# Ising model

1. Sep 2, 2012

### LagrangeEuler

1. The problem statement, all variables and given/known data
Calculate magnetisation for partition function
$Z=12+4\cosh (8\beta J)$ for Ising model 2x2 lattice.

2. Relevant equations
$$F=-k_BTln Z$$
$$M(H,T)=-\frac{\partial}{\partial H}(\frac{F}{k_BT})$$

3. The attempt at a solution
For me it looks that magnetisation is zero. By just doing the derivatives. But solution of problem is $\frac{16+8e^{\beta J}}{Z}$. If I combine relation from relevant equation it looks like for me that solution is zero. Where I making the mistake.

2. Sep 2, 2012

### M Quack

Where does the partition function come from?

There is no H in the partition function, so the system does not react to any external field, just as you calculated.

What are the units of J?

3. Sep 2, 2012

### LagrangeEuler

$J$ is constant. Well you have 2x2 lattice. You calculate
$$Z=\sum e^{-\beta H}$$

4. Sep 4, 2012

### kreil

You are confusing notation here. The derivative to find the magnetization should be with respect to h, the external magnetic field. H is the Hamiltonian used to find the partition function.

If there is no term dependent on h in your partition function expression, then you have used an incorrect Hamiltonian. Assuming a constant interaction J and constant external field h, the 2D Ising Model has a Hamiltonian of the form

$$H=-J \sum_{i,j} \sigma_i \sigma_j - h\sum_j \sigma_j$$

The first sum is over all nearest neighbors.

As written your partition function trivially has no magnetization.

Last edited: Sep 4, 2012
5. Sep 5, 2012

### Oxvillian

I think there are a few sources of confusion here...

It seems that the problem presents a certain partition function as a given. Presumably it's some kind of approximation for the 2-d Ising model Z.

Also it looks like J is some parameter that's proportional to the applied magnetic field. So dZ/dH won't be zero.

6. Sep 6, 2012

### Mute

J is typically the spin-spin coupling in Ising models, and I suspect that is the case here. The partition function given simply doesn't depend on the magnetic field.

To the OP: why should one expect the magnetization to be non-zero in the absence of a magnetic field? You have a 2x2 Ising model at finite temperature. Furthermore, I assume that the spin-spin interactions are nearest neighbor only, is that correct?

If so, consider this question: what is the difference between your 2x2 Ising model and a 4-spin 1d Ising model with periodic boundary conditions (and no external field)?