Ising model

The model assigns each site of a lattice a spin variable that can point in one of two directions, conventionally written as s_i = +1 or −1. Its energy is commonly written H = −J∑_⟨ij⟩s_is_j − h∑_is_i, where J measures neighboring-spin coupling and h is an external magnetic field. Positive J favors parallel alignment and models a ferromagnet; negative J favors antiparallel alignment and models an antiferromagnet.⁴

The equilibrium behavior is determined not by a single lowest-energy arrangement but by the Boltzmann distribution, which gives thermally accessible configurations weights proportional to exp(−H/k_BT). At low temperature, aligned domains can produce spontaneous magnetization; at high temperature, thermal fluctuations usually destroy long-range order. The model was proposed by Wilhelm Lenz and developed in its one-dimensional form by Ernst Ising in 1925.¹

The dimensionality of the lattice controls the model’s qualitative behavior. The one-dimensional nearest-neighbor model has an exact solution, but its magnetization vanishes at every nonzero temperature in the thermodynamic limit, so it has no finite-temperature phase transition.¹ This result showed that short-range interactions alone do not guarantee persistent ferromagnetic order.

The two-dimensional square-lattice model undergoes a continuous phase transition at a critical temperature, separating an ordered low-temperature phase from a disordered high-temperature phase. Lars Onsager obtained the zero-field free energy in 1944, and Bruria Kaufman supplied an important transfer-matrix and fermionic formulation soon afterward.²⁵

At the critical point, correlation lengths diverge and thermodynamic quantities follow power laws. The three-dimensional model is central to the theory of critical phenomena, but it has no comparable closed-form exact solution; numerical, series, and renormalization-group methods are therefore essential.

The Ising model helped establish universality: systems with different microscopic details can share the same critical exponents when they have the same dimensionality, symmetry, and range of interactions. Its two-dimensional zero-field magnetization was derived by C. N. Yang in 1952, completing a major part of the exact description of the square-lattice transition.³

Analytical approaches include transfer matrices, high- and low-temperature expansions, duality, and the renormalization group. The Kramers–Wannier duality relates the high-temperature and low-temperature expansions of the two-dimensional model and identifies the critical point for the square lattice.⁶ Numerical studies commonly use Metropolis updates, cluster algorithms such as Wolff or Swendsen–Wang methods, and finite-size scaling. These techniques estimate critical temperatures, exponents, correlations, and domain statistics in dimensions or geometries that resist exact calculation.

The model’s simplicity makes it a broad theoretical laboratory rather than a literal microscopic description of every magnet. Its spins may represent coarse-grained local moments, while variants incorporate anisotropic couplings, external fields, random bonds, random fields, competing interactions, or antiferromagnetic order. Boundary conditions—periodic, open, fixed, or mixed—can substantially affect finite lattices even when they do not alter bulk critical behavior.

A less obvious connection is the exact correspondence between an Ising lattice gas and a binary fluid: spin-up and spin-down sites can be reinterpreted as occupied and empty sites, with magnetization becoming a density difference. The model also underlies graphical expansions, including domain-wall and high-temperature loop representations, which expose its relation to combinatorics and quantum field theory. In statistical mechanics, it remains a benchmark for testing simulations, scaling hypotheses, and approximate methods because its limiting cases and critical properties are unusually well characterized.⁷

Spontaneous magnetization

Nonzero bulk magnetization that persists without an applied magnetic field below a phase-transition temperature.

Critical temperature

The temperature at which a thermodynamic system changes between ordered and disordered phases.

Universality

The occurrence of identical large-scale critical behavior in systems with different microscopic details.

Transfer matrix

A matrix method that propagates statistical weights from one lattice row or slice to the next.