Ising ferromagnets and antiferromagnets in an imaginary magnetic field

TL;DR

This paper investigates 2D Ising models under imaginary magnetic fields, mapping them onto symmetric vertex models to enable accurate numerical analysis of phase transitions and critical behavior.

Contribution

It introduces a mapping to symmetric vertex models that allows the use of tensor network methods for complex magnetic field Ising models, revealing detailed critical properties.

Findings

Critical points follow the 2D Ising universality class with known exponents.

Critical line features constant critical exponents and central charge.

Low-temperature behavior exhibits singularities and chaotic magnetization variations.

Abstract

We study classical Ising spin-$\frac{1}{2}$ models on the 2D square lattice with ferromagnetic or antiferromagnetic nearest-neighbor interactions, under the effect of a pure imaginary magnetic field. The complex Boltzmann weights of spin configurations cannot be interpreted as a probability distribution which prevents from application of standard statistical algorithms. In this work, the mapping of the Ising spin models under consideration onto symmetric vertex models leads to real (positive or negative) Boltzmann weights. This enables us to apply accurate numerical methods based on the renormalization of the density matrix, namely the corner transfer matrix renormalization group and the higher-order tensor renormalization group. For the 2D antiferromagnet, varying the imaginary magnetic field we calculate with a high accuracy the curve of critical points related to the symmetry breaking of magnetizations on the interwoven sublattices. The critical exponent $\beta$ and the anomaly number $c$ are shown to be constant along the critical line, equal to their values $\beta=\frac{1}{8}$ and $c=\frac{1}{2}$ for the 2D Ising in a zero magnetic field. The 2D ferromagnets behave in analogy with their 1D counterparts defined on a chain of sites, namely there exists a transient temperature which splits the temperature range into its high-temperature and low-temperature parts. The free energy and the magnetization are well defined in the high-temperature region. In the low-temperature region, the free energy exhibits singularities at the Yang-Lee zeros of the partition function and the magnetization is also ill-defined: it varies chaotically with the size of the system.