Joint estimation of parameters in Ising model

DOI10.1214/19-AOS1822MaRDI QIDQ2196194zbMATH OpenFDO

Authors Promit Ghosal, Sumit Mukherjee

Publication date 28 August 2020

Published in The Annals of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1801.06570, https://projecteuclid.org/euclid.aos/1590480034

zbMATH Keywords

Ising modelpseudo-likelihood

Mathematics Subject Classification ID

Large deviations (60F10) Asymptotic properties of parametric estimators (62F12) Estimation in multivariate analysis (62H12) Applications of statistics to physics (62P35)

Abstract: We study joint estimation of the inverse temperature and magnetization parameters $$ of an Ising model with a non-negative coupling matrix $A_n$ of size $nimesn$, given one sample from the Ising model. We give a general bound on the rate of consistency of the bi-variate pseudolikelihood estimator. Using this, we show that estimation at rate $n^{-1/2}$ is always possible if $A_n$ is the adjacency matrix of a bounded degree graph. If $A_n$ is the scaled adjacency matrix of a graph whose average degree goes to $+infty$, the situation is a bit more delicate. In this case estimation at rate $n^{-1/2}$ is still possible if the graph is not regular (in an asymptotic sense). Finally, we show that consistent estimation of both parameters is impossible if the graph is Erd"os-Renyi with parameter $p>0$ free of $n$, thus confirming that estimation is harder on approximately regular graphs with large degree.

Review text

The authors consider the problem of the joint estimation of the parameters \((\beta,B)\) in an Ising model of the form: \[ {\mathbf P}_{n,\beta,B} (\mathbf{X}=\mathbf{x}) = \frac 1 {Z_n(\beta,B)} e^{\frac \beta 2 \mathbf{x}'A_n\mathbf{x} + B\sum_{i=1}^n x_i} \] for \(\mathbf{x}\in\{-1,1\}^n\), based on a single observation \(\mathbf{X}\) from the model. Under mean-field assumptions and additional conditions on the row sums of the matrix \(A_n\), the consistency of the pseudo-likelihood estimator is proven. When the conditions on the row sums are removed, the pseudo-likelihood estimator may be inconsistent. This result has a notable consequence in graph theory, because the estimation of the parameters is harder when a regular graph of large degree is considered. This fact is also confirmed through a simulation study. All the theory is introduced in Section 1, while the remaining sections and the supplementary material are devoted to the rigorous proofs of the results.

Recommendations

• Inference in Ising models
• Parameter estimation for two-dimensional Ising fields corrupted by noise
• scientific article; zbMATH DE number 218217
• High-dimensional Ising model selection using \(\ell _{1}\)-regularized logistic regression
• On testing for parameters in Ising models

Cites work

• Asymptotics of maximum likelihood estimators for the Curie-Weiss model
• Concentration inequalities. A nonasymptotic theory of independence
• Decomposition of mean-field Gibbs distributions into product measures
• Efficiently learning Ising models on arbitrary graphs (extended abstract)
• Estimation in spin glasses: a first step
• Gibbs measures and phase transitions on sparse random graphs
• Global testing against sparse alternatives under Ising models
• High-dimensional Ising model selection using \(\ell _{1}\)-regularized logistic regression
• High-dimensional structure estimation in Ising models: local separation criterion
• scientific article; zbMATH DE number 4058704 (Why is no real title available?)
• scientific article; zbMATH DE number 3513115 (Why is no real title available?)
• scientific article; zbMATH DE number 218217 (Why is no real title available?)
• Inference in Ising models
• Large networks and graph limits
• Nonconcave penalized composite conditional likelihood estimation of sparse Ising models
• Nonlinear large deviations
• On consistency of a class of estimators for exponential families of Markov random fields on the lattice
• Universality of the mean-field for the Potts model

Cited in

(25)

• On testing for parameters in Ising models
• Statistics of the two star ERGM
• Discussion to: ``Bayesian graphical models for modern biological applications by Y. Ni, V. Baladandayuthapani, M. Vannucci and F. C. Stingo
• Bayesian model selection for high-dimensional Ising models, with applications to educational data
• Inference in Ising models
• Estimation in spin glasses: a first step
• Asymptotically unbiased inference for Ising models
• scientific article; zbMATH DE number 218217 (Why is no real title available?)
• An expectation maximization algorithm for high-dimensional model selection for the Ising model with misclassified states*
• Inference in Ising models on dense regular graphs
• Discussion of “A Tale of Two Datasets: Representativeness and Generalisability of Inference for Samples of Networks” by Pavel N. Krivitsky, Pietro Coletti, and Niel Hens
• Sub-critical exponential random graphs: concentration of measure and some applications
• Tensor recovery in high-dimensional Ising models
• Efficient estimation in tensor Curie-Weiss and Erdős-Rényi Ising models
• Pseudo-likelihood-based M-estimation of random graphs with dependent edges and parameter vectors of increasing dimension
• Causal effect estimation under network interference with mean-field methods
• One-shot learning for k-SAT
• Limit theorems and phase transitions in the tensor Curie-Weiss Potts model
• High dimensional logistic regression under network dependence
• Phase transitions of the maximum likelihood estimators in the p-spin Curie-Weiss model
• Variational Inference Aided Variable Selection For Spatially Structured High Dimensional Covariates
• A clustered and sparse Ising regression model for multivariate binary data
• Fluctuations in random field Ising models
• Mean-field spin models -- fluctuation of the magnetization and maximum likelihood estimator
• One-shot learning for \(k\)-SAT

This page was built for publication: Joint estimation of parameters in Ising model

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2196194)