\to X is defined by S\omega=(\omega_{i+1})_{i} and \mathcal{M}_{1,S}(X) denotes the set of the translation-invariant probability measures on X . One of the main objectives in this paper is to study the Ruelle-Perron-Frobenius transfer operator \mathcal{L}_{\beta\phi} for the potential \beta\phi , which acts on the space of functions on X_{+} by (7) \mathcal{L}_{\beta\phi}f(x):=\sum_{a\in\{\pm 1\}}e^{\beta\phi(ax)}f(ax),\;\;x\in X_{+}, here f\in\mathbb{R}^{X_{+}} and ax\in X_{+} denotes concatenated configuration. First, we note that for the spectral radius \rho(\mathcal{L}_{\beta\phi}) of the operator \mathcal{L}_{\beta\phi} , one has \rho(\mathcal{L}_{\beta\phi})=e^{P(\beta\phi)} , where P(\beta\phi) is the topological pressure of \beta\phi , which also coincides with the infinite-volume pressure of the interaction \beta\Phi . Second, the adjoint operator \mathcal{L}_{\beta\phi}^{*} , which acts on the space of signed measures \mathcal{M}(X_{+}) , always has an eigenprobability \nu\in\mathcal{M}_{1}(X_{+}) corresponding to the spectral radius \lambda=\rho(\mathcal{L}_{\beta\phi}) , i.e., \mathcal{L}_{\beta\phi}^{*}\nu=\lambda\nu . Throughout the paper, we also refer to the eigenprobabilities of \mathcal{L}_{\beta\phi}^{*} as the one-sided Gibbs measures for \beta\phi (or \beta\Phi ) since they are exactly the probability measures satisfying the DLR equations corresponding to \beta\phi (or \beta\Phi|_{\mathbb{Z}_{+}} ) [ 4 ] . In the half-line \mathbb{Z}_{+} , a similar phase diagram to that of the interactions \beta\Phi on \mathbb{Z} is also established. Indeed, in [ 13 ] , it was proven for the standard Dyson model that there exists a critical temperature \beta_{c}^{+}(\alpha)\in(0,\infty) such that for every \beta\in[0,\beta^{+}_{c}(\alpha)) there exists a unique eigenprobability of \mathcal{L}_{\beta\phi}^{*} and for \beta>\beta^{+}_{c}(\alpha) there are multiple eigenprobabilities. In [ 13 ] , Johansson, Öberg and Pollicott also showed that for every \alpha\in(1,2] , \beta_{c}^{+}(\alpha)\leq 8\beta_{c}(\alpha) and conjectured that \beta^{+}_{c}(\alpha)=\beta_{c}(\alpha) . We note that part of this conjecture – namely, that \beta_{c}(\alpha)\leq\beta_{c}^{+}(\alpha) for every \alpha\in(1,2] – follows directly from the Griffiths inequalities [ 9 , 11 , 12 ] (see Proposition 2.2 ). The eigenprobabilities correspond, in a dual sense, to the principal eigenfunctions of the transfer operator \mathcal{L}_{\beta\phi} . While the existence of an eigenprobability of the adjoint operator \mathcal{L}_{\beta\phi}^{*} follows directly from the DLR equations, establishing the existence of a principal eigenfunction for \mathcal{L}_{\beta\phi} is considerably more subtle and technically involved. Important advancements in this direction have been made in [ 19 , 14 , 16 ] , whose relevant results — specialised here to the standard Dyson model — are summarised in the following theorem. Theorem 1.1 . [ 19 , 14 ] Let \Phi be the standard Dyson interaction ( 1 ). Then (i) for all \alpha>1 and for \beta\geq 0 sufficiently small , the unique Gibbs state \mu (in fact, the restriction to X_{+} ) for \beta\Phi is equivalent to the unique half-line Gibbs state \nu for \beta\phi , i.e., \mu\ll\nu and \nu\ll\mu . In particular, the Perron-Frobenius transfer operator \mathcal{L}_{\beta\phi} has a non-negative integrable eigenfunction in L^{1}(X_{+},\nu) ; (ii) if \alpha>\frac{3}{2} , then for all 0\leq\beta<\beta_{c}(\alpha) , the Gibbs state \mu for \beta\Phi is not only equivalent to the half-line Gibbs state \nu , but there even exists a continuous version of the Radon-Nikodym density \frac{d\mu}{d\nu} . Thus, the Perron-Frobenius transfer operator \mathcal{L}_{\beta\phi} has a positive continuous eigenfunction. The main objective of this paper, in the context of transfer operators, is to extend the above result to the entire supercritical region, that is, to all pairs (\alpha,\beta) with \alpha\in(1,2] and \beta\in[0,\beta_{c}(\alpha)) . We, in fact, state the main result of this paper for interactions given by ( 1 ), which are slightly more general than the standard Dyson interaction. A broader generalisation was considered in [ 19 ] , where the authors studied a class of long-range interactions – potentially including multi-body and non-ferromagnetic cases. However, the analysis in [ 19 ] was restricted to sufficiently small values of \beta\geq 0 , ensuring that the interaction \Phi satisfies the Dobrushin uniqueness condition [ 10 ] *Chapter 8. This condition corresponds to the high-temperature regime and is known to be strictly contained within the uniqueness region. Near-critical regimes , i.e., when 0\ll\beta<\beta_{c}(\alpha) , are typically more delicate to analyse in statistical mechanics and often require different techniques. This phenomenon is also reflected in our setting. We also note that [ 14 ] approached the problem using the random cluster representation – a method fundamentally different from the techniques used in [ 19 ] – and successfully treated both the high-temperatures 0\leq\beta\ll\beta_{c}(\alpha) and the near-critical regimes 0\ll\beta<\beta_{c}(\alpha) , under the assumption that \alpha>\frac{3}{2} . However, the methods in [ 14 ] fundamentally depend on the square summability of the variations of the potential \phi , a condition that holds only for \alpha>\frac{3}{2} . Consequently, their approach does not extend to the strongly interacting setting \alpha\leq\frac{3}{2} . In another