Ising model

The Ising model, also known as the Lenz–Ising model, is a mathematical framework used in statistical mechanics. It provides a way to study ferromagnetism, which is the physical property that allows certain materials to act as magnets. By using discrete variables to represent the magnetic dipole moments of atomic spins, scientists can model how large groups of particles behave. This model is highly significant because it allows researchers to understand how microscopic interactions lead to macroscopic changes in a material.

At the heart of the model is the concept of a spin. Each spin is a discrete variable that can exist in one of two states: +1 or -1. These spins are arranged on a graph, which is usually a lattice where the local structure repeats in all directions. Each spin interacts with its nearest neighbors. In a ferromagnetic Ising model, neighboring spins prefer to be aligned. When adjacent spins have the same sign, the system reaches a lower energy state. However, heat acts as a disruptive force. As temperature increases, thermal energy can disturb this tendency toward alignment, creating different structural phases.

Scientists classify these models based on how the spins interact. In a ferromagnetic model, the interaction favors alignment, meaning adjacent spins tend to point in the same direction. In an antiferromagnetic model, the interaction favors opposite signs, so adjacent spins tend to point in opposite directions. The model can also include an external magnetic field, denoted as h. When an external field is present, the spins attempt to line up with that field. If the field is zero, the system remains symmetric, meaning you could switch all spins from +1 to -1 without changing the energy of the system.

The history of the model began with the physicist Wilhelm Lenz. He proposed the problem to his student, Ernst Ising. In his 1924 PhD thesis, Ising solved the one-dimensional version of the model. A one-dimensional model is a simple line of spins where each site only interacts with its left and right neighbors. Ising found that the one-dimensional model has no phase transition, meaning it does not move from a disordered to an ordered state. Based on this, Ising incorrectly concluded that no version of the model would ever show phase behavior.

Later discoveries proved that dimensionality changes everything. In 1936, Rudolf Peierls used what is now called the Peierls argument to prove that the two-dimensional model does undergo a phase transition. This was a major breakthrough in understanding how order emerges. In 1949, Lars Onsager provided an analytic description of the two-dimensional square-lattice Ising model. He announced a formula for spontaneous magnetization, which describes how a material becomes magnetic without an external field. In 1951, Bruria Kaufman published the first formal proof of Onsager's formula using a limit formula for Fredholm determinants.

The Ising model is much more complex in higher dimensions. In dimensions greater than four, the phase transition is described by mean-field theory. In the late 1970s, researchers explored the model using various tree topologies. One specific study looked at closed Cayley trees, which are branching structures. The solution for these trees showed unusual phase transition behavior and long-range spin-spin correlations. These findings are considered relevant to the study of large neural networks.

Beyond physics, the Ising model connects to various fields of mathematics and computer science. For example, the Ising problem without an external field can be formulated as a graph maximum cut (Max-Cut) problem. This is a task in combinatorial optimization where one seeks to divide a graph into two subsets to maximize the weight of the edges between them. This connection allows mathematicians to use tools from graph theory to solve physics problems. The model remains a fundamental tool for studying how infinitesimal changes in parameters can lead to massive, aggregate changes in a system.