Metastability for the Curie–Weiss–Potts model with unbounded random interactions
Johan L. A. Dubbeldam, Vicente Lenz Burnier, Elena Pulvirenti, Martin Slowik · 2025
Read Metastability for the Curie–Weiss–Potts model with unbounded random interactions by Johan L. A. Dubbeldam, Vicente Lenz Burnier, Elena Pulvirenti, Martin Slowik (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-2505-11260.txt.
Introduction
Some systems linger near one state for ages before jumping to another, a behavior called metastability. Johan Dubbeldam and co-authors study it in a mean-field magnet model where each spin can take several values and the couplings between spins are random and unbounded. Spins flip one at a time under heat-driven rules.
First the authors settle the orderly version of the model, where all couplings match. Then they carry the result over to the disordered one, showing the lingering behavior survives with high probability. The core output is the ratio between the jump times of the two models. They pin down its long-run tail shape and its average values. The proof mixes potential theory with concentration tools. The work extends metastability theory to messier, more realistic randomness.
Catalog record
| CREATOR | Johan L. A. Dubbeldam, Vicente Lenz Burnier, Elena Pulvirenti, Martin Slowik |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 144 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-11260
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Metastability for the Curie–Weiss–Potts model with unbounded random interactions Johan L. A. Dubbeldam, Vicente Lenz Burnier, Elena Pulvirenti, Martin Slowik Abstract . We analyse the metastable behaviour of the disordered Curie–Weiss–Potts (DCWP) model subject to a Glauber dynamics. The model is a randomly disordered version of the mean-field q q -spin Potts model (CWP), where the interaction coefficients between spins are general independent random variables. These random variables are chosen to have fixed mean (for simplicity taken to be 1 1 ) and well defined cumulant generating function, with a fixed distribution not depending on the number of particles. The system evolves as a discrete-time Markov chain with single spin flip Metropolis dynamics at finite inverse temperature β \beta . We provide a comparison of the metastable behaviour of the CWP and DCWP models, when N → ∞ N\to\infty . First, we establish the metastability of the CWP model and, using this result, prove metastability for the DCWP model (with high probability). We then determine the ratio between the metastable transition time for the DCWP model and the corresponding time for the CWP model. Specifically, we derive the asymptotic tail behavior and moments of this ratio. Our proof combines the potential-theoretic approach to metastability with concentration of measure techniques, the latter adapted to our specific context. 2010 Mathematics Subject Classification 60K35, 60K37, 82B20, 82B44 1. Introduction Over the past 50 years, the mathematical study of statical and dynamical aspects of disordered mean-field spin systems has attracted considerable interest. In this paper, we continue the analysis of metastable behaviour of these systems, as initiated in [ 7 ] , [ 4 ] , [ 11 ] and [ 5 ] , by examining the disordered Curie–Weiss–Potts (DCWP) model with unbounded interactions. This model generalises the disordered mean-field Ising model to q\geq 2 spins. Here, “disordered” refers to the fact that spin interactions are independent and identically distributed random variables. These random variables are chosen to have a fixed mean and a well-defined cumulant generating function, with a distribution independent of the number of spins. In particular, this model also encompasses the Potts model on homogeneous dense random graphs. Specific examples that fit within our framework include the Potts model on Erdős–Rényi random graphs, the Potts model on multi-edge random graphs, and the Potts model with Gaussian noise. As a first result, we prove metastability in the sense of [ 22 ] for the Curie–Weiss–Potts (CWP) model at fixed temperature in large volumes. Further, we show that metastability of the CWP model implies metastability of the DCWP model with respect to the same metastable sets, for almost all realisations of the random interactions. After identifying specific metastable sets for both models, we derive estimates for the ratio of mean metastable transition times in the DCWP and CWP models in the regime of large-volumes and fixed temperatures. These estimates are of two types: the first one provides insight into the tail behaviour, showing that, asymptotically in N , this ratio behaves like a random variable of order an exponential of a sub-Gaussian random variable. Moreover, we derive moment estimates for this random ratio, again in large volumes and at fixed temperatures. Our strategy is based on techniques developed in [ 5 ] and [ 22 ] . Where adaptations are necessary, references are provided to the specific results in these publications. Our methods are based on the potential-theoretic approach to metastability. This method was initiated in [ 6 ] and allows us to express mean metastable exit times in terms of capacities and weighted sums of the equilibrium potential, the latter frequently referred to as harmonic sums (for a general overview of this method, we refer to [ 3 ] ). Estimates on the former can be obtained with the help of well-known variational principles, while estimates on the latter are generally more involved and, in this manuscript, rely on a new definition of metastability given by [ 22 ] . This definition differs slightly from the standard one given in [ 3 ] , yet it provides crucial insights, particularly regarding the localisation of the harmonic sums around the initial metastable set. Additionally, our proof offers a strategy for verifying metastability in other similar mean-field models. Similar to previous works on metastability in disordered models, the use of concentration of measure is pivotal in the comparison of the disordered and mean-field model. However, in contrast to these studies, we allow for potentially unbounded random interactions. To handle this, we develop concentration inequalities using Chernoff-type bounds tailored to our setting, inspired by results from [ 15 ] . Furthermore, the presence of multiple critical temperatures, unlike the single critical temperature of the Curie–Weiss model, necessitates a careful analysis of the free energy landscape and its phase transition structure (we refer to [ 19 ] for a complete description of the free energy landscape for the CWP model). This in turn means that different temperature regimes are linked to different properties of the critical points of the free energy landscape and therefore need to be treated in different manners. Another class of disordered models is the random-field Curie–Weiss–Potts (RFCWP) model, in which the magnetic field is given by i.i.d. random variables. To our knowledge, the metastability of the RFCWP model has only been studied in [ 23 ] . The techniques developed in this paper are also applicable to the study of the metastable behaviour of the disordered Curie–Weiss–Potts model with an additional random field, under a suitable assumption on the structure of the free-energy landscape of the underlying RFCWP model. 1.1. The model The disordered Curie–Weiss–Potts (DCWP) model is a generalisation of the disordered mean-field Ising model to q components. For any N\in\mathbb{N} , consider an enumeration of the vertex set consisting of N elements. To each vertex i\in\{1,\ldots,N\} we associate a spin variable \sigma_{i} taking values in \{1,\ldots,q\} , q\geq 2 , the so-called set of colours. We write \mathcal{S}_{N}=\{1,\ldots,q\}^{N} to denote the corresponding state space. Elements of \mathcal{S}_{N} are denoted by Greek letters \sigma,\eta , and will be called configurations . Let (\Omega,\mathcal{F},\prob) be an abstract probability space and let \mean and \var denote expectation and variance with respect to \prob . Let J\equiv(J_{ij})_{1\leq i<j\leq\infty} be a triangular array of real random variables on (\Omega,\mathcal{F},\prob) whose law satisfies the following assumption. Assumption 1.1 . For some v\in(0,\infty) assume that the triangular array (J_{ij})_{1\leq i<j\leq\infty} consists of i.i.d. random variables with (i) \mean\bigl[J_{12}\bigr]=1 and \var[J_{12}]=v , (ii) the set \mathscr{D}=\left\{\lambda\in\mathbb{R}\colon\mean\bigl[\exp(\lambda J_{12})\bigr]<\infty\right\} has non-empty interior containing 0 . Given a realisation of J and N\in\mathbb{N}\setminus\{1\} , we consider the following random Hamiltonian, H_{N}\colon\mathcal{S}_{N}\to\mathbb{R} , given by (1.1) \displaystyle H_{N}(\sigma)\;\mathrel{\mathop{:}}=\;-\frac{1}{N}\sum_{1\leq i<j\leq N}J_{ij}\,\mathbbm{1}_{\{\sigma_{i}=\sigma_{j}\}}. The corresponding random Gibbs measure, \mu_{N} , at inverse temperature \beta\geq 0 is defined by (1.2) \displaystyle\mu_{N}(\sigma)\;\equiv\;\mu_{N,\beta}(\sigma)\;\mathrel{\mathop{:}}=\;\frac{\mathrm{e}^{-\beta H_{N}(\sigma)}}{Z_{N}}, where Z_{N}\equiv Z_{N,\beta} denotes the partition function. In view of Assumption 1.1 -(ii), the expected value of the partition function is finite, for all values of \beta and provided that N is chosen large enough. Notice that for q=2 , the model becomes the disordered Curie–Weiss model. The spin configuration evolves as a discrete-time Markov chain \Sigma^{N}\equiv(\Sigma^{N}(t))_{t\geq 0} , with state space \mathcal{S}_{N} and Glauber-Metropolis transition probabilities given for any \sigma,\eta\in\mathcal{S}_{N} by (1.3) \displaystyle\pi_{N}(\sigma,\eta)\;\equiv\;\pi_{N,\beta}(\sigma,\eta)\;=\;\begin{cases}\,(Nq)^{-1}\mathrm{e}^{-\beta\left[H_{N}(\eta)-H_{N}(\sigma)\right]_{+}},&\text{if