Infinite-dimensional models are core to statistical physics. They can be used to understand liquids and glasses, as they are in this book [153, 206, 292, 362], but also strongly coupled electrons [164], atomic physics [337] and gauge field theory [132], to name a few. The reason is that infinite-dimensional models are exactly solvable using mean field methods. The aim of this chapter is to give an example of this construction in the context of the Ising model of magnetism.It will be assumed that the reader is already familiar with the basic properties of the Ising model as presented, for example, in the first chapters of [69]: its main observables (magnetisation, magnetic susceptibility), its phase diagram and phase transitions, and its dynamics. The aim of this chapter is mostly to present these properties in the context of a large dimensional expansion and to introduce the concept of a thermodynamic (stable or metastable) state, identified with a local minimum of a suitable free energy function.
The Ising Model
DefinitionsAlthough some of the concepts presented in this chapter are fairly general, it is instructive to focus on a specific setting. We thus consider a model of N Ising spins {S i } i=1,...,N , with S i = ±1. The energy of a spin configuration S = {S i } is given by the Hamiltonian functionwhere J ij = J ji ∈ R denotes the (symmetric) exchange coupling between spins i and j (with J ii = 0,∀i), and B i ∈ R denotes an external magnetic field acting on spin i. The factor 1/2 in front of the exchange energy ensures that each pair ij is counted only once. The Hamiltonian in Eq. (1.1) summarises many cases of interest, 1