Skip to content
Tutorials Center
My learning

Topics

Lesson 30 of 39 Β· Phase Transitions and the Ising Model

The 2D Ising Model: Onsager's Exact Results

Meet Onsager's exact solution of the square-lattice Ising model: critical temperature, spontaneous magnetization, critical exponents, and a brute-force 4x4 enumeration you can run.

In 1944 Lars Onsager solved the two-dimensional square-lattice Ising model exactly in zero field. It was the first rigorous demonstration that statistical mechanics, applied to a simple Hamiltonian, produces a true phase transition with non-classical critical behavior. The full derivation is hard (it uses transfer matrices of growing size and fermion-like techniques), so in this lesson we use the results and test their meaning with calculations you can run.

The headline results

For the square lattice with coupling J between nearest neighbors, zero field, and N β†’ ∞:

Text
k T_c / J = 2 / ln(1 + √2) = 2.269185...        (sinh(2J/kT_c) = 1)

Below T_c the spontaneous magnetization, as found by Onsager and proved later by Yang (1952), is

Text
m(T) = [ 1 - 1 / sinh⁴(2J/kT) ]^(1/8)       (T < T_c),     m = 0 above T_c

Compare with mean-field theory in the previous lesson: that predicted kT_c = 4J and m ∝ (T_c - T)^(1/2). The exact solution gives a lower T_c and a much flatter onset, m ∝ (T_c - T)^(1/8): the magnetization rises almost like a cliff.

Critical exponents

Near T_c thermodynamic quantities follow power laws, with t = |T - T_c|/T_c:

Quantity Behavior 2D Ising (exact) Mean-field
Specific heat C t^(-Ξ±) Ξ± = 0 (logarithmic divergence) 0 (finite jump)
Magnetization m t^Ξ² Ξ² = 1/8 1/2
Susceptibility Ο‡ t^(-Ξ³) Ξ³ = 7/4 1
Critical isotherm m(h) at T_c h^(1/Ξ΄) Ξ΄ = 15 3
Correlation length ΞΎ t^(-Ξ½) Ξ½ = 1 1/2
Correlation function at T_c r^(-(d-2+Ξ·)) Ξ· = 1/4 0

The exponents are not independent: scaling relations such as Ξ± + 2Ξ² + Ξ³ = 2 (check: 0 + 1/4 + 7/4 = 2) and Ξ³ = Ξ½(2 - Ξ·) (check: 1 Γ— 7/4) hold. They are shared by every system in the 2D Ising universality class: lattice gases, binary alloys and thin-film magnets, regardless of microscopic detail. This is the idea of universality, which Landau-style arguments cannot predict.

The logarithmic divergence of the specific heat at T_c is a hallmark of the 2D Ising model. A finite system never diverges, but its peak grows like ln L with system size L.

Evaluating Onsager's magnetization

The first example plots the exact spontaneous magnetization against the mean-field curve (in units of their own critical temperatures), and fits the exponent close to T_c to confirm Ξ² = 1/8.

Python Onsager's spontaneous magnetization and its exponent

Press Run (or Ctrl + Enter) to see the result here.

Program input Β· what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The fitted exponent should agree with 0.125 to three digits.

A finite lattice, solved by brute force

Onsager's solution is for an infinite lattice. A small lattice, however, can be solved by enumerating every configuration: a 4Γ—4 lattice with periodic boundaries has 2^16 = 65,536 states, easily handled with NumPy. This shows how a finite system smooths the transition, and sets up the Monte Carlo methods of the next chapter section, which are needed when 2^N is too large to enumerate.

Python Exact enumeration of a 4x4 Ising lattice

Press Run (or Ctrl + Enter) to see the result here.

Program input Β· what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

On such a small lattice the magnetization does not drop sharply at T_c; it decays smoothly because there is no true symmetry breaking in a finite system (<m> = 0 exactly, which is why we average |m|). The specific heat shows a broad peak close to, but not exactly at, the infinite-lattice value. As L grows, the peak sharpens and moves toward T_c: finite-size scaling, which you will use in the projects.

Check your understanding

Quick check Β· 4 questions

of 4 answered You scored / 4 Β· perfect!

  1. 1. What is the critical temperature of the square-lattice Ising model?

    This is Onsager's exact result; mean-field theory overestimates it as 4J.
  2. 2. The exact 2D Ising critical exponent Ξ² is

    m ∝ (T_c - T)^(1/8): much flatter than mean-field's 1/2.
  3. 3. Universality means that

    Exponents depend only on dimension and symmetry, not on lattice type or interaction details.
  4. 4. Why do we average |m| in a finite-lattice enumeration?

    Flipping all spins maps a configuration to one of equal energy, so the signed average vanishes.

Summary

  • Onsager (1944): kT_c = 2J/ln(1 + √2) β‰ˆ 2.269 J for the square lattice; m = [1 - sinh⁻⁴(2J/kT)]^(1/8) below T_c.
  • Exponents: Ξ± = 0 (log), Ξ² = 1/8, Ξ³ = 7/4, Ξ΄ = 15, Ξ½ = 1, Ξ· = 1/4.
  • Mean-field values are different, and universality groups systems by shared exponents.
  • Small lattices can be enumerated exactly; large ones need Monte Carlo.

Next: Landau Theory and Universality