Shared information in classical mean-field models

TL;DR

This paper investigates how bipartite information estimators behave in classical mean-field spin models near phase transitions, revealing universal logarithmic divergence at criticality and ensemble-dependent scaling.

Contribution

It provides a detailed analysis of bipartite information estimators in classical mean-field models, highlighting their universal critical behavior and differences between ensembles.

Findings

Estimators remain finite away from criticality.

Logarithmic divergence at the critical line with a model-dependent coefficient.

Different universal behavior in micro-canonical ensemble.

Abstract

Universal scaling of entanglement estimators of critical quantum systems has drawn a lot of attention in the past. Recent studies indicate that similar universal properties can be found for bipartite information estimators of classical systems near phase transitions, opening a new direction in the study of critical phenomena. We explore this subject by studying the information estimators of classical spin chains with general mean-field interactions. In our explicit analysis of two different bipartite information estimators in the canonical ensemble we find that, away from criticality both the estimators remain finite in the thermodynamic limit. On the other hand, along the critical line there is a logarithmic divergence with increasing system-size. The coefficient of the logarithm is fully determined by the mean-field interaction and it is the same for the class of models we consider. The scaling function, however, depends on the details of each model. In addition, we study the information estimators in the micro-canonical ensemble, where they are shown to exhibit a different universal behavior. We verify our results using numerical calculations of two specific cases of the general Hamiltonian.